The big idea: America turns 250. It almost ended in Brooklyn.

Reading time: 5 minutes

This is my last Reading Ambitiously before America celebrates its 250th birthday.

From this distance, the American story can feel inevitable. A revolution, a republic, a superpower. We compress two and a half centuries into a clean arc and forget how fragile the beginning really was.

A few weeks after the Declaration of Independence, the whole experiment nearly collapsed.

David McCullough tells the story in 1776. Washington’s army had been badly beaten in Brooklyn. The British had him trapped with the East River at his back. The rebellion was still more idea than reality, and the idea was in danger. There was no guarantee the army would survive the summer, let alone that the country would survive the next two and a half centuries.

The British had the superior army and the momentum. Washington had a battered force, a narrow window, and one possible way out: across the river to Manhattan. Even that depended on the weather. If the wind shifted, British warships could move into the East River and close the trap. Washington’s army would have nowhere to go.

But the wind held.

As night fell, Washington made his move. Boats crossed back and forth in the dark, carrying men, horses, artillery, and supplies across the river. A heavy fog settled over Brooklyn and concealed the retreat. By morning, the British looked across the East River and realized the Americans were gone. The Continental Army had escaped, and the Revolution had survived.

Two hundred and fifty years later, we call that history. At the time, it was a near miss.

We tend to clean up the past when we tell it backward. We turn uncertainty into destiny. We draw straight lines through events that were anything but straight to the people living through them. We say the Revolution led to the Republic, the Republic endured, and the United States became the most powerful economic engine in human history.

All true. It also almost ended in Brooklyn.

Living through history is different from reading it afterward. People make decisions with incomplete information. Leaders carry doubt they cannot show. Good luck and good judgment become tangled together. The line between survival and collapse can be thinner than anyone wants to admit.

Life works this way, too. A chance meeting becomes a marriage. A job you almost did not take becomes the place where your life’s work begins. A child is born, and the world divides into before and after. These moments rarely announce themselves as turning points. Most of the time, they arrive disguised as ordinary days. Only later do we understand what changed.

America at 250

America is turning 250 at a strange moment. Technology is moving faster than institutions can absorb it. AI is changing the shape of work before most companies have figured out how to use it well. Markets are concentrated. Politics are polarized. Trust is low. The global order feels less stable than it did a decade ago. The future feels unusually hard to read.

When the future gets hard to read, fear fills the empty space.

The decline narrative has a certain appeal because it gives anxiety a plot. It takes every bad headline, every institutional failure, every political dysfunction, and every technological disruption, then arranges them into a story with a clear destination.

That story shows up in almost every major technological era. The new thing is too powerful. Institutions are too weak. Incumbents are too entrenched. Jobs will disappear. This generation will be the one that finally loses control of the machine.

The printing press threatened existing authority. The steam engine upended labor. Electricity remade the factory. The mainframe changed the enterprise. The PC challenged the mainframe. The internet rewired distribution. Now the world’s information is available 24/7 in our pockets.

AI is forcing the next question: what happens when software stops waiting for us to push buttons and starts pushing the buttons for us?

That is a serious question. AI will disrupt. It will expose weak business models, institutions, and strategies. It will create winners and losers. It will move faster than many people want, and probably slower than the loudest people predict.

The mistake is treating the first pain of a transition as the final verdict on it.

New technologies break old patterns before the new ones are fully visible. The future arrives unevenly. People experience the disruption before they experience the abundance. That is why turning points feel so unstable. The old world has begun to break, but the new one has not yet become useful enough to feel inevitable.

This is where history helps. It does not give us a script, but it can give us a pattern.

Each era looks different on the surface. Underneath, the same arguments keep coming back: centralized vs. decentralized, open vs. closed, regulation vs. innovation, fear vs. adoption. Every cycle has its dominant players. Every cycle has its skeptics. Every cycle has people who insist the incumbents are unbeatable and people who insist the whole thing is a bubble.

Every cycle also produces a period where the technology is visible, awkward, expensive, and overhyped before it becomes useful, embedded, and eventually ordinary.

The most powerful technologies eventually disappear into the background. Most of us did not wake up this morning and thank electricity. It is just on. AI will likely follow a similar path. It will move from spectacle to infrastructure, and from the thing everyone talks about to the thing everyone uses.

The path is messy at the moment as we approach the turning point. It often comes with speculation, monopoly power, inequality, political stress, institutional lag, and a general feeling that the economy has become detached from common sense.

The people who do well in these periods usually do not have perfect foresight. They keep building when the story is unresolved. They learn the new tools. They are curious. They adapt their organizations. They protect what matters and let go of what no longer works. They resist the temptation to confuse uncertainty with helplessness.

And they choose optimism over fear.

Across centuries, mankind has a remarkable track record of turning tools into capacity. We invent, then we adapt. We lower costs. We expand access. We use new capabilities to do work that previously seemed impossible, expensive, or reserved for the few.

That record is imperfect and full of painful transitions. But it is still the record. Again and again, human ingenuity has taken the raw material of disruption and turned it into progress.

The American story itself follows that pattern. It has never been a straight line. It has been a long argument between fear and possibility, failure and renewal, contradiction and progress. Again and again, the country has required reinvention from people who could not know the ending, but chose to act anyway.

That is the posture I invite you to carry into this next era.

We do not know exactly what AI will do to work. We do not know how institutions will respond. We do not know which companies will matter in ten years. We do not know what the next chapter of the American story will look like.

Still, all of us, not just Americans, have agency to live our ideals. We can build. We can learn. We can reskill. We can invest in people. We can make the new thing useful. We can preserve what deserves to endure while finding the courage to change what no longer works.

History is shaped by the people living inside the uncertainty. It is shaped by the people who keep rowing in the fog, who move before the outcome is obvious, who carry the idea forward long enough for it to become real.

Washington did not know how the story would end when he crossed the East River. He only knew the army had to survive the night. That is often how progress works. The grand historical arc depends on people making the next right move before they have the comfort of knowing how it all turns out.

Maybe years from now, people will look back at this strange, anxious, accelerated period and say the next era was being born in plain sight. They will make it cleaner than it felt. They will draw the line backward and give the story a shape none of us can see yet.

But we are the ones living in the fragile part, before the fog lifts, while the future is still being made. The work is to keep moving before the story makes sense. To learn the new tools and use them for good. To build what we couldn’t before. To carry the idea forward long enough for it to become real. And to be optimistic that it will.

History hangs by a thread.

And sometimes, the wind holds just long enough for the courageous ones to act.

Happy birthday, America.

I’ll see you on July 10th.

Stay ambitious.

Subscribe now

Best of the rest:

🔁 The AI world is getting ‘loopy’ – Agentic AI is moving from one-off prompts to always-on feedback loops, where agents monitor, improve, and prompt other agents, a powerful shift that could make AI more useful and much more expensive. – TechCrunch

💰 How to Earn a Billion Dollars – Paul Graham makes the founder’s case that extreme startup wealth is less about extraction than exponential growth, created by building something users love in a market big enough to compound. – Paul Graham

📚 Has AI Already Killed How-To Nonfiction? Sales Trends, My Personal Data, and What It Might Mean for the Future – Tim Ferriss uses his own collapsing book sales to make a blunt case that prescriptive nonfiction is becoming raw material for chatbots, while voice, sequencing, and transformation may be the last durable moats. – The Blog of Author Tim Ferriss

🏗️ Owning vs. Renting Intelligence — Lin Qiao argues that AI companies shouldn’t just rent frontier model APIs, they need to turn their own data, workflows, and domain expertise into proprietary intelligence they actually control. — Lin Qiao on X

🌶️ OpenAI unveils its first custom chip, built by Broadcom – OpenAI unveiled Jalapeño, its first custom inference chip, built with Broadcom and designed to lower the cost of running its AI models as the company doubles down on its own infrastructure to reduce its dependence on Nvidia GPUs. – TechCrunch

Charts that caught my eye:

→ Why does it matter? People often describe AI applications as “thin wrappers” around the model. The concern is easy to understand: if the model does most of the work, the product layer can look thin, and thin products rarely build deep moats. Maybe. But accurate output takes more than a prompt box and an API call. The model needs the right context, workflow, permissions, integrations, evaluation loop, and domain-specific definition of “good.” The moat may not come from the model itself. It may come from everything required to make the model useful, reliable, and trusted in production.

→ Why does it matter? Either we need a lot more data centers, or AI needs to get much more efficient at using the compute we already have.

→ Why does it matter? Some interesting reporting from OpenAI on how Codex is spreading beyond software engineering. Coding was the first killer use case, but the usage pattern is broadening: Legal, Finance, and Recruiting accounted for the majority of Codex usage around April 2026, and non-developer usage has grown faster than developer usage among individual, organizational, and internal OpenAI users. The next wave of agentic work may start with code, but it will not stay there.

Tweets that stopped my scroll:

→ Why does it matter? Now you can tag Claude directly inside Slack and bring AI into the work already happening. It can read the conversation, use the context, remember your instructions, and take action with the data it can access. It starts to feel less like a tool you open and more like another person helping the team.

→ Why does it matter? I remember visiting Beijing about ten years ago and stumbling across a street vendor selling software. At the time, IBM had a multi-billion-dollar software business, and one of its middleware products could sell for millions of dollars per year. The vendor had a hard drive with the source code for $30. Different technology, same story: American companies spend years and billions building valuable software, and China finds ways to copy it. More here. 👇

Worth a watch or listen at 1x:

→ Why does it matter? "To me, a model is like the database market." - Satya Nadella. Applied Computer is helping companies do that. We’re seeing early attempts with Cursor Composer and Harvey. How many models should there be? Companies should be able to embed their proprietary knowledge inside models they control.

→ Why does it matter? Nikesh Arora is the Chairman and CEO of Palo Alto Networks. Tons in here on AI inside the enterprise, including Mythos as an accelerant in cybersecurity, falling token costs, memory as the moat, and why AI in cyber creates a different set of stakes than consumer AI.

Subscribe now

Quotes & eyewash:

→ Why does it matter? The journey is the reward!

Subscribe now

The mission:

The Wall Street Journal once used “Read Ambitiously” as a slogan, but I took it as a personal challenge. Our mission is to give you a point of view in a noisy, changing world. To unpack big ideas that sharpen your edge and show why they matter. To fit ambition-sized insight into your busy life and channel the zeitgeist into the stories and signals that fuel your next move. Above all, we aim to give you power, the kind that comes from having the words, insight, and legitimacy to lead with confidence. Together, we read to grow, keep learning, and refine our lens to spot the best opportunities. As Jamie Dimon says, “Great leaders are readers.”

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, or legal advice. Readers should do their own research and consult with a qualified professional before making any decisions.

The big idea: What a Diamond Mine in South Africa Taught Me About the Build Trap

Reading time: 5 minutes

In 2015, IBM sent me to South Africa.

A few years prior, IBM had debuted Watson on the game show Jeopardy! The grand challenge made AI feel real to businesses for the first time.

We were meeting with one of the world’s largest diamond miners. The problem was both serious and impossible to forget. Diamonds were being stolen from the mines. There was even a version of the story involving carrier pigeons being used to carry diamonds out of the mines, and the company’s response was creating its own environmental problem.

Someone had mentioned to the diamond miner and the South African government that AI might help. I was sent to figure that out.

Most of my Watson meetings followed the same arc. We had this new, incredibly powerful technology. It seemed like it could do anything, so the conversation naturally moved toward what we could build. I found myself in this dynamic so often that I coined a term with myself to describe it: the “art of the possible” meeting.

I love those conversations. There is something energizing about watching people realize a new technology may let them solve a problem they had stopped believing could be solved.

But there is also a trap hiding inside the art of the possible. When technology makes more things possible, the default question becomes whether we can build something. That is a reasonable starting place, but it is incomplete. The better questions are more practical: how important is this, what will it cost to get the outcome, how much control do we need, and who will own it after the demo works?

Ultimately, the South African diamond mine decided on a multi-part solution that included biometric scans and a lightweight “lie detector” type assessment. It turned out that if you showed up for work outside your normal baseline, you were more likely to commit a lapse of integrity. AI was part of the solution, but mostly basic statistics.

The lesson I took from that meeting was not that AI was overhyped. It was that possibility has to be paired with discipline. When a new technology expands what can be done, leaders have to become much clearer about what is worth doing.

The New Build Trap

I think this is the part of the AI conversation we are entering right now. CEOs have given their teams the mandate to figure it out. Get access. Experiment. Nobody wants to be the company watching this happen from the sidelines.

That instinct is right. Companies should be experimenting with AI. They should be learning what the tools can do. The risk is that a company can create a lot of visible activity without meaningfully improving the work.

The old Build Trap was confusing shipping more features with creating more value. The new Build Trap is confusing more AI usage with better outcomes.

AI makes the Build Trap more dangerous by making building feel cheaper. What once required a product manager, designer, a few engineers, and a quarter of planning can now be prototyped in an afternoon. That is incredible, but it changes the management problem. When more things become buildable, deciding what deserves to be built matters even more.

AI also gets close enough to make a bad allocation feel reasonable. It can generate the draft, write the code or automate a small task. The demo works. The room gets excited. The team feels momentum. Sometimes that momentum is real. Sometimes it is just a faster path to spending time and money on something that should have been left alone.

Imagine canceling a $10 per month calendar scheduling subscription and replacing it with a custom AI workflow that costs $1,000 to build and run. Maybe it was a killer demo. As a business decision, something still went wrong. A cheap, working solution became more expensive.

That is the 2026 version of the art of the possible. The risk is not only that AI fails. The risk is that AI succeeds well enough to make a questionable decision feel smart.

That is why “what’s this going to cost?” should not be dismissed as a boring CFO question. It is a question that forces clarity. A budget is not just a limit. It is a forcing function. It makes you decide what matters, what does not, and where the next dollar should go.

Now that intelligence feels increasingly available, scarcity and constraints matter more. When everything feels buildable, the scarce resource is knowing what is worth building.

Importance, Cost, Control & Ownership

A team asks, “Could we use AI here?” Increasingly, the answer will be yes. But yes does not tell you what the work is worth, how much control you need, whether the economics make sense, or who will maintain it after the demo works.

Start with strategic importance. Is this workflow core to how the company creates economic value, or is it a generic business process? If the workflow is close to the economics of the business you may want more ownership. If it is a common task that every company performs, you probably do not.

Then look at cost per outcome. The question is what it costs to get good work done, not how much AI you are using. The frontier models may be worth it for the hardest tasks, but they will be wasteful for simple ones. You would not hire a brain surgeon to cut your lawn.

Next is control. Who decides which models get used, what data they can see, what gets logged, what gets remembered, and when a human steps in? This is where AI is different from ordinary software. You are not only renting a tool. You may be renting part of the way work gets interpreted, what data is retained, and where its routed.

Finally, ask who will own the thing after it works. AI makes prototypes feel cheap, but it does not make production free. Someone has to secure it, monitor it, update it, explain it, govern it, and fix it when it breaks.

These are simple tests, but require discipline in the moment to be asked.

Capital Allocation Comes for AI

Ultimately, this comes back to capital allocation. It always does.

The job of a business is to allocate scarce resources toward the work that matters most. AI does not change that. It creates a new kind of resource to allocate: intelligence.

These questions about quality, cost, control, and accountability are not as energizing as the art of the possible, but they are where AI strategy becomes real. At first, the conversation seems to be about tools. Over time, it becomes a conversation about how work should move through the company.

The art of the possible is still where the best ideas begin. Protect that energy. Companies need imagination, experimentation, and people willing to ask whether an old problem can finally be solved in a new way.

But possibility is not a strategy. Strategy is choosing. The goal is not to use the most AI, it’s to know where AI is worth it.

This is what leaders have always had to do. AI just makes the decision more important.

Subscribe now

Best of the rest:

🤝 Introducing the OpenAI Partner Network – OpenAI is admitting the enterprise bottleneck is no longer model quality but implementation, turning partners, integrators, and forward-deployed experts into the distribution layer for applied AI. – OpenAI

⚖️ Anthropic Sued Over Limits on Its $200-a-Month AI Plans – Anthropic’s premium Claude backlash shows how AI companies are learning the hard way that unlimited-feeling subscriptions collide with real inference costs once power users start pushing the limits. – The Wall Street Journal

🚀 SpaceX to acquire the AI coding startup Cursor for $60 billion – A reported $60 billion Cursor deal would mark the clearest signal yet that AI coding has become strategic infrastructure, not just a developer productivity wedge. – CNBC

🧠 GLM-5.2: Built for Long-Horizon Tasks – Z.ai’s 1M-context open model keeps pressure on the frontier labs by showing that coding agents are becoming a global, cost-sensitive, open-weights race. – Z.ai

🤖 Salesforce acquires AI customer service platform Fin for $3.6B – Salesforce is buying its way deeper into agentic customer service, reinforcing that support automation may be one of the first enterprise AI markets to consolidate. – TechCrunch

🔁 When AI builds itself – Anthropic lays out the uncomfortable next phase of frontier AI, where models increasingly help build their successors and safety questions move from theoretical to operational. – Anthropic

💸 Ramp at $44 Billion: The Third Pillar – Ramp frames AI spend as a new corporate cost category, making token governance and intelligence procurement feel like the next major finance workflow. – Ramp

🧮 Microsoft eyes DeepSeek for enterprise AI – Microsoft’s flirtation with DeepSeek shows enterprise AI is becoming a multi-model margin game, where cost, control, and token efficiency matter as much as capability. – Axios

🌐 A frontier without an ecosystem is not stable – Satya Nadella’s line is the strategy in miniature: frontier models need developers, partners, distribution, and trust to become durable platforms. – X

Charts that caught my eye:

→ Why does it matter? Incredible chart from Coatue. Innovation waves are often attributed to new technology. The real trigger is when great talent becomes available to build. I think about SpaceX. NASA at one point had 40,000 employees, many of them went to work for SpaceX.

→ Why does it matter? According to TechCrunch, ChatGPT remains the world’s most-used AI assistant with more than 1.1 billion monthly users, but its market share has slipped below 50% for the first time as Google’s Gemini and Anthropic’s Claude have gained ground.

→ Why does it matter? Live and die by the Power Law.

→ Why does it matter? SalesForce.com’s Headless 360 architecture.

Tweets that stopped my scroll:

→ Why does it matter? Hello, open source. Or, technically, open weight. It’s incredible that a model almost nobody had heard of can shoot up the leaderboard this quickly. These charts can be easy to game, but the broader point still stands: open source is catching up fast.

→ Why does it matter? This is a wild stat. Tim’s book sales are down 80% because people are going to LLMs to get what they need. OpEd by Bernie Sanders in the NYT a few weeks ago illuminates this issue. Where do you think all the intelligence came from in the first place?

→ Why does it matter? Yes, this is financial advice.

Worth a watch or listen at 1x:

→ Why does it matter? A complete guide to Nick Milo’s AI OS for personal notetaking and knowledge management. I have been a part of the second brain movement for years. LLMs are a game-changer. The beauty of Nick’s approach is that you own the underlying data.

→ Why does it matter? At Harvey, token usage jumped from 1 trillion in January to a projected 12-13 trillion this month. Molly O’Shea in conversation with their CEO, Winston Weinberg.

Subscribe now

Quotes & eyewash:

“You cannot be successful without failure.” — John F. Kennedy

→ Why does it matter? It is how you respond from that failure that is most important.

Subscribe now

The mission:

The Wall Street Journal once used “Read Ambitiously” as a slogan, but I took it as a personal challenge. Our mission is to give you a point of view in a noisy, changing world. To unpack big ideas that sharpen your edge and show why they matter. To fit ambition-sized insight into your busy life and channel the zeitgeist into the stories and signals that fuel your next move. Above all, we aim to give you power, the kind that comes from having the words, insight, and legitimacy to lead with confidence. Together, we read to grow, keep learning, and refine our lens to spot the best opportunities. As Jamie Dimon says, “Great leaders are readers.”

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, or legal advice. Readers should do their own research and consult with a qualified professional before making any decisions.

The big idea: The Seven Primitives of an AI-Native PersonalOS

Reading time: 8 minutes

“When was the first time I ever wrote about a personal AI operating system?”

Bam. A few seconds later, I had the answer.

January 30, 2026 — Building AmbitiousOS & AI Agents w/ Claude Code

In the old world, I would have gone to ReadingAmbitiously.com, searched the archive, and eventually found it. Not a difficult task, but an annoying one.

And right now, I think most of the world still thinks of AI as an answer engine. A technology a tad better than Google.

That’s a version of AI. In my opinion, the least interesting.

So let’s ask a different question.

“Hey AmbitiousOS, how has my thinking on AI changed since we initially launched?”

Now we’re cooking.

AmbitiousOS did not just retrieve the first post. It reconstructed the arc across 30 data points and five editions:

1. January 30 — The initial build. A personal operating system to help with the newsletter.

2. February 20 — Memory. The idea shifted toward memory, and the implication that near-perfect recall would belong to people who choose to document.

3. February 27 — The dataset. The “it” in AI: the dataset, often selected, created, and curated by people.

4. April 10 — Always-on intelligence. AI proactively working in the background and the hidden tax of broken memory in knowledge work.

5. May 22 — The new knowledge work. A new approach to knowledge work: externalizing your process, turning it into skills, and letting AI agents run the parts of the job that are really the “work about the work.”

Now this is the power of AI.

In the old world, I could have eventually identified this pattern. I would have read through the old editions. In the new world, AmbitiousOS already had the context.

There is a great line from John Culkin, the Jesuit priest: “We shape our tools, and thereafter our tools shape us.” Silicon Valley loves this quote because it captures something that feels obviously true once you see it.

For much of 2026, I have been writing about the shape of these tools. Now, in a small but increasingly real way, many of those ideas are making their way into the PersonalOS I’ve built, AmbitiousOS.

The path (only 4 months in length) looked something like this:

• The Reading Ambitiously archive, now more than 100 editions, became searchable.

• The searchable archive became a knowledge graph that shows how concepts connect.

• The graph evolved into the first version of system memory.

• Memory expanded into claims, predictions, and trajectories.

• The tools we developed became callable.

• The workflows involved in manufacturing the newsletter became skills.

• The routines and scheduled jobs started running those skills in the background.

• MCP made the same AmbitiousOS brain accessible across other AI environments, such as ChatGPT, Claude, and Codex.

I know that sounds technical. Great products ultimately feel magical and frictionless. Those are coming. But in a hobbyist way, I think these capabilities are starting to reveal the emerging system architecture for AI-native work.

What becomes clearer to me every day is that every knowledge worker, team, and company will want some version of this.

The primitives of AmbitiousOS

AmbitiousOS is currently built around seven primitives: context, memory, skills, tools, routines, portability, and the build loop.

1. Context: what are the inputs, and how do they relate?

The original dataset in AmbitiousOS consisted of about every edition of the newsletter and all the primary research that went into writing them. With the help of Claude Code, we built a database and labeled the data. Very quickly, it became clear that the AI needed help understanding all of this new context.

A knowledge graph is a map of how things relate. People, companies, ideas and themes all “live” near other things. Useful knowledge is rarely an isolated fact. It is usually about connection.

In the Reading Ambitiously context, OpenAI is not just a mention in an old issue. It sits near ChatGPT, Anthropic, Sam Altman, xAI, DeepSeek, Microsoft, Nvidia, tokens, agents, and the broader question of where AI value will accrue. Since I started writing, OpenAI has been mentioned 54 times in the archive. That is no longer a single note. It is a neighborhood of thought.

The power comes when your ideas are no longer disconnected notes. Once the relationships are visible, both you and the AI can start seeing the shape of your thinking. As of this writing, we track 1,071 entities and 9,831 connections amongst them.

Context provides AI with the right inputs and an understanding of how they relate.

2. Memory: what do I believe, and has it changed?

Memory is different from storage. Memory helps preserve the meaning.

A good memory system should let you rewind the tape and see how your thinking changed over time. What did you believe? When did you believe it? What changed? Does the claim still hold? Was the prediction right, wrong, partial, or unresolved?

The latest version of AmbitiousOS now tags claims, predictions, recommendations, opinions, confidence, dates, and resolution status. That means the system can help me reason across my own thinking, not just retrieve old words from the archive.

That is a meaningful shift. If I ask, “What have I written about OpenAI?” I get retrieval. If I ask, “How has my thinking on OpenAI changed?” I get memory. The first question finds artifacts. The second question helps me reason through how my thoughts have changed over time.

Memory is what helps the AI understand how context has evolved over time.

Right now, memory is a new and emerging space. Next time you’re using ChatGPT or Claude, ask it to “commit this to memory” which is a start. Better solutions are coming.

3. Skills: how should this work be done?

Skills are repeatable procedures. Think of a skill as a good onboarding document for your agents. You spell out the task, explain the process, define what good looks like, and make clear what “done” means.

This is important because much knowledge work is repetitive. Every week, I have to gather inputs for Reading Ambitiously, evaluate links, decide which charts matter, check whether I have used something before, draft “why it matters” blurbs, copyedit, source check, and assemble the final issue.

For years, most of that process lived in my head. AmbitiousOS is helping me turn that process into skills.

Once you write down a skill, it can improve. It can be versioned. It can be reused. It can be handed to an agent when the same work comes up again. A skill turns one piece of a messy process into something an AI can run more reliably. Many skills together start to look like a new way to manufacture knowledge work.

The practical lesson is straightforward: look for the work you repeat. What do you do every week? What do you explain over and over again? What process would you teach a new hire if they joined tomorrow? Write that down. That is the beginning of a skill.

Skills tell the AI how the work should be done.

4. Tools: what can the system actually do?

If skills are the onboarding instructions, tools are how the AI becomes operational.

A model can reason. A tool lets it do something. Search the archive. Query the knowledge graph. Pull recent editions. Check prior claims.

Without tools, the AI mostly talks. With tools, it can operate inside a workflow.

When I start a new edition of Reading Ambitiously, these tools help the model move through the archive and retrieve grounded context from previous editions. Instead of vaguely remembering that I talked about “the transition tax,” it can fetch the issue. Instead of saying OpenAI has been important in the archive, it can query the graph and show how often it appeared and what it connects to.

Tools are where the action happens. This is the question to ask in your own work: what would an AI need besides the answer? Should it pull files? Search a note-taking app? Check a CRM? Query a database? Draft a follow-up? Compare a document against prior decisions?

Tools give the AI the ability to act.

5. Routines: what should happen without me asking?

Routines are the heartbeat of the system.

A system that only works when I remember to prompt it is still dependent on me to start the work. That is useful, but it is not always on.

Scheduled jobs and automations create background loops. AmbitiousOS can ingest new information, prepare briefings, update the graph, refresh memory, and keep the knowledge base current. Every night while I am asleep, AmbitiousOS can go through a kind of “dream cycle” in which it gardens the underlying knowledge base and ensures it is up to date.

In software, these are often called cron jobs. The word Cron derives from Chronus, meaning time. A cron is a time-based job. The system does not need to wait for me to constantly update its knowledge base. It can routinely run jobs to do that itself.

This is one of the most important primitives because so much knowledge work is maintenance. Keeping track of what changed. Revisiting old assumptions. Connecting the new thing to the old thing.

Routines are how AI becomes a system working alongside you that’s “always on”.

6. Portability: where else should the brain be accessible?

The current best version of this is the Model Context Protocol (MCP). MCP is a technical phrase, but the plain-English version is simple: it lets the brain travel.

Early versions of AmbitiousOS were trapped inside our own interface. If you wanted to use the system, you had to go to AmbitiousOS. Sometimes I want to think in ChatGPT. Sometimes I want to build in Codex. Sometimes I want to work in Claude.

Now, via MCP, AmbitiousOS can be accessed from other AI environments.

AI becomes more valuable at the moment of work. If I am drafting a reply to an email, I do not want to leave the workflow, go to a separate archive, search for context, copy it back, and then write. I want the AI working with me to be able to tap into the relevant brain at the right moment.

In a harness like Claude, ChatGPT, or Codex, AmbitiousOS can be connected alongside other MCP’s that bring in email, calendar, Slack, documents, or code.

This is also why people talk about software becoming “headless.” The interface becomes less important than the capability. The value is not always the screen you log into. Increasingly, the value lies in the service, context, or action an agent can call on when it needs it.

Portability puts the brain where the work happens.

7. Build Loop: how does the system improve itself?

Codex and Claude Code drive the build loop.

This is the newest and most exciting part of AmbitiousOS. I can now use Codex to build the next version of AmbitiousOS while giving Codex access to what AmbitiousOS is supposed to become.

The first version of AI coding was: ask AI to write code.

The next version was: let it work inside your code repository.

The emerging version is: give it the latest edition of Reading Ambitiously and let it improve itself.

After I publish an edition of Reading Ambitiously, Codex can pick it up, understand the concepts, translate them into a product roadmap, and start helping build the feature set. If I write about the seven primitives, those ideas can be taken up by the AmbitiousOS and Codex.

The system I am building becomes part of the environment used to build the system. That is sometimes called a recursive system.

This one is mind-boggling, I know. We are only touching the tip of the iceberg.

But even at this early stage, this is pretty cool. The build loop improves the system with its own context.

Zooming out

The applied lessons are these:

• AI needs the right context and an understanding of how concepts relate to one another.

• Memory helps explain how that context has evolved over time.

• Skills give the AI instructions for how to do the job.

• Tools give the AI what it needs to do the job well.

• Routines let the jobs run in the background.

• Portability puts the AI where the work happens.

• The build loop lets the system improve with its own context.

As of June 12, that is my best description of the emerging architecture of AI-native work.

Make sure the AI captures that claim.

The more I build in this direction, the more I believe this is where knowledge work is going.

AI changes what is possible, but only if the AI has access to the right context, in the right place, at the right time, with the right tools to act on it.

We shape our tools, and thereafter our tools shape us.

Stay ambitious, my friends. And if you’re really ambitious and ask your favorite AI to give this edition a read and start building yours today.

Request access to AmbitiousOS today.

Subscribe now

Best of the rest:

💸 OpenAI Files Confidentially for Its IPO – One week after Anthropic, with an $852B private mark, a Musk lawsuit cleared, and Microsoft’s 27% stake set for a public price. – Yahoo Finance

🔒 The 24-Year-Old Running $20B on an AI Thesis – Leopold Aschenbrenner’s Situational Awareness is up 1,000%+ since launch, and Jane Street, which rarely backs outsiders, invested. – WSJ

💼 The Moat Was Never the Code – Agents made translating a domain into software cheap; knowing what right looks like is the part you still can’t prompt for. – Aaron Brethorst

🧠 The Case for Cancelling Your AI Subscription – A prolific builder lists 17 projects he never needed, calls the tooling a “thermonuclear ADHD amplifier,” and argues friction was the product. – David Wilson

🔒 OpenAI Eyes a 10-Gigawatt Ohio Campus – A 20-year lease on federal land, Nvidia guaranteeing the financing, and a build that could run $500B. – The Information

🤖 From Wizard to Patron – Mollick’s Fable review: nine-hour autonomous builds, finished work back, and his unease that steering is no longer the same as doing. – One Useful Thing

🏛️ Dario’s Policy Playbook for the Exponential – Amodei moves past transparency: FAA-style testing for frontier models, wage insurance for displacement, and a democratic coalition locking down the supply chain. – Dario Amodei

Charts that caught my eye:

→ Why does it matter? According to Coatue, a business valued between $100B and $1T has a higher statistical likelihood (31%) of multiplying its value by 10x compared to smaller, earlier-stage unicorns (8%).

→ Why does it matter? The median RAMP customer is paying $12/month per employee on AI. The top completely AI pilled decile? Closer to $9k/month per employee. The top 0.01%? $90k/month per employee.

→ Why does it matter? The Silicon Data LLM Token Expenditure Index, a benchmark for how much the market is actually spending on AI tokens, has started rolling over. According to Citadel this is a shift toward cheaper models. Companies substituting away from expensive frontier AI toward "good enough" alternatives.

→ Why does it matter? Fable (the publicly available version of Mythos) is a beast!

→ Why does it matter? This chart doesn’t include Anthropic’s Fable out this week but it’s a comprehensive LLM landscape!

Tweets that stopped my scroll:

→ Why does it matter? So, spend the the time in the meetings you’re most stressed as those are likely the ones where you’re making the hard decisions? Amazing how AI can rapidly create and personalize a dashboard like this.

→ Why does it matter? I too get the impression OpenAI will soon be back to #1 on the LLM leaderboard. Remember, we’re just starting to train models on Nvidia’s latest Blackwell series, and Vera Rubin is next.

→ Why does it matter? An important list to keep an eye on.

→ Why does it matter? A lot of talk about Elon Web Services (EWS). Well, here it is. 40% of SpaceX’s 2025 revenue.

Worth a watch or listen at 1x:

→ Why does it matter? AI may not hurt the people who “did everything wrong.” Tyler Cowen’s worry is that it hurts the people who did everything right. The consulting partner making $1.4 million does not become unemployed. He becomes a $300,000 energy executive in Houston. Financially, he is fine. Psychologically, he is not. Politically, that may matter a lot.

→ Why does it matter? Dan Shipper, CEO of Every, spent a week testing Anthropic’s Fable 5, a Mythos-class model that scored a 91 out of 100 on Every’s senior-engineer benchmark!

→ Why does it matter? If you know me, you know I’ve never seen an S-Curve I didn’t like!

Subscribe now

Quotes & eyewash:

→ Why does it matter? Haters going to hate.

Subscribe now

The mission:

The Wall Street Journal once used “Read Ambitiously” as a slogan, but I took it as a personal challenge. Our mission is to give you a point of view in a noisy, changing world. To unpack big ideas that sharpen your edge and show why they matter. To fit ambition-sized insight into your busy life and channel the zeitgeist into the stories and signals that fuel your next move. Above all, we aim to give you power, the kind that comes from having the words, insight, and legitimacy to lead with confidence. Together, we read to grow, keep learning, and refine our lens to spot the best opportunities. As Jamie Dimon says, “Great leaders are readers.”

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, or legal advice. Readers should do their own research and consult with a qualified professional before making any decisions.

The big idea: Electrons → Tokens → Outcomes

Reading time: 7 minutes

Every technology era finds a unit of account.

Small units have a way of becoming organizing principles. A barrel of oil was just a barrel of oil until the world learned how to refine and consume it. A kilowatt-hour was just electricity until factories, homes, and grids reorganized around it.

The token may be the same kind of unit for the AI economy.

In a recent interview, Jensen Huang compressed Nvidia’s mission into one sentence:

“We convert electrons into tokens.”

Electrons go in. Tokens come out. In the middle sits the world’s most valuable company.

Tokens are the units AI consumes to read, reason, and generate a result. The more information you give the model, and the more work it produces, the more tokens it uses.

Most AI initiatives are focused on the same basic thing: using tokens to generate an outcome.

The simplest frame I can come up with to put it all together is: Electrons → Tokens → Outcomes

Nvidia converts electrons into tokens. The model providers, OpenAI, Anthropic, and Google, build the engines that use those tokens. Everyone else converts tokens into outcomes.

That is overly simplified. It is also useful.

Because isn’t that what we’re all trying to figure out? How to get the AI to do what we want?

The Harness

The current best answer is the harness.

The harness is what it sounds like. A horse has power, but the harness lets you direct it. A model has intelligence, but the harness lets you aim it at work.

There is a new term called harness engineering. It is the discipline of building systems around AI, so the model can perform useful work repeatedly.

A good harness provides the model with the richest inputs possible, in the correct order. Those inputs go into the model’s scarcest resource: the context window.

Think of the context window as the model’s working memory. Context windows are measured in token size. The current start-of-the-art models offer 1M token windows. If something is not in the context window, it is not reliably in play.

The context window is the AI’s mind’s eye.

The idea is not to give the model an ocean of inputs. It is to provide the model with the essential inputs, in the right order, at the right time. Token by token.

This is why so many of the recent AI breakthroughs rhyme. MCPs, skills, memory, proprietary datasets, config files, voice, and vision can sound like separate technical ideas. They are better understood as parts of the same emerging AI native system architecture.

Each one ultimately expresses tokens that shape context.

That is the harness’s job. It determines what the model sees, what it remembers, what tools it can use, what rules it follows, and when a human needs to step in.

That is where much of the applied work in AI is happening. If you hear about a new company raising money to “build AI for X,” there is a good chance it is not training a frontier model. It is renting models from OpenAI, Anthropic, and Google, and spending its R&D dollars on the harness. This is likely what internal tech teams are doing as well.

The Product Is Not the Model

Claude Code is a useful example of a harness. Sierra, Harvey, Legora, Cursor, Decagon, and Glean are others.

There is a big difference between Claude in a blank chat window and Claude inside a software development environment. In the blank chat window, the model can answer questions and write code. Inside Claude Code, the model can see the code repository, understand surrounding files, edit code, run commands, respond to errors, and keep moving through a task.

The difference is the harness. The harness helps input all of the coding context into the model.

Ultimately, we want the work done. In the case of Claude Code, high-quality code generated.

The Business of AI

Augmentation is the easier one to understand because it’s familiar. Give people AI products and help them do more. Let engineers write more code. Let analysts read more filings. Let customer service teams resolve more tickets.

This is already happening. R&D teams using AI coding tools can build more software.

But augmentation keeps the current structure mostly intact. The user still owns the outcome. The AI is a tool. The customer still pays for seats or usage.

That version feels familiar because it looks like software with new killer features.

Service-as-software sells the work itself. The software performs the service and delivers the outcome.

When you sell traditional software, the core questions are product questions. Is it useful? Does it meet my requirements? Is it easy to install?

When you sell an outcome, the questions become operating questions. Can you do the work better than the customer can? Can you do it repeatedly? Is it of high quality? Who is liable when it’s wrong?

That is not a small shift. It changes pricing, gross margins, customer expectations, and liability.

A SaaS company can say, “Here is the tool.” A service-as-software company says, “These tools also do the job.” Then it must back it up.

And likely, future application providers will be a combination of both.

The Thrive Bet

This week, it was reported that Thrive Holdings plans to invest more than $1 billion in an AI-powered accounting roll-up. OpenAI had already taken an ownership stake with a focus on accounting services.

I do not know exactly what they are building. But here is my best guess.

They are not going to sell software to accountants. They are building a harness that is exceptionally good at accounting, then put it into the hands of their CPAs to deliver the accounting outcome faster, cheaper, and with the same or better service level.

That is service-as-software. The accounting firm is not just a distribution channel. It is the laboratory.

Inside that laboratory are the ingredients that matter: proprietary data, domain expertise, process knowledge, and a customer who expects the job to be done correctly.

And what are they going to do? Convert as much of it as possible into tokens, then drive better outcomes than the competition.

If a typical accountant can handle only 10 clients today and 100 or 1,000 tomorrow with the right harness, that is the opportunity.

When AI performs the work, someone still has to stand behind the work. That is where audit trails, human review, service levels, and liability come into play. The companies that can prove reliability will not just have better products. They will be easier to trust.

That trust is what allows the outcome to become the business.

Adding It Up

So the current best Reading Ambitiously frame for how to think about the business of AI is this:

Electrons → Tokens → Outcomes

The harness is the conversion system between tokens and outcomes.

That is where many AI companies will live. It is also where buyers and investors should spend more of their time. Not asking whether a company “uses AI,” but asking what it can reliably produce with it.

The hard part of AI right now is adding it all up without losing the plot.

Electrons become tokens. Tokens move through a harness. The harness produces work. The work creates an outcome. The outcome creates value when someone trusts it enough to depend on it.

That last step is the hard one.

It is also where the business lives.

And unless you’re building picks and shovels, that’s how you’re going to make money.

The trusted outcome.

If you plan to build an AI business, this is what you need to work backward from.

Subscribe now

Best of the rest:

💸 Anthropic Confidentially Files Its Draft S-1 — Anthropic submitted a confidential draft S-1 to the SEC on June 1, putting it in the 2026 IPO race alongside SpaceX and OpenAI, on a run rate north of $47B. — Anthropic

⚖️ Kirkland to Spend $500M Building Its Own AI Platform — The world’s highest-grossing law firm is building, not buying: $500M over three to four years to own its AI outright, a direct bet against renting from Harvey or Legora. — Reuters

🧮 Building Finance AI Agents — OnlyCFO breaks a finance agent into its four parts (a skill, a system prompt, MCP connectors, a trigger) and shows one flow turning a two-hour reconciliation into five minutes. — CFOpilot

🛠️ 11 Ready-to-Use AI Skills — Khe Hy hands over eleven copy-paste Claude skills for everyday knowledge work, the kind you install once and forget you are running. — Khe Hy

🧠 Where the Winners Get Built — Brett Queener argues the next durable software companies get built inside specific, messy domain workflows, not horizontal AI, because that is where context and trust compound. — Tales from the Bonfire

🐆 Riding the Leopard — Packy McCormick’s talk-turned-essay on composure and creative nerve: if you are a piece of the universe experiencing itself, the job is to be the fullest, strangest version of you. — Not Boring

Charts that caught my eye:

→ Why does it matter? Anthropic engineers, on average, ship 8x as much code per quarter as they did in 2021-2025.

→ Why does it matter? Just 10 SaaS companies accounted for ~80% of the market cap regained over the last 34 days. Full update from Meritech here.

→ Why does it matter? AI-related companies have issued ~$140 billion in investment-grade bonds year-to-date and attracted ~$220 billion in venture capital funding, making up 87% of the total. H/t: The Kobeissi Letter.

→ Why does it matter? DeepSeek (open-source Chinese LLM) on top of the trending?! 👀

→ Why does it matter? Do we have the compute?

Tweets that stopped my scroll:

→ Why does it matter? $84 vs $954 across the same 100 tasks. 11x cheaper! This is the power of open-source alternatives to the state-of-the-art closed source. The question is: would you trust a Chinese open source model to do it?

→ Why does it matter? Imagine what Siri will be capable of in 1-2 years? And Apple plans to run these models locally. The rumored MacBook Ultra (an unreleased Apple laptop) is considered to be even more powerful than the MacBook Pro.

→ Why does it matter? This is AI for good. My first grader and I are already using Claude Code together. Be sure to check out Reading Ambitiously 4-17-26 if you missed it.

→ Why does it matter? Jonny Ive’s rumored OpenAI device?! During Jon’s years at Apple, the leaks weren’t always unintentional. Jonny loved to build the hype.

Worth a watch or listen at 1x:

→ Why does it matter? Sarah is one of the best CFOs in the game and no doubt spent a ton of time preparing for her remarks here. There has been a media driven story about whether she and Sam are on the same page about the spending. Lots to unpack!

→ Why does it matter? Remember when Uber IPO’ed and people said it’d never be profitable? Dara has masterfully executed and shares a lot of the story with Patrick.

→ Why does it matter? Fred Wilson co-founded Union Square Ventures in 2004. USV has backed Twitter, Etsy, Coinbase, Kickstarter, and MongoDB. After 40-years in the industry, Fred is rethinking how to organize the firm and how AI is going to play a major role in it.

Subscribe now

Quotes & eyewash:

→ Why does it matter? Earlier this week, I had the opportunity to meet Coach Jack Clark. The Cal Rugby coach with the highest winning % of all time in collegiate rugby, #2 place isn’t even close. Jack offered some incredible thoughts about leadership. Non-cognitive grit, mental toughness, grateful for everything, entitled to nothing, team. He sends his team through a tunnel every game (a reminder that you’re coming into our world) a world where your deeds speak louder than your words, a world where we all win together. An incredible model for leadership which you can read more about here.

Subscribe now

The mission:

The Wall Street Journal once used “Read Ambitiously” as a slogan, but I took it as a personal challenge. Our mission is to give you a point of view in a noisy, changing world. To unpack big ideas that sharpen your edge and show why they matter. To fit ambition-sized insight into your busy life and channel the zeitgeist into the stories and signals that fuel your next move. Above all, we aim to give you power, the kind that comes from having the words, insight, and legitimacy to lead with confidence. Together, we read to grow, keep learning, and refine our lens to spot the best opportunities. As Jamie Dimon says, “Great leaders are readers.”

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, or legal advice. Readers should do their own research and consult with a qualified professional before making any decisions.

The big idea: I asked AI about a dead tree limb. The next morning, it was gone.

Reading time: 4 minutes

With three kids, there is always something happening at the Lynch house.

There is also always something happening literally with the house.

Every homeowner has some version of the project list. Not the well-organized one you write down on an ambitious Saturday morning. The other one. The one that lives in your head and follows you around.

Fix the thing in the garage. Call the person about the window. Figure out why that outlet doesn’t work.

For months, there had been a dead limb hanging from the tree in front of our house. I saw it all the time. I knew it needed handling. I’m almost embarrassed to admit this, but I had no idea whose problem it was.

Did we own the tree? Or was this the city’s responsibility, buried somewhere in a municipal gray area that would take twenty minutes of research to understand?

And because it required twenty minutes of research, I did what any reasonable adult with too much on his plate would do.

Nothing.

Not because it was hard. Because it was annoying. It was one more thing on the long list of things I knew I should handle but kept putting off.

Until Tuesday night, when I opened ChatGPT and said, in effect: you figure it out.

A few seconds later, the AI told me exactly what to do. It identified the likely responsible party, found the right place to submit the request, wrote the ticket language, and assigned me one 30-second task: snap a photo.

I copied the language into the town portal, attached the photo, and hit submit.

The next morning at 7:53 a.m., to my surprise, guess who was out front.

The public works department.

I texted my wife: “You’re not going to believe it. The AI really did it.”

The AI did not operate the bucket truck. A human crew did the real work.

But it gave me the capacity to get something done when I personally was at 0%. It turned a vague annoyance into an action. And, not for nothing, it gave me the creative spark for this week’s Reading Ambitiously.

So much of modern work is not the work itself. It is the work about the work.

Find the website. Research the rules. Draft the form. Upload the document. Submit the request.

Small tasks, maybe. But they create a huge amount of friction. They are not strategically important enough to dominate the day, but not trivial enough to disappear. So they accumulate.

That is what the dead tree branch represented. Not a major life problem. A category.

A list of things that sit unfinished because the first step is annoying. We all have metaphorical dead branches in our lives.

And there are a lot of them at work.

The morning briefing

I have always loved the idea of the President’s Daily Brief.

Before the President wakes up, an entire intelligence apparatus has been working to answer one question: what needs to be on the President’s radar today?

That is a pretty good mental model for how I want AI to work for me.

So I told the AI to start acting like my own small army of intelligence officers. Hours before I wake up, it starts preparing Jack’s morning briefing. The inputs are simple: email, calendar, Slack, and the context of what is going on in my world.

It arrives at 7 a.m.

The briefing is useful. It tells me what happened overnight, what meetings matter, what I owe people, what needs preparation, and what might otherwise slip through the cracks.

But the briefing is only the beginning.

These days, I ask the AI a follow-up question:

“What can you do for me?”

• If I have a customer meeting, the AI pulls together background on the attendees.

• If I am double-booked, it suggests a reschedule and drafts the email.

• If I need a colleague’s help, it drafts the Slack message and asks for fifteen minutes.

• If an important email slipped through the cracks, it suggests the follow-up.

• If I have a meeting I accepted three weeks ago and have not thought about since, it writes the pre-read.

That is the work about the work. The layer of small, necessary actions that make the real work possible.

Honestly, most days I print the briefing and review it with my notebook and pen. And then record a voice memo back to the AI to get to work on the to-do list. That combination feels amazing to me: offline thinking, powered by technology.

In the wise words of Ferris Bueller, “If you have the means, I highly recommend it.”

That is the difference between a chatbot and an agentic workflow. Once AI can operate where work already lives, the interface starts to feel less like search and more like delegation.

Mine the mundane for leverage

The first spreadsheets did not replace CFOs. They replaced the manual recalculation.

Early behavior change is often practical before it is profound.

That seems to be happening here. AI is starting to collapse the distance between noticing a task and acting on it. The AI did not just answer my question about who owned the tree. It helped create the ticket, move the process forward, and actually got the branch removed. The outcome.

Of course, AI still makes mistakes. Proceed with caution.

But there is a lot of work that should not require so much human friction.

Ask AI to take the first pass.

It might surprise you how many things on your list are not actually hard. They are just annoying to start.

And it might surprise you what comes off the list.

Stay ambitious, my friends.

Subscribe now

Best of the rest:

💸 The magic of doing $10,000 per hour work – Khe Hy’s leverage framework is a useful reminder that the highest-value work often feels slow, ambiguous, and unrewarding in the moment, but compounds into the systems, relationships, and judgment that actually move a life or business forward. – RadReads

🕊️ Magnifica Humanitas – Pope Leo XIV’s first major AI encyclical frames artificial intelligence not as a technical novelty, but as a civilizational test of whether power, work, truth, and human dignity can still be ordered toward the common good. – The Holy See

🏦 Your AI agent now has a brokerage account – Robinhood opened native MCP servers letting agents trade equities and spend on a ring-fenced virtual card, with limits you set. – Robinhood

🤖 Two Claude diehards now reach for Codex first – GPT-5.5 needs less hand-holding and compacts long threads better; they still keep Claude for design and writing. – Technically Curious

👋 Salesforce starts waving bye to the UI – A headless, agent-first turn for the CRM giant. One read: enterprise software gets consumed by agents, not screens. – The Register

🛡️ AI is finding critical bugs faster than humans can patch them – Anthropic’s Project Glasswing used Mythos Preview to surface over ten thousand high-severity vulnerabilities. Finding flaws is now easy; the bottleneck has shifted to patching. – Anthropic

🧠 Being rude to ChatGPT made it more accurate – A small study found very rude prompts beat very polite ones on accuracy, 84.8% to 80.8%. The opposite of what older work showed. – arXiv

Charts that caught my eye:

→ Why does it matter? 4.1M recipes. 7 languages. 1,790 ingredients. 300 dimensions!

→ Why does it matter? Interesting data indicating that the majority of AI tokens are being spent on bug fixes and rework, with only 18% going to shipped products.

Tweets that stopped my scroll:

→ Why does it matter? Turns out “slop” was the Merriam-Webster Dictionary Word of the Year in 2025, which we predicted! Ethan Mollick’s latest One Useful Thing piece, “Choosing to Stay Human,” is a useful reminder that the real question is not whether to use AI, but what parts of our thinking we want to keep for ourselves. The choice ahead is not AI adoption or AI avoidance. It is deciding, task by task, what we are willing to hand over and what we still want to earn ourselves.

→ Why does it matter? In just 24 months, Slack predicts more agents will use their platform than humans. SFDC also had a big Q1!

→ Why does it matter? I’m feeling this vibe shift as well. OpenAI’s Codex (competitor to Claude Code & Cowork) is getting really good. Maybe it's personal preference, but I do enjoy GPT 5.5 as a model and interacting with it. Not sure how to describe it, Opus is colder, more rigid. GPT 5.5 feels like a better all-purpose model. Doesn't matter when coding, but it is noticeable, almost off-putting in the non-coding workflows. I feel like I can get into a better flow state with GPT 5.5 for knowledge work, less so with Sonnet/Opus.

→ Why does it matter? 🙈

Worth a watch or listen at 1x:

→ Why does it matter? Dan Shipper is living in the future and has committed to leaning into that future in the most possible way. Listening to Dan talk about how he completes his work, his “AI-native workspace,” and his thoughts on his recent article, “After Automation,” provides many thought starters.

→ Why does it matter? David Senra is at the top of his game. His interview with Rick Rubin on the creative process is outstanding because David does the work: he reads the books, studies the guest, and shows up with real curiosity. You can feel his passion and preparation in the conversation, and Rubin is the perfect guest for that style. This one is thoughtful, patient, and full of insight on creativity, taste, and the discipline of making things. I may need to listen to it again. It is that good.

Subscribe now

Quotes & eyewash:

"Until death, all defeat is psychological." - Marcus Aurelius

→ Why does it matter? Never give up!

Subscribe now

The mission:

The Wall Street Journal once used “Read Ambitiously” as a slogan, but I took it as a personal challenge. Our mission is to give you a point of view in a noisy, changing world. To unpack big ideas that sharpen your edge and show why they matter. To fit ambition-sized insight into your busy life and channel the zeitgeist into the stories and signals that fuel your next move. Above all, we aim to give you power, the kind that comes from having the words, insight, and legitimacy to lead with confidence. Together, we read to grow, keep learning, and refine our lens to spot the best opportunities. As Jamie Dimon says, “Great leaders are readers.”

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, or legal advice. Readers should do their own research and consult with a qualified professional before making any decisions.

The big idea: The new knowledge work

Reading time: 7 minutes

Last week I wrote about the transition tax. On Saturday, after the kids went to bed, I started paying it.

You might think, given how much I write about AI, that the manufacturing of this newsletter is entirely automated. It is not. Actually, it is more manual than I’d like to admit.

I am not talking about the Big Ideas or the content I have tastefully curated for you. I am talking about assembling HTML, screenshotting tweets because Substack is in a feud with X, formatting, carefully selecting the right emojis for Best of the Rest, and attributing the quotes.

This is not the thinking part. It is not the creative process, nor the ultimate expression of making something from nothing. That is by far the most rewarding part of why I do this every week.

This is the work about the work. And it takes a lot of time that I want back.

The central promise of AI is to give it back. To let your Agents take on the mundane.

To elevate you. And to get there, you need a new form of knowledge work.

The new operating system and applications

In the old world, the knowledge worker has access to a computer, an operating system, and software. A file system, your favorite applications, productivity tools, and a browser. You use these tools to move the work.

Software historically gave you the tools to do the job. What is different with AI is that the tools can do the job.

The new world is powered by AI, and its primary operating system harnesses intelligence to drive outcomes. Your new job as a knowledge worker is to get in the driver’s seat, tell it what to do, and give it the vital context it needs.

One example of a harness is Claude Cowork from Anthropic. When you open it, you see a mosaic of capabilities working together:

• A voice or text-based experience powered by a large language model you select.

• Connectors that wire the brain to the tools where your work already lives, such as your email, calendar, files, and applications.

• Plugins, which are pre-built collections of capabilities for a specific kind of work.

• Memory management (although it is very nascent today).

• And skills.

Skills are one unit of automation. They are the onboarding documents for your AI agents.

Skills are to the new world as applications were to the old world.

You write down how a task should be done, the way you would explain it to a new hire on their first day. The agent reads it when the task comes up. The next time the work comes up, it runs the same way.

If a prompt is a conversation, a skill is a procedure.

Software used to come in a box. You bought it, installed it, and shaped your work to fit its menus. The unit of software was the application, and the work occurred within it. Like those icons on the desktop.

The unit of software in the new world is now the skill. You write it, shape it, and version it. The work happens around it, in the harness, with the agent reaching for the skill and the context at the right time.

A working example, ready to install

Last week, Anthropic released something called Claude for Small Business. What is it? You guessed it, a big library of skills. Fifteen of them, all written for the work a typical small business does.

There are Monday morning briefings, tax prep, month-end closing, cash forecasting, and payroll.

Each of these is a paragraph of plain English sitting in a markdown file, ready to run when you ask for it. The files are version-controlled and updated by Anthropic. You install these skills into your harness, wire it up to your systems like HubSpot, QuickBooks, and PayPal, and you are ready to start automating knowledge work.

It is clear to me that an app store of skills will come. I see a world where you develop a skill, share it with a colleague, and make it available to others, free or for a fee, through a marketplace built for security and compliance.

You do not have to wait for that to arrive. If you can make Word documents, you can start developing skills today.

The agentic factory

I have been building skills to automate the manufacturing process of this newsletter. I liken the compilation of skills to the periodic table.

The elements are discrete. Each one does a very specific thing. Together, they make everything.

There is:

• A sweep skill that pulls every candidate from the private Slack channel where I flag items every week. This channel is accessible to Claude via a Slack MCP connector.

• A classifier that interprets the content as either Best of the Rest, charts, tweets, or podcasts.

• A fetcher that fetches data and metadata from the internet about the content.

• An archive checker that uses AmbitiousOS to search every previous issue and see if I have already used the content.

• A drafter who starts coming up with ideas in my voice for the why it matters.

• A builder that handles Substack’s rich HTML formatting requirements.

None of these is doing the thinking or the heavy lift of writing the Big Idea. That is the part I love and the human-only part. What they are doing more and more of is the manufacturing, the part of the job that takes hours every week.

Your new job

The knowledge worker’s job is to do the work we are doing here.

The hardest part is thinking in a new way. It is taking a step back and thinking through the processes you have today to manufacture your work. Most knowledge workers have never written this down. We do the work because we know the work, but the way we do it lives in our heads.

Your next job is to externalize that knowledge. Break the processes into elements, write each element into a skill, and let the agents run the elements you should not be running yourself.

This is the practice of paying the transition tax. And you can start right now by opening your favorite AI tool and saying “Hey, let’s create a skill together”.

Done right, you can construct and assemble the skills your agents need to automate the mundane, so that you can move to higher and higher value work. The most creative and rewarding part of the human experience.

It’s worth it, let’s pay the tax together.

Stay ambitious.

Subscribe now

Best of the rest:

📚 You Need AI That Reduces Maintenance Costs — James Shore drops a spreadsheet on “Rock Lobster,” his fictional 2x-output coding agent, and shows what we’ve all suspected: if AI doubles your output without halving the cost of maintaining that output, you’ve quadrupled your future maintenance burden. Turn the agent off, and the productivity bump evaporates while the debt stays. Hotel California, but for code. — James Shore

🎯 “AI-Powered” Isn’t a Position — Arielle Jackson returns to First Round Review with the playbook for differentiating when every category is converging on “AI-powered X.” The bottled-water frame is the keeper (Evian and Liquid Death built distinct brands on functionally identical water) and the Cursor case is sharper — they won by repositioning every few months as the terrain moved, and the day they stopped, Claude Code ate their frame. — First Round Review

📊 AI eats the world — Spring 2026 — The new Ben Evans macro deck just dropped, titled “AI eats the world”. — Benedict Evans

🧠 After Automation – Dan Shipper argues that AI will not eliminate expert work so much as flood the zone with cheap competence, making human judgment, taste, and framing more valuable than ever. – Every

🧱 The Inference Shift — Ben Thompson suggests the chip future bifurcates along a more useful axis than training vs inference: “answer inference” and “agentic inference”. Three sideways implications worth chewing on: China already has everything it needs for agentic inference today, space data centers suddenly make sense, and the way we get more compute might just be realizing the compute we already have is good enough. — Stratechery

Charts that caught my eye:

→ Why does it matter? SpaceX in IPO filing: "We believe we have identified the largest actionable total addressable market in human history. We estimate that our quantifiable TAM is $28.5 trillion! 🤯 It was also reported via the S-1 that Anthropic is paying SpaceX $1.25B/month to rent their datacenters. What’s a $15B/ACV deal for SpaceX! Oh, it includes a 90-day cancellation clause. Buckle up!

→ Why does it matter? Back in January 2025, the conversation was: Will the AI revenue come? Well, here it is. Palantir, Snowflake, and Databricks are the three highest-profile SaaS companies founded in the last 10 to 12 years, and each spent roughly 10 years building its business. In comparison, Anthropic added the combined annual recurring revenue (ARR) of those three companies in just one month.

→ Why does it matter? Thiel sold out of Facebook early and missed roughly $100B of upside. The SpaceX IPO could turn that into the tightest comeback story in venture — ~$80B personally on a $3T print. The lesson isn’t that he was right about SpaceX; it’s that the power law lets you be right once and rich forever.

→ Why does it matter? When we first wrote about Google’s AI token volume, the number was already staggering: 480 trillion tokens per month. Today, it is 3.2 quadrillion, yes, quadrillion.

→ Why does it matter? Be the person who doesn’t wait for the conditions to be perfect or a victim of your circumstances to go get what you want.

→ Why does it matter? Newest roadmap on the AI data center stack from Bessemer Ventures!

→ Why does it matter? Average LLM token costs are now $2.12/mil tokens,+12% this week alone and +65% since the end of Feb.

Tweets that stopped my scroll:

→ Why does it matter? Karpathy was the unofficial standard-bearer for the OpenAI alumni network. He’s now wearing orange. Rumor has it this Tweet got more engagement than Google I/O (which cost Google millions). Karpathy is going to work on building AI Research agents that can generate new AI Research. Once you have the AI making the AI better, this is the endgame.

→ Why does it matter? $1.3M in tokens from a single user in one month — roughly $43k a day. Two years ago, that was a startup’s monthly inference budget. Today, it’s an individual. Either we’re radically underestimating the per-user upper bound, or radically underpricing tokens. Probably both.

→ Why does it matter? Twenty-year arc from six Dell boxes called “the hive,” sitting out in the room with the traders so they could be unplugged in a crisis (someone once vacuumed one off mid-trade), to 4,032 liquid-cooled GPUs in Texas processing trades in under 100 nanoseconds.

Worth a watch or listen at 1x:

→ Why does it matter? The Acquired guys are THE best, and this one was so good. Bogle gets fired by his own partners. He turns around and builds a shareholder-less mutual that's now the biggest equity owner in most of the S&P 500. No wonder Acquired is considered to be the most expensive ad space in podcast land.

→ Why does it matter? Gavin Baker is at the top of his game right now. He’s running a deep understanding of the entire AI supply chain in his head, updated daily. Nobody better than Patrick to tease it out. Worth 1x for the walkthrough of how every layer connects.

Subscribe now

Quotes & eyewash:

→ Why does it matter? And it was a masterpiece that we still watch with the family today!

Subscribe now

The mission:

The Wall Street Journal once used “Read Ambitiously” as a slogan, but I took it as a personal challenge. Our mission is to give you a point of view in a noisy, changing world. To unpack big ideas that sharpen your edge and show why they matter. To fit ambition-sized insight into your busy life and channel the zeitgeist into the stories and signals that fuel your next move. Above all, we aim to give you power, the kind that comes from having the words, insight, and legitimacy to lead with confidence. Together, we read to grow, keep learning, and refine our lens to spot the best opportunities. As Jamie Dimon says, “Great leaders are readers.”

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, or legal advice. Readers should do their own research and consult with a qualified professional before making any decisions.

The big idea: The transition tax

Reading time: 5 minutes

“The future is here, it’s just not evenly distributed.”

William Gibson wrote that way before AI arrived. But the quote feels especially relevant as I write this week’s edition from 35,000 feet, using United’s Starlink Wi-Fi, powered by SpaceX, to send instructions to my AI agents working across multiple projects in parallel.

The stack itself feels improbable. SpaceX powers the high-speed satellite Wi-Fi. OpenAI and Anthropic power the AI agents. Three companies, all sitting near the center of trillion-dollar-scale speculation about where the next platform shift is headed.

Yet these technologies remain astonishingly scarce in practice. Out of curiosity, I asked ChatGPT to estimate how many people on Earth might be using Claude Code on a Starlink-enabled United flight at the same time.

Its answer: maybe 20.

The future often arrives this way, slowly as a strange little pocket of improbability that barely anyone notices. Then suddenly.

That future is here. But increasingly, the technologies themselves are not the only things unevenly distributed. The more I live in this new world, the more hyperaware I become of another uneven distribution: mankind’s capacity to absorb change.

Two Parallel Realities

Getting off the plane, I stepped back into the old world, which was fittingly O’Hare’s Terminal 2. It has not been updated in decades. You can still see where the telephone booths once were.

Spend a couple of hours using all this new AI, then try to do the same work the old way. No way. The new way is just too good.

But here’s the thing: the old systems have not disappeared just because the new ones arrived.

Knowledge workers are now balancing two parallel realities: the existing job they are still accountable for, and the emerging AI-native workflow they suspect may eventually redefine the job entirely.

It is energizing, invigorating, concerning, and exhausting all at once.

I feel this tension personally every week.

Most Friday mornings, after a long week of travel, meetings, operational cadence, and finally hitting publish on Reading Ambitiously, I let out a sigh of relief. Then I typically message our CTO and VP of AI some version of the same thought:

“Gosh, I had so much to get done this week based on my traditional set of responsibilities, but I can’t wait to continue building out all the AI to redefine my next set.”

More often than not, I do not feel like I have enough hours in the day for the latter.

This AI revolution is not arriving into empty calendars.

Microsoft calls this the “Transformation Paradox.” Organizations want all the benefits of AI while still measuring performance through legacy systems and incentives.

And unfortunately, new technology tends to add complexity before it reduces it, leading to burnout and what feels like a cognitive tax.

This is normal. When email arrived, it increased the volume of communication and raised expectations for timely response. Now that AI is here, it is incredibly good at generating content. But reviewing and validating that content is becoming more demanding.

Ahead of a recent meeting, I received more than 50 pages of pre-reading material, much of it carefully curated by our teams and some of it helpfully generated by AI.

What am I supposed to do? Have AI summarize the AI-assisted pre-read?

The Deployment Layer

Companies are beginning to realize they may need help with this transition. For many, AI-native work is still happening in the shadows. But the market is beginning to professionalize that shadow work.

Carlota Perez, my spirit guide for technological revolutions, would recognize this moment immediately. We are somewhere past the frenzy, approaching the turning point, and the deployment period is beginning to take shape. That is not a metaphor. It is literally what they are calling it. OpenAI announced the OpenAI Deployment Company this week, a new venture staffed with forward-deployed engineers who will help companies install AI.

Stripe seems to understand this, too. The company recently posted a new role, “Forward Deployed AI Accelerator,” describing it as embedding AI-native operators directly into its marketing department to help teams of roughly 20 make AI the default mode for all work.

The hard part is no longer simply getting access to AI, although it is astonishing to me how many people still do not have it in the workplace. The hard part is integrating AI into the business's daily operations and embracing change.

Why It Matters

The lesson is that new technologies do not create an advantage simply because they exist. They create an advantage when people reorganize around them.

At first, the new tool gets bolted onto the old system. Email becomes faster mail. Websites become digital brochures. Cloud becomes rented servers. AI becomes a better search engine.

Useful, but the real leverage shows up when the work changes shape.

Electricity is the classic example. Before factories were electrified, many were organized around a central steam engine. The engine turned a shaft. The shaft ran through the building. Belts and pulleys carried power to individual machines. The entire factory was designed around the constraints of the power source.

Then electricity arrived. But many factory owners did not immediately redesign the factory. They simply replaced the steam engine with an electric motor. They had a new power source running an old operating model.

The real gains came later, when factories put smaller electric motors on individual machines. Once power no longer had to flow through a single central shaft, the work could be reorganized. Machines could move. Production lines could change. Buildings could change. The flow of work could change.

The technology mattered, but the reorganization mattered more.

A century later, Tesla gave the same idea a modern form. The car mattered, but so did the machine that built the car. Tesla framed it as an effort to turn the factory itself into a product and build the machine that makes the machine.

That may be the best way to think about AI.

The tech is impressive. But the long-term advantage will likely come from the operating system around it.

Right now, many companies are using AI like an electric motor attached to the old shaft. AI-generated pre-reads. AI-assisted summaries. AI pilots. AI copilots layered on top of work designed before AI existed.

That is useful. It is not the unlock.

The better question is not: how do we use AI to do the old work faster?

The better question is: what would this work look like if AI were assumed from the beginning?

The old world is still here. The new one has arrived. And most of us are living in the gap between them, landing in Terminal 2 while our agents are still working somewhere above the clouds.

The advantage goes to the people and companies willing to close that gap. Not by electrifying the shaft. By redesigning the factory.

The ultimate long-term advantage will go to individuals and companies that pay the transition tax.

Stay ambitious!

Subscribe now

Best of the rest:

🤖 OpenAI launches the OpenAI Deployment Company to help businesses build around intelligence — OpenAI is turning enterprise AI deployment into a company-level bet, combining FDEs, Tomoro’s applied AI talent, and a heavyweight partner network to move customers from AI experiments to production systems. — OpenAI

💸 How a Job at OpenAI Became the Greatest Lottery Ticket of the AI Boom – OpenAI’s latest employee stock sale shows just how much wealth the AI boom is creating before IPO, turning hundreds of employees into multimillionaires and giving Silicon Valley a preview of the economic shockwave still to come. – The Wall Street Journal

🍽️ QR Codes Didn't Kill Waiters. And Neither Will AI. — Sophie Bakalar makes the clearest case yet that AI will automate tasks, not jobs, because the human layer still matters wherever trust, care, and accountability are part of the product. — Sophie’s Notes

🛠️ Learning on the Shop floor — Shopify’s River shows that the real unlock in enterprise AI may not be private productivity, but public apprenticeship, where every agent interaction becomes searchable, teachable, and compounding company knowledge. — Tobi Lütke

⚡ Compute Futures Are Here – As AI demand turns GPUs and advanced chips into strategic infrastructure, financial markets are beginning to treat semiconductors like oil or electricity, creating new ways to hedge price volatility and speculate on the backbone of the AI economy. – CNBC

Charts that caught my eye:

→ Why does it matter? Tech startups are 5% of active U.S. entrepreneurs using ChatGPT. The other 95%: agency owners, plumbers, Etsy sellers, dentists. One read: the AI adoption story for small businesses is already further along than the coverage of it.

→ Why does it matter? SalesForce.com has announced they’re going headless. Seema Amble argues that as agents bypass the UI, software defensibility shifts from human workflow and muscle memory to data models, permissions, proprietary context, compliance, and real-world execution. — a16z

→ Why does it matter? The clearest snapshot I've seen of where software adoption is actually moving: 33 of 69 major vendors declining YoY. Great work from Deedy Das — worth spending time in every quadrant.

Tweets that stopped my scroll:

→ Why does it matter? In the pursuit of organizing around the work, some CEOs are trying entirely new designs. Dorsey's version: collapse Block's five management layers to two or three, with AI doing the information routing that middle management used to own. (Fortune)

→ Why does it matter? Few read the market better than Coatue, and their take: the market is pricing the supply chain, not the applications.

→ Why does it matter? 🤯

→ Why does it matter? The mouse pointer is a 50-year-old interface, and Google DeepMind is experimenting with replacing it with motion, voice, and gesture.

→ Why does it matter? Shopify's internal AI agent increased its merge rate from 36% to 77% in two months. Not by switching models, but by making all AI work visible in public Slack channels. Everyone watching, everyone learning.

Worth a watch or listen at 1x:

→ Why does it matter? Naval's blunt take: pure software is uninvestable. If your only edge is code that others can't build, that edge is gone. Worth 1x for the contrarian framing on venture and his walkthrough of building a personal app store on his iPhone.

→ Why does it matter? Anthropic's CFO walks through how they went from a $9B to a $30B run rate in a single quarter. Krishna Rao's view from inside: compute is the canvas, not a variable cost, and 90% of Anthropic's own code is now written by Claude Code.

Subscribe now

Quotes & eyewash:

→ Why does it matter? There is simply not enough time in the day to do all of the things you want (or should) do. What can you say “NO” to?

Subscribe now

The mission:

The Wall Street Journal once used “Read Ambitiously” as a slogan, but I took it as a personal challenge. Our mission is to give you a point of view in a noisy, changing world. To unpack big ideas that sharpen your edge and show why they matter. To fit ambition-sized insight into your busy life and channel the zeitgeist into the stories and signals that fuel your next move. Above all, we aim to give you power, the kind that comes from having the words, insight, and legitimacy to lead with confidence. Together, we read to grow, keep learning, and refine our lens to spot the best opportunities. As Jamie Dimon says, “Great leaders are readers.”

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, or legal advice. Readers should do their own research and consult with a qualified professional before making any decisions.

The big idea: Building AI-Native Products. Where’s the IP?

Reading time: 8 minutes

A strange thing is happening in software.

For most of my career, the origin of value creation was relatively clear. The company owned the source code. The code became the product. The product created the workflow. The workflow created the business.

AI has made that answer less obvious.

That is why I keep coming back to this topic. Not because I have the answer. I don’t. But because I think this is the question sitting underneath almost every AI-native product right now:

Where’s the IP?

If you’re one of our weekly readers, by now you might be thinking I’ve become a bit obsessive with this topic.

I think you’re right. It’s also for good reason.

Software is, was, or will be again a $4 trillion asset class. Software, now mostly delivered “as a Service,” has matured over the course of my career. I remember the early days of cloud, well before AWS, GCP, and Azure became the dominant hyperscalers. I remember when Salesforce.com and Workday paved the way.

Today, you can practically get a bachelor’s degree in building a SaaS company. The business model has a well-understood set of unit economics: CAC, LTV, net retention, gross margins, the Rule of 40, and other metrics that have helped create a common language for valuing software companies.

If a SaaS company delivered a product, found the market, and demonstrated efficient growth, investors felt reasonably confident in its ability to generate long-term, predictable cash flows.

AI has put that idea into question.

Some people call this the “SaaSpocalypse”: the notion that SaaS becomes less valuable as AI can write software, automate workflows, and compress the value of many point solutions. We’ve written about that before and believe that not all software is created equally.

And it is an important context.

Because my obsession with AI-native products is really a pursuit of understanding the fundamentals of how value gets created in the new era.

If the above is the well understood, time tested way to create value in SaaS? What is that in AI?

The harness is the product

Many of us assumed AI products might be monetized in ways that look familiar. SaaS companies launched AI features and priced them as an add-on.

But the pattern emerging for building and monetizing AI-native products may differ significantly from SaaS.

I am not talking about model producers like OpenAI or Anthropic. I’m talking about the application companies that rent the model through an API and then harness it to deliver predictable output.

Glean is a good example: a harness around Anthropic and OpenAI models designed to help enterprises find answers across their knowledge base, including Google Drive, Slack, and the other systems where work already happens.

There is now a rapidly growing ecosystem of AI-native applications companies doing something similar. These are the application companies built on this new technology.

What they are doing is harnessing the LLM and pointing it at a specific output.

The harness is the AI-native product.

At least today.

Think of it this way. You strap the LLM into an exoskeleton. Then you build everything into the suit that the AI needs to deliver reliable work.

The suit needs memory, so it can remember what matters.

It needs tools to know when to use a calculator instead of asking the model to guess.

It needs plug-ins and integrations, because real work doesn’t happen in a blank chat window. It happens in Google Drive, Slack, Salesforce, Outlook, Word, Excel, and the systems where the business already lives.

It needs proprietary data, so the model can be augmented with examples and documents that improve the quality of the output.

It needs permissions to know what it can touch, what it can change, and what needs approval.

It needs skills, so repeated work can be encoded into reusable procedures with your secret sauce baked in.

And it needs evals, so the system can learn the difference between output that sounds right and output that actually is right.

Building this exoskeleton is what R&D teams at AI-native application companies are working on. If you are building one of these products, technical breakthroughs across these areas, and the specific way they are combined, can add up to real competitive differentiation.

That is the theory.

But a strange thing is happening in legal AI that puts the theory under pressure.

A real-world test case

Legal AI is becoming the perfect test case because the work is valuable, text-heavy, high-stakes, and still deeply dependent on trust.

There is currently a race in the market between Harvey and Legora. Recently, Legora hired Jude Law to be its spokesperson.

Harvey has done an incredible job creating urgency in the legal industry, turning a profession that is typically late to adopt technology into one of the earliest enterprise adopters of AI.

Harvey recently raised $200 million at an $11 billion valuation. Legora raised $550 million at a $5.55 billion valuation. You might think they are running away with the market.

Then last week, former Latham & Watkins lawyer Will Chen launched Mike, an open-source legal AI platform. Mike describes itself as an open-source alternative to Harvey and Legora, with “feature parity,” zero cost, and a self-hostable codebase that law firms can own and extend.

Mike’s point was not subtle: much of the visible application layer can now be rebuilt and distributed for free.

To pile on, Microsoft introduced a Legal Agent directly inside Word, where much of legal work already happens today.

So it begs the question sitting underneath almost every AI application company right now, and those valuations:

Where’s the IP?

In SaaS, this question was easier to answer. The company owned the code. The code became the product. The product created the workflow. The workflow created switching costs. Customers paid for access.

AI-native software is harder to understand.

The model may belong to OpenAI, Anthropic, Google, Meta, or xAI. Those companies are making a lot of money, or at least they are today, although they also face pressure from open-source alternatives.

The interface may be easy to copy. The integrations may sit on top of APIs available to everyone. And the customer may not actually care about the software at all.

They may just care about the work getting done.

That is why the legal AI moment forces the harder question: is the value of the product in the visible application layer, or does it lie deeper?

This is where I keep getting stuck.

The easy answer is that the harness is the IP. I think that is directionally right. But it is also incomplete.

Because not every harness is equally defensible.

If the harness is just a chat interface, a project folder, a tabular review workflow, and a few saved prompts, then maybe the critics are right. Maybe that application layer is thinner than we thought. Maybe it can be rebuilt faster than investors want to admit. Maybe Microsoft can absorb it into a Word feature. Maybe an open-source project can give customers enough of the visible product to make the premium vendor harder to justify.

That is the uncomfortable version of the argument.

But there is another version.

Maybe the visible application layer is not the real product. Maybe it is just the part we can see.

In most serious enterprise workflows, the value is not just the screen. It is the accumulated understanding of how work gets done: the weird edge cases, approval paths, partner preferences, security model, integrations, and trust required to put the system into production.

That stuff does not show up cleanly in a demo. It sits underneath the iceberg.

In enterprise software, the visible product has never been the whole story. The real product is often the system of work behind it.

AI makes that more true, not less.

A good harness does not just help the model answer questions. It teaches the model how a specific kind of work should be done. It encodes the difference between a plausible answer and an acceptable one. It knows what should never be touched without a human in the loop.

That is a different kind of product.

It is less like selling software and more like building an operating system for a category of work.

The question is whether that operating system becomes defensible enough to support large standalone companies, or whether the value gets pulled apart by platforms, model companies, open source, and customers building for themselves.

I do not know the answer.

But I think this is the right question.

Because the next generation of AI application companies will be judged on whether they can turn that output into a repeatable, trusted, governed workflow that customers are willing to depend on.

That is where the moat may get seriously deep.

At the point where the system becomes trusted enough to own part of the work.

The “business” of AI

This is the part I keep coming back to.

In SaaS, the product was the software. In AI-native companies, the product may be the harness. But the business of AI may be delivering the outcome.

That distinction feels important because the buyer may not actually want another application.

The buyer wants the work done.

There is an old adage:

“People don’t want to buy a quarter-inch drill. They want a quarter-inch hole.”

A law firm needs documents reviewed, contracts redlined, and work product delivered to a standard that partners and clients will trust.

A CFO does not need another accounting copilot. She needs the books closed.

An insurer does not need another claims assistant. It needs claims to be processed accurately.

This is where the IP question starts to move beyond software.

Maybe the most defensible AI-native companies will not just sell the harness. Maybe they will use the harness to do more of the work.

My friend Sean O’Connell, who is fittingly a lawyer, came up with a phrase for this: tech-shoring.

Offshoring moved labor to a cheaper geography. Tech-shoring moves labor into a technology-mediated system. The work is still supervised by humans where judgment, trust, and accountability matter. But the leverage comes from the AI and how it is harnessed.

This is not pure software. It is not traditional managed services. It is not simply giving people a copilot and hoping productivity improves.

It is using software to deliver the service itself.

That changes the business model.

If you sell the harness, the customer can ask whether they should build it. If you sell the outcome, the customer has to ask a different question: Can you deliver the work better than I can, and are you willing to stand behind it?

Better may mean cheaper. It may mean faster. It may mean more consistent. It may mean more auditable. It may mean less dependence on hard-to-find labor. It may mean the service improves over time because every job teaches the harness something new.

That is a more interesting moat because the value is not just in the tool. It is in the operating system around the work: the process, the data, the human review, the liability, the quality control, the service levels, the customer-specific knowledge, and the feedback loop from doing the work repeatedly.

This is where the IP may start to look less like software IP and more like institutional capability.

That is harder to see from the outside.

It is also harder to copy if it compounds.

Why it matters

We are still using an old software map to navigate a new market.

In SaaS, the questions were relatively familiar: Who owns the code? How sticky is the workflow? What is the retention? How efficient is the growth?

In AI-native products, the questions are different.

If the model is rented, the interface can be copied. The company has to prove it owns something deeper than the screen.

That might be a harness that compounds. It might be proprietary workflows. It might be customer-specific data. It might be trust, governance, or liability. It might be the ability to deliver the outcome better than the customer can.

For builders, this changes what is worth building.

For buyers, it changes what is worth buying.

For investors, it changes what is worth valuing.

The easy answer is to call everything a wrapper. The harder answer is to understand which systems are merely wrapping the model and which are learning to own the work.

The most valuable AI-native companies may not be the ones with the best demo. They may be the ones who turn rented intelligence into work.

The model may be rented. The harness may be the product. The outcome may be the business.

And the IP may live in the system that connects all three.

Stay ambitious, my friends.

Subscribe now

Best of the rest:

🤖 DeepSeek nears $50 billion valuation in first fundraising round — China is signaling serious intent in the global AI race, with state-backed capital and tech giants lining up behind a breakout model player that could reshape competitive dynamics. — Reuters

💸 Corporate Card Startup Ramp Raising Funds at $40 Billion Valuation — Ramp’s reported $750 million raise shows investor appetite for AI-powered finance automation is still running hot, with fintech’s category winners continuing to pull away from the pack. — The Wall Street Journal

🧠 Do I belong in tech anymore? – A raw, searching essay on AI burnout and the collapse of tech’s old self-image, arguing that speed without care is hollow because the real work of building software is human judgment, friction, and craft. – ky.fyi

⚙️ A.I. Should Elevate Your Thinking, Not Replace It – A sharp line in the sand for the AI era, arguing that the best engineers will use AI to remove drudgery and deepen judgment, while the ones who outsource thinking will quietly make themselves irrelevant. – Koshy John

Charts that caught my eye:

→ Why does it matter? Financial services went from a rounding error to Anthropic’s #2 vertical in about 18 months, with 40% of top customers being financial institutions. Banks tend to be slow buyers of frontier tech. Not this time.

→ Why does it matter? Talent density is everything when capital is abundant. Paraform’s ranking is a view into where the market thinks the best builders are clustered.

Tweets that stopped my scroll:

→ Why does it matter? Sierra went from four design partners to 40% of the Fortune 50 in roughly two years. And now they’re raising a $1B warchest. Also, Bret is the current chairman of OpenAI and is seen as a logical successor to Sam Altman.

→ Why does it matter? Perplexity Computer is an always-on AI harness, and this week, they’re announcing a new offering focused on Financial Services, giving access to datasets from MorningStar, Daloopa, and beyond. This is a direct shot at Bloomberg Terminal, and it’s compelling.

→ Why does it matter? Sam Altman texts then-OpenAI CTO Mira Murati during the BOD meeting where he was ousted.

→ Why does it matter? Incredible growth from Palantir. A nice architectural overview of AIP, their AI-focused front-end to the Palantir platform.

Worth a watch or listen at 1x:

→ Why does it matter? You’re doing something right if your first customer conversation is with Jamie Dimon. Anthropic laid out a compelling vision for financial services, its second-fastest-growing market behind technology, and made clear why the category matters: high-stakes workflows, dense information, and customers with both the urgency and resources to put AI to work.

→ Why does it matter? A master class in presenting company strategy from Bill McDermott and the ServiceNow team, with cameos from Jensen Huang and Anthropic’s lead engineer on Claude Code. ServiceNow told a clear story about where enterprise AI is headed and why workflow sits at its center. Financial Analyst Days are always incredible forcing functions. They make companies do the hard work of turning strategy into story, and story into conviction.

Subscribe now

Quotes & eyewash:

→ Why does it matter? Build a high agency team and strive for intellectual honesty. A few years ago, I stumbled upon this great book on the subject: Mistakes Were Made (but Not by Me): Why We Justify Foolish Beliefs, Bad Decisions, and Hurtful Acts.

Subscribe now

The mission:

The Wall Street Journal once used “Read Ambitiously” as a slogan, but I took it as a personal challenge. Our mission is to give you a point of view in a noisy, changing world. To unpack big ideas that sharpen your edge and show why they matter. To fit ambition-sized insight into your busy life and channel the zeitgeist into the stories and signals that fuel your next move. Above all, we aim to give you power, the kind that comes from having the words, insight, and legitimacy to lead with confidence. Together, we read to grow, keep learning, and refine our lens to spot the best opportunities. As Jamie Dimon says, “Great leaders are readers.”

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, or legal advice. Readers should do their own research and consult with a qualified professional before making any decisions.

The big idea: The next killer app is AI computer use

Reading time: 8 minutes

The next killer application for AI may sound like a bad idea.

Give it access to your computer.

Access to the actual machine where knowledge work happens: your browser, file system, productivity tools like Excel and Word, inbox, calendar, and all the half-finished context that makes modern work feel modern.

That sounds reckless.

It may also be the most logical next step toward realizing AI’s full potential.

It’s called computer use. The AI uses a keyboard, mouse, browser, and operating system just like humans do.

If AI is going to automate knowledge work, it eventually has to get “hands on” with the knowledge worker’s primary tools.

The strange part is not that this is technically possible, although the first time you see AI driving around your computer, it’s mind-boggling.

The strange part is that it may make sense to let AI operate the machine where work already happens.

The factory floor of knowledge work

The Industrial Revolution did not eliminate work. It reorganized it.

Labor moved from farms to factories, then from factories to offices. The modern knowledge worker was not born in opposition to automation. In many ways, knowledge work was created by it.

Today, the factory floor of knowledge work is a screen.

The machinery is familiar: inboxes, spreadsheets, decks, agendas, tabs, calendar invites, PDFs.

This is where most work happens.

And the dominant provider of the tools for this work is Microsoft, anchored in Office and the broader Microsoft 365. Microsoft 365 has more than 80% enterprise market share, more than 500 million paid seats, and more than $80 billion in revenue.

Paul Graham once called Microsoft “a mountain that can walk.”

AI computer use is not circling a small category. It is circling this mountain.

And this makes a lot of sense. The AI is moving out of the chat window and into the environments where real work happens.

This has already proven to work in agentic coding, which is nearly perfect terrain for AI: codebases, terminals, tests, logs, and GitHub.

Knowledge work is messier. It does not have the cleaner, more structured world of software development.

It has the computer.

The first wave of products is here

For decades, productivity software assumed humans pushed the buttons. Users opened the application, wrote the words, attached the file, updated the spreadsheet, and sent the email.

Computer use gives AI the ability to press the buttons.

Humans still provide the desired outcome, but the AI performs as many steps as possible.

Claude Cowork is a great example. It moves beyond the prompt window and starts working across your files, browser, folders, and applications to complete tasks autonomously.

Microsoft recently released its competitor: Copilot Cowork.

Other examples include Perplexity Computer, Factory’s Droid Computers, OpenClaw, and Hermes. Just yesterday, Manus (acquired by Meta) launched their competitor.

And, of course, OpenAI is going with Codex here. In this demo, you can actually see the AI using a mouse.

Different products, same idea.

Move the AI into an “always on” position and into the place where work happens.

The harness is the product

A framework for building these types of AI computer-use products is coming into clarity.

A few weeks ago, we wrote that the harness is the product. A 2x user and a 100x user can be using the same AI model. The difference is the harness.

The harness is the program that wraps the model and helps it do useful work. A good harness gives the model context, tools, and memory management. It decides what belongs in the context window and when. It decides what the model should handle and what should be routed to more precise software.

With a good harness, the AI can make better decisions to achieve accurate, high-quality outcomes.

A simple example is arithmetic.

It makes no sense to ask the AI to do arithmetic when a deterministic calculator already exists. The AI should know when to call the calculator. It should know when to query the database. It should know when to use the browser, open the file, run the script, or ask the human for approval.

That is the difference between a model with access to tools and a system that knows how to work.

The model is useful for judgment: reading, synthesis, and classification.

Deterministic tools are useful for precision: queries, calculations, and workflows where the same input should produce the same output every time.

The worst agent systems confuse the two. They ask the model to do arithmetic. They force it to reason through tasks that should have been handled by software.

The best systems push intelligence up into skills and push execution down into deterministic tools.

This is how AI computer use becomes a lot more than AI clicking around on your screen.

It becomes a system that lives in the knowledge worker’s environment and can get work done.

Building a moat into the wrapper

If the model is not the whole product, the next question is where the IP lives.

• It may live in the harness: tooling, memory, permissions, routing, evals, and integration.

• It may live in the skills: reusable procedures that encode how work gets done.

• It may live in the proprietary datasets: examples, corrections, approvals, exceptions, and workflows that teach the system what “right” looks like.

• It may live in the integrations: the ability to touch the systems you need to drive outcomes.

• It may live in the tacit knowledge that gets captured over time: your deep understanding of the domain, the judgment embedded in your workflows, the “secret sauce” of how you do what you do.

This is not traditional software IP. The crown jewels used to be the code.

This is not code.

It is institutional knowledge that augments the intelligence of the AI.

A skill is a reusable procedure. It captures the process, judgment, and context required to perform a task. The user supplies the specific request. The skill supplies the method.

Imagine a skill for investigating a company. It might tell the model to pull filings, build a timeline, compare management commentary against financial results, identify contradictions, summarize both sides of the argument, and cite the sources.

The same skill can be applied to different companies, questions, and datasets.

That is not prompt engineering. It is software design in a language the model understands.

If the AI can help perform the work, observe the pattern, and codify the process, the workflow itself becomes an asset.

Trust is essential

By now, you’ve probably formulated an opinion on whether you’d give AI access to your computer.

My hunch is that most readers are a “no.”

You could buy a new computer and give the AI its own. That’s what I’m doing. The base Mac mini was recently unavailable in Apple’s online store, and TechCrunch reported that marked-up Mac minis were appearing on eBay amid demand for machines that can run on-device AI.

Giving AI access to your computer is not a small permission. It is like giving a junior employee your laptop, your badge, and your password, then telling them to log in and help you get things done.

That employee may be capable. They may also misunderstand the task, click the wrong button, expose sensitive data, or post the wrong thing in the wrong place with confidence.

In the human world, we manage that risk with roles, permissions, training, audit trails, and supervision. In security, ideas like Zero Trust start from a similar premise: do not assume trust just because someone or something is inside the system. Verify access. Limit permissions. Require the right approval at the right time.

Maybe unglamorous, but AI will need this version of trust machinery for both consumer and enterprise usage.

And this is a big reason enterprise incumbents have a leg up. They have already built much of the machinery that makes software usable in real companies.

Computer use will raise the bar here.

The AI cannot just use the computer. It has to use the computer as someone, with a specific role, a specific set of permissions, and a specific boundary around what it can and cannot do.

Without that, it is not a product that companies can trust.

It is a very impressive prototype.

From productivity tools to supervising the work

The knowledge worker’s current tools give humans what they need to do the work.

Documents. Spreadsheets. Slides. Email. Calendars. Chat.

The new productivity layer may use all of these tools. It may not replace Excel. It may use Excel. It may not replace Outlook. It may use Outlook.

This is the strategic tension for Microsoft and Google. They may be mountains, but the terrain around them could change quickly.

The question is not whether AI gets added to productivity software. It will. The question is whether the knowledge worker’s primary tools remain the place where work happens, or become the surface on which AI operates.

For decades, the model was simple: learn the software, push the buttons, produce the work, share it.

The next model may be different: describe the outcome, watch the system work, approve what matters, share it.

From pushing the buttons yourself to supervising the work.

We may have to give the AI access to a computer to get there.

Stay ambitious.

Subscribe now

Best of the rest:

🤖 OpenAI Misses Key Revenue, User Targets in High-Stakes Sprint Toward IPO — OpenAI’s growth story is colliding with the brutal economics of compute, as missed user and revenue targets raise harder questions about whether frontier AI can scale fast enough to justify its infrastructure arms race. — The Wall Street Journal

💰 Sources: Anthropic could raise a new $50B round at a valuation of $900B — Anthropic’s reported mega-round shows the AI capital race shifting from model capability to balance sheet capacity, with Claude’s coding momentum turning investor demand into a direct challenge to OpenAI’s scale advantage. — TechCrunch

🛡️ Our evaluation of OpenAI's GPT-5.5 cyber capabilities – AISI’s latest cyber eval shows GPT-5.5 pushing into frontier-risk territory, with one of the strongest performances they’ve measured and a rare end-to-end solve of a multi-step attack simulation. – AISI

⚠️ Cursor's warchest, xai's redemption – Ethan Ding argues Cursor’s rumored sale to xAI is less a victory lap than a warning shot for the AI application layer, where even breakout products can get squeezed when their model suppliers decide to own the margin pool. – mandates

Charts that caught my eye:

→ Why does it matter? Acquirers are paying scarcity prices for AI Native companies. At a 17.7x median and half the deals clearing 20x, buyers are betting that quality AI assets won’t get cheaper.

→ Why does it matter? AI labs built consumption pricing from day one. Traditional SaaS spent decades optimizing seat contracts, and that model is sticky: 3.8% consumption share after 11 months of AI pressure.

→ Why does it matter? Ten companies are now larger than the entire G7 national economies. Eight of those ten are American tech firms.

→ Why does it matter? Half the new $5B+ private companies in 2025 are AI apps, with the application layer (Cursor, Cognition, Harvey) clustering at the top. Capital is rewarding the layer closest to end users, not just the model labs underneath!

→ Why does it matter? The spike lines up almost exactly with ChatGPT’s public release. LLMs love dramatic rhetorical pivots and engagement-seeking styles. Your earnings calls may already sound like everyone else’s.

Tweets that stopped my scroll:

→ Why does it matter? Jane Street generates $8M in profit per employee, with 3,500 people. 🤯

→ Why does it matter? OpenAI building a phone with Luxshare (Apple’s own manufacturer) signals ambition well beyond ChatGPT and Codex. And they, of course, have Jonny Ive!

→ Why does it matter? Telling people they slept well (even falsely) measurably improved their cognitive test scores. What you believe about your condition shapes your actual performance, not just your mood.

Worth a watch or listen at 1x:

→ Why does it matter? CEO’s have the mandate from their BODs to figure out AI. A pleasure to join Marley Kayden and Sam Vadas on Schwab’s Market on Close to discuss a wide range of topics, including Ridgeline’s AIOS for Asset & Wealth Management, AI computer use, the “it” in AI is the data set, and the SaaS apocalypse. Complete with Harry Potter references.

→ Why does it matter? Karpathy reframes LLMs as ghosts rather than animals: jagged, statistical, summoned. He also separates vibe coding from agentic engineering and argues you can outsource thinking but not understanding.

→ Why does it matter? Tudor Jones walks through the trader-versus-investor split, lessons from 1987 and the 1980 silver collapse, and his current read on the debt bubble and AI risk. This is a wide-ranging conversation with an industry legend. Must watch!

→ Why does it matter? Dwarkesh is on a hot streak (NYT profile this week), and this episode is a good example of why. Reiner Pope walks through how frontier models actually get trained and served, and a striking amount of it reduces to a handful of equations.

Subscribe now

Quotes & eyewash:

→ Why does it matter? I pre-ordered David Epstein’s new book, Inside the Box, this week. Epstein’s previous bestseller, Range, made the case for generalists. His new book asks a different question: why do limits often produce better work? That feels relevant to AI right now. Look at DeepSeek. The company did not have the cleanest supply chain, the largest American-style lab budget, or unfettered access to Nvidia’s best chips. And yet, DeepSeek trained V3 on 2,048 H800 GPUs and claimed performance comparable to leading closed-source models at a fraction of the cost. The question is not whether DeepSeek is the best model in the world. It isn’t. The question is: why did a constrained team get this close? One answer: scarcity forced better architecture, better training efficiency, and a sharper focus on what mattered.

Subscribe now

The mission:

The Wall Street Journal once used “Read Ambitiously” as a slogan, but I took it as a personal challenge. Our mission is to give you a point of view in a noisy, changing world. To unpack big ideas that sharpen your edge and show why they matter. To fit ambition-sized insight into your busy life and channel the zeitgeist into the stories and signals that fuel your next move. Above all, we aim to give you power, the kind that comes from having the words, insight, and legitimacy to lead with confidence. Together, we read to grow, keep learning, and refine our lens to spot the best opportunities. As Jamie Dimon says, “Great leaders are readers.”

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, or legal advice. Readers should do their own research and consult with a qualified professional before making any decisions.

The big idea: How I use AI in my daily life

Reading time: 10 minutes

The most frequent question I get from friends, family, and readers is simple:

“Jack, how do you actually use AI?”

A friend asked me that on Saturday night, standing at the kitchen counter.

What this person meant, though, is something deeper. They do not want to fall behind.

I think a lot of people feel that right now. The world is moving quickly, AI is suddenly everywhere, and there is a quiet fear that everyone else has already figured it out.

They haven’t.

The data makes that clear. Gallup now shows that half of U.S. employees use AI in their role. But only 28% use it frequently, and just 13% use it daily. It is still early. Very early.

Most people are still using AI like a slightly better version of Google.

That alone is useful. I do it too. This morning I asked, “How many days until Mother’s Day?”

But AI is much more powerful than that. The more interesting shift happens when AI stops being a better search box and starts becoming your assistant. When it starts taking real work off your plate.

A lot of people may have tried AI now. Very few are using it anywhere close to its potential. In fact, the group using Agentic Coding is tiny, about 0.04% of the world’s total population.

So, good news. We are still super early and can learn this together.

Getting Started

By now, I assume most readers have tried at least one of the major AI products: ChatGPT, Claude, Gemini, or Grok.

I use all four.

ChatGPT is my all-around daily driver. Best overall product experience across desktop and mobile, broad tooling, strong multimodal capabilities, meaning text, photos, and video, and the easiest place to start.

Gemini is especially useful if you use Google Workspace. Gmail, Docs, Sheets, and Drive are all right there. Its image generation, Nano Banana, is also excellent.

Claude, and especially Claude Code, is where I go when I want to build software or create workflows that persist beyond a chat window. It is the difference between getting an answer and building a working system.

Grok is useful when I want a real-time read on the internet, especially around X, headlines, and what people are actually talking about right now.

You do not need all four. You just need to understand what each is good at.

One practical tip before anything else: turn on voice.

The best way to interact with an LLM is often to talk to it. MacOS has native voice dictation. The apps do too. There are also paid tools like WhisperFlow that are even better. Over the course of a day, I often record a two- or three-minute stream-of-consciousness voice memo just to get the model started on my thinking.

Context and Prompts

Before I start any real project with AI, I think about two things: context and prompts.

Start with context.

The model can only work with what it can see. So before I ever type anything, I ask myself: what would I put on the table if I were doing this project manually?

A spreadsheet. A few notes. A draft. An email thread. A PDF. Maybe some screenshots. Maybe a voice memo.

I think of context as the model’s working memory for the task at hand.

I often create a folder on my desktop and put everything in it before I start. Then I upload the whole thing to the model. If I want a good answer, I need to give it a good view of the problem.

Think of it like setting up shop at a library table. You want the right books open before you start working.

One small habit that helps: when I switch topics, I usually start a new chat. Context is useful until it becomes a distraction.

Then come prompts.

Yes, better prompts usually lead to better output. But people overcomplicate this.

One of the best prompting tricks I know is to ask the AI to help me write the prompt. Open a fresh chat and say: “I want your help writing a prompt for this project.”

Then paste in your rough notes, or, better yet, use a 2-minute voice memo to talk through what you are trying to do.

You will be surprised how good the model is at turning messy thinking into a clear instruction set.

When people ask how to begin, my advice is simple: start with work you already understand.

The best first projects are usually in areas where you already have expertise. That is how you build confidence. You know what good looks like, so you can tell whether the model is actually helping.

In many ways, what I built with AmbitiousOS is basically Reading Ambitiously turned into software.

That is a useful way to think about AI more broadly. Start with tasks where you already know the destination. Let the model help you get there faster.

The hardest part of AI is not understanding the theory. It is seeing where it fits into the flow of an ordinary day.

Which is why I like starting with a to-do list.

The To-Do List

I love a good to-do list. I make a lot of them.

I have tried every application out there to help me organize and make lists. And maybe it is the technology overload of 2026, but these days I still like to do my best thinking with a pen and paper, away from the electrons.

There is an old story about a man who walked up to J.P. Morgan and said, “I have the secret to success in this envelope, but it will cost you $10,000.” Morgan asked to see it first. He opened the envelope. Inside was a single line: “Make a list and do it.” Morgan shook his head and paid the man.

That is usually where my day begins, with a pen, a legal pad, and a list of what is actually on my mind.

On Monday, it looked something like this.

Before AI helps me do the work, I need to know what the work actually is. A handwritten list helps me think through it.

Then I make a simple move. I take a picture of the list, drop it into ChatGPT, and say: “Turn this into a plan. Do the first pass on anything you can.”

That is the first handoff to AI.

From there, each line item starts to change shape. One becomes a research brief. One becomes a prioritization engine. One becomes a cleanup job I have been putting off. And in a few cases, once I know the workflow is worth repeating, if I need to move from ChatGPT into Claude Code and build something more durable, I will.

Here is what that looked like across a real day.

1. 451 Attendees, 20 Conversations That Matter

A spreadsheet with 451 names is not a plan. It is just a lot of rows.

So I start by giving ChatGPT the file and a little context on the conference, who I want to meet, and what a good outcome would look like.

“Hey ChatGPT, I’m headed to a conference and 451 people are registered. Help me break this list down, prioritize the people I should care most about, find their LinkedIns, and give me a short note on why each one matters. Then draft outreach for the top group.”

I am no longer staring at 451 names. I am reviewing 20 decisions.

If I want that to become a reusable system, with enrichment, scoring, and tracking, that is when I move it into Claude Code.

2. The Meeting Brief

This is one of my favorite ChatGPT use cases because it is so close to how a real assistant should work.

I gather the emails, notes, research, and background materials, then I hand over the problem.

“Hey ChatGPT, I have an important meeting coming up. I’m uploading the background here. Put together a brief for me. Give me the key themes, what has changed recently, what I should watch out for, five questions I should be ready for, and a draft of the point of view I want to open with.”

What comes back is usually a much more complete view of the room I am walking into.

This one is mostly ChatGPT end-to-end.

I treat much of this the same way I would a smart junior analyst. Incredibly useful for a first pass. Still worth checking before I bet something important on it.

3. The Mailing List, or the Work You Avoid

This is the kind of task that sits around longer than it should.

An old spreadsheet. Outdated addresses. Random fixes scattered across texts and emails. Not hard work, just annoying work.

“Hey ChatGPT, this mailing list is a mess. I’m giving you the old Excel file, some emails, and text updates. Reconcile everything you can, clean it up, standardize the format, and flag anything that still needs a human look. Oh, and please give me the final output in XLS. Thank you.”

What comes back is a cleaned-up file. Instead of being something I will get to eventually, it gets finished.

4. Trip Planning, From Searching to Steering

Travel is a perfect example of something AI handles well because the inputs are clear and you have to do a ton of research.

Dates. Preferences. Budget. Meetings. Restaurants. Logistics. Too many Chrome tabs.

“Hey ChatGPT, I need help planning a trip. Here are my dates, where I need to be, who’s coming, and what matters most. Give me a plan. Start with flights and hotel options, then help me map the trip so I’m not wasting time bouncing around.”

What comes back is not a final answer so much as a strong first draft.

5. The Follow-Up, Where Value Is Won or Lost

This is where good work often dies. Not in the meeting, but after it.

You have notes, business cards, partial memory, and every intention of following up well.

“Hey ChatGPT, I just got back from a conference. I’m uploading my notes and the people I met. Help me group them by priority, draft follow-up emails, and suggest the right next step and timing for each one.”

What comes back is a system where there used to be a pile.

People get grouped. Emails get drafted. Next steps get named. Nothing falls through the cracks.

Where Each Tool Fits

When people ask how I use AI, they are usually asking about which tool.

The more useful way to think about it is layers.

Some work is about thinking. Some is about doing. And some is about building something you can reuse.

ChatGPT is where most workflows begin and often end. It is my default for research and synthesis, structuring messy problems, drafting and summarizing, and getting to a strong first version quickly.

That is why it shows up so clearly in the meeting brief and trip planning. Those are thinking problems.

Claude Code shows up when the work needs to persist.

When something moves from a one-time task to something I will do again, I start to think differently. Can this become a tool? Can this run the same way every time? Can I remove myself from the loop entirely?

That is where Claude Code becomes useful.

The conference list if I want a reusable prioritization system. The mailing list if I want a repeatable transformation pipeline. The follow-up process if I want a system that tracks status and cadence.

ChatGPT gets me to the answer. Claude Code helps me build the system.

Most people do not need to start with Claude Code. They need to start with ChatGPT or Gemini and get comfortable handing off pieces of their work. But once a workflow proves valuable, the next question becomes obvious: should this be a one-off, or should it run the same way every time?

That is where the second layer begins and where Claude Code can help you turn that conference attendee list into an AI-based system you can use every time.

Why it matters

When I step back, three things consistently show up: context, clarity, and tools.

Context means gathering everything into one place. Documents, notes, files, screenshots, voice memos. The model works best when it has the full picture.

Clarity means not overengineering prompts. Explain the task, what good looks like, and where you need help.

Tools means accepting that you are not choosing one product. You are assembling your personal AI stack. ChatGPT is my default. Grok for real-time awareness. Gemini for images. Claude Code, when I want to build something repeatable.

Half of people may have tried AI. Most are still using it like Google 2.0.

The shift is learning what can be handed off and what is worth turning into a system. That second step is where things start to compound.

To my friend at the kitchen counter, the answer is not really which AI tool to use. It is time to think differently about your work.

Start with your list and ask a simple question: What can the AI do for me?

Stay ambitious.

Subscribe now

Best of the rest:

🔐 Anthropic’s Mythos Model Is Being Accessed by Unauthorized Users — If frontier models can leak before controlled rollouts even begin, the real bottleneck in AI may be security and distribution, not just raw capability. — Bloomberg

⚙️ Thin Harness, Fat Skills — The real unlock in AI coding isn’t smarter models but better architecture, where lightweight wrappers and reusable “skills” turn LLMs into compounding systems rather than one-off tools. – Garry Tan

🍎 Community Letter from Tim – Tim Cook’s handoff to John Ternus is more than a succession note, it is Apple signaling continuity, stability, and a product-led future at one of the world’s most important companies. – Apple

🤖 Sign of the future: GPT-5.5 – Ethan Mollick’s real point is not just that GPT-5.5 is better, but that the compounding gains across models, apps, and tooling are making AI meaningfully more useful even as the jagged frontier still has not disappeared. – One Useful Thing

🏢 Inside BlackRock’s AI Transformation - BlackRock is moving from AI experimentation to enterprise-wide execution, with a no-code agent platform that lets employees across the firm build and deploy agents, a sign that the real AI race inside large institutions is shifting from models to workflow redesign. - The Wall Street Journal

🚀 xAI Explored Collaborating With Mistral and Cursor — Musk’s AI ambitions are colliding with reality, forcing xAI to look outward for partnerships as the frontier race consolidates around a few dominant players. — Business Insider

☁️ Google Deepens Thinking Machines Lab Ties With New Multi-Billion-Dollar Deal — Google is locking in scarce AI demand early, betting that owning the compute layer for next-gen labs like Mira Murati’s will pay off as model training scales exponentially. — TechCrunch

🧠 The peril of laziness lost — LLMs are supercharging output but eroding the core engineering virtue of thoughtful abstraction, risking a future of bloated systems optimized for speed over simplicity. – Bryan Cantrill

Charts that caught my eye:

→ Why does it matter? All-time highs feel like bad entry points. The data disagrees. Buying at peaks has historically beaten buying on a random day across every time horizon shown, which suggests “wait for a pullback” is often just dressed-up hesitation.

→ Why does it matter? At 75%, Google is approaching a point where human engineers review AI code more than they write it.

→ Why does it matter? The public SaaS market has effectively split into two economies: 96 companies grinding below 10x, and 7 commanding 10x or more. The separating factor looks less like growth and more like AI relevance. If your product isn’t in the labor-replacement conversation, the market is pricing you like a legacy business.

Tweets that stopped my scroll:

→ Why does it matter? This one lives rent-free in my head often: if FTX had not collapsed, its startup investment portfolio would be worth over $100 billion today. 🤯

→ Why does it matter? Thoma Bravo paid $6.4B for Medallia in 2021 and is handing it over to creditors, leaving nothing for equity. 2021 was a tough vintage.

→ Why does it matter? Perplexity’s AI computer looks genuinely so cool. Cannot wait for my Mac Mini to arrive.

→ Why does it matter? Kimi 2.6 is the first open-weight model competitive with top closed models on agentic tasks, at cheaper inference costs. I truly think open source could be the black swan in this whole thing.

→ Why does it matter? Enterprise AI adoption is a services problem as much as a software one. Legacy systems, fragmented data, and change management mean consultants.

→ Why does it matter? OpenAI is moving ChatGPT from a personal tool to a team infrastructure play. Shared agents that persist across workflows put it in direct competition with Zapier, Notion, and enterprise automation software. Rumor has it that Anthropic might be buying Atlassian for this reason.

Worth a watch or listen at 1x:

→ Why does it matter? Rabois has strong opinions on which AI startups are building on borrowed time and which archetypes will actually survive. Many hot takes from someone who has seen enough cycles to back them up.

→ Why does it matter? OpenClaw creator Peter Steinberger takes us back to the transformative moment he let his AI agent loose on the internet, igniting one of the world's fastest-growing open-source projects.

→ Why does it matter? This was one of the most talked-about 1-minute segments of any podcast in the past ~10 days.

→ Why does it matter? Can’t wait!

Subscribe now

Quotes & eyewash:

→ Why does it matter? Seven Dunkin stops turn a legend into a commute. Have to love the USA.

Subscribe now

The mission:

The Wall Street Journal once used “Read Ambitiously” as a slogan, but I took it as a personal challenge. Our mission is to give you a point of view in a noisy, changing world. To unpack big ideas that sharpen your edge and show why they matter. To fit ambition-sized insight into your busy life and channel the zeitgeist into the stories and signals that fuel your next move. Above all, we aim to give you power, the kind that comes from having the words, insight, and legitimacy to lead with confidence. Together, we read to grow, keep learning, and refine our lens to spot the best opportunities. As Jamie Dimon says, “Great leaders are readers.”

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, or legal advice. Readers should do their own research and consult with a qualified professional before making any decisions.

The big idea: What I am teaching my kids about AI

Reading time: 7 minutes

My first-grader and I have been building software to help her learn to spell and type.

She has been doing more of the heavy lifting than you might think. She is six. Give her a whiteboard and Claude Code, and she can now move from idea to software fast.

What fascinates me most, as her dad, is the combination. A six-year-old’s imagination is already wild. Claude turns that imagination into something tangible.

Today, we have a Trolls-centric, Sing 2-themed spelling game with different modes, levels, and mini-game rewards, including her own version of Doodle Jump. And while we wait for Claude to write the code, what do you think we do? We spell the words. Again and again. She has been getting 10 out of 10 on her spelling tests.

When a close friend recently asked me what I am teaching my kids about AI, I found myself coming back to this story.

Not because I think every six-year-old should be building software. And not because I think AI will make the hard parts of learning disappear. It will not. But something has changed. The distance between imagination and execution is shorter than it used to be.

The edge may belong less to the strongest or the smartest, and more to the children who learn how to respond to change without losing themselves, ask better questions, and turn ideas into something real.

That is what I find myself wanting to reinforce at home.

Not a fixed view of which job titles will matter most. Not a prediction about which degrees will age best. Something more basic. A responsible relationship with powerful technology. Character. Curiosity. Communication. Resilience. Judgment.

The new street smarts.

A responsible relationship

I spend a lot of time thinking about what a responsible relationship with technology actually looks like.

With phones and social media, the parental instinct is delay. Hold the line. Buy more time. As a founding member of the Wait til 8 pledge at our daughter’s school, I know that instinct well. It makes sense to me.

But AI feels different.

You may be able to delay a smartphone. I doubt you will be able to delay AI in the same way. AI will not live in one app. It will show up in the software our kids use at school, the devices in our homes, the cars we drive, and the tools they use for work. Software is already everywhere. AI is going to spread through that same surface area.

Which means the job is not just delay. It is preparation.

And preparation, at least to me, does not mean making every decision for them. It means giving them age-appropriate room to make decisions, feel the weight of those decisions, and learn from the outcomes. That is how street smarts are built. Not through lectures alone. Through giving it a go. Through small choices, some good and some bad, while the stakes are still low enough for a parent to help them make sense of what happened.

Too often, what we call protection is really substitution. We make the decision for the child, and then wonder why judgment never strengthens.

I keep coming back to the idea of street smarts. My mom has them. She grew up on the South Side of Chicago. Street smarts meant knowing how to read a situation before it turned on you. Keep your head up. Know who you are with. Pay attention to tone. Leave early when something feels off.

Our kids need a modern version of that.

Not fear. Not techno-doomerism. Good judgment.

They need to know when a confident answer is still wrong. They need to know what information should stay private. They need to know when a tool is helping them think, and when it is doing the thinking for them. They need to know when speed helps, and when speed hides a mistake. They need to know when to close the laptop and ask a human being.

I am nowhere near a complete list. But a few traits keep bubbling to the top, mostly because they help kids build judgment.

The new street smarts

Character.
Every morning on the drive to school, we walk through the Dad Rules. Be kind. Remember people’s names. Treat others how you want to be treated. Do the right thing when nobody is looking.

I ask them because repetition matters. I also ask them because no technology will ever remove the need for character. Powerful tools raise the stakes. AI can help you move faster. It can also help you cut corners, hide bad intent, and avoid responsibility.

Before I care what my kids can do with a tool, I care who they are when they use it.

Curiosity.
A few weeks before we built the spelling game, my oldest walked into my office and pointed at the little door in the ceiling.

“What is that?”

“The attic,” I said.

“Can we go up there?”

“Absolutely. Right now.”

I dropped what I was doing and grabbed the ladder.

Curiosity has a short half-life if adults keep putting it off. Later today. This weekend. Not now. Do that often enough, and a child learns that wonder belongs behind more practical things.

I do not want to teach that lesson by accident.

The world their generation inherits will not hold still for long. The kids who do well in it will be the ones who stay open long enough to notice what changed, ask what it means, and try something new.

Communication.
Speak well. Write clearly. Listen closely. Say what you mean.

More and more, the interface is natural language. The person who can describe what they want, frame the problem cleanly, and hear what is missing has leverage.

My daughter was not just learning words. She was learning how to describe what she wanted to Claude Code. She was learning how to point to an idea that did not yet exist and make it legible enough to build.

There is nothing soft about that skill.

Resilience.
The happy story about AI is that you type a prompt, get a miracle, and move on with your life.

That has not been my experience. The model still gets things wrong. The first version still disappoints. Building software is still messy.

That is part of why I like making things with my kids instead of just handing them polished software. They get to see the real process. You try something. It fails. You revise. You try again. The point is not frustration for its own sake. The point is learning not to fall apart when the first pass is bad. In fact, it’s even better if you can start to love the “bad”.

Curiosity gets you started. Resilience gets you through the broken middle, or what I called a few weeks ago, The Stall.

Knowing which question to ask.
For a long time, a lot of value came from knowing how to answer hard questions with scarce technical skill.

AI is changing the economics of answering questions.

One of the most honest reactions I have seen to recent progress in AI came from someone who had built part of his identity around being very good at math. The models started solving problems that once felt like the province of a tiny elite. What hit him was not just surprise. It was grief.

If your whole identity rests on being the person who can produce the answer, the ground may start to feel shaky. If it rests on knowing what matters, which questions are worth asking, and where to point your attention, you stand on firmer ground.

That is what I saw on the whiteboard with my daughter.

She was not demonstrating mastery of software engineering as a formal discipline. She was doing something more basic, and maybe more durable. She was noticing what she liked. She was imagining what could be better. She was helping shape the thing before it existed. And asking the questions.

That skill will age well.

Why it matters

I am not an absolutist about any of this. I am not an expert on how the next twenty years unfold. I am still making sense of it myself.

But I am starting to know what I want to reinforce in my kids while the ground is moving.

I want them to build. I want them to ask questions. I want them to speak clearly. I want them to treat people well. I want them to stay curious long enough to notice what others miss, and resilient enough to keep going when the first version does not work.

I also want to give them enough agency to feel the consequences of their decisions while the stakes are still manageable. That is how judgment gets built.

I want them to understand that they are more than any one skill the market happens to prize at a given moment. Titles change. Tools change. Status changes. The deeper work is learning how to adapt without losing yourself.

That is what I am teaching my kids about AI.

Not a theory of the labor market. Not a prediction about which degrees will win. Certainly not fear.

Something more basic.

A responsible relationship with powerful tools. Character. Curiosity. Communication. Resilience. Judgment. The instinct to ask better questions.

Those, to me, are the new street smarts.

Subscribe now

Best of the rest:

💸 Beyond the Sky — Jeffrey Yan’s Hyperliquid story shows how a brutally focused team can turn crypto market structure, community distribution, and product obsession into a startup so efficient it makes traditional notions of scale look outdated. — Colossus

🤖 Sam Altman May Control Our Future—Can He Be Trusted? - This New Yorker profile captures the growing tension at the heart of the AI era: one of the most consequential technologies in the world may be shaped by a leader whose ambition, judgment, and governance still inspire deep unease. - The New Yorker

⚙️ How to get your company AI pilled — Ramp’s internal playbook is a vivid case study in what real AI adoption looks like when leadership sets the expectation, removes friction, and treats company-wide experimentation as a cultural advantage instead of an IT project. — X

🛠️ Eight years of wanting, three months of building with AI - Lalit Maganti offers one of the most credible firsthand accounts yet of what AI coding actually does well, what it quietly breaks, and why the real leverage comes from pairing agentic speed with human taste, architecture, and discipline. - Lalit Maganti

Charts that caught my eye:

→ Why does it matter? Nearly half of all private equity deals now target software and tech services, a share that has doubled over the last 15 years and is at an all-time high. This is also what has people concerned about private credit. If AI compresses the value of legacy SaaS products the way SaaS compressed the value of on-premise software before it, the unwind may not be contained to a single fund or vintage year.

→ Why does it matter? A year ago, the question was whether AI revenue would ever show up. Anthropic answered: $87M to $30B in ARR in roughly 15 months. Salesforce took 22 years to cover the same ground.

→ Why does it matter? IBM tracked every shot at the Masters. The result is mostly blue for bogey. Even at golf’s highest level, bogeys outnumber birdies by a lot. Watching the broadcast, you’d never know.

→ Why does it matter? Two years ago, top models stalled at basic reconnaissance. Mythos Preview now reaches the infrastructure compromise threshold with enough tokens. Capable enough to threaten real weakly-defended systems, and the curve is still climbing.

Tweets that stopped my scroll:

→ Why does it matter? McIlroy’s heart rate hit 150 bpm celebrating, higher than his 135 bpm on the 18th tee. Joy is apparently more taxing than pressure.

→ Why does it matter? A user logged actual behavior data showing Claude’s reasoning depth dropped 67% in late February. Anthropic stayed quiet until the numbers were too documented to dismiss. The working theory among many users: Anthropic is throttling Opus to push traffic toward training its next flagship model.

→ Why does it matter? Anthropic posted benchmark numbers that are turning heads with Mythos. This is the first model trained on Blackwells. Buckle up!

→ Why does it matter? Most big-company ideas die by committee veto. One-yes-wins flips the default: contrarian ideas get a fair shot without needing unanimous buy-in up the chain.

→ Why does it matter? Intercom’s R&D team tripled output in 16 months without tripling headcount. That’s the AI productivity story most engineering leaders are chasing right now.

Worth a watch or listen at 1x:

→ Why does it matter? Jesse Genet is running 11 AI agents in the background of real family life. Not a lab. While homeschooling four young kids. Katherine Boyle and Sarah Wang draw out what that actually looks and feels like: the friction, the workflow decisions, and the mindset required to treat agents like an extension of your parental capacity.

→ Why does it matter? Fair warning: This is a highly technical conversation between Dwarkesh Patel and Jensen Huang. But if you want to nerd out on the economics of AI, this is the one for you. I loved Jensen’s explanation of Nvidia: “We turn electrons into tokens”.

Subscribe now

Quotes & eyewash:

“It is not the strongest of the species that survives, nor the most intelligent, but the one most responsive to change.” – Charles Darwin

→ Why does it matter? One to live by with how much change our world is experiencing right now.

“The best way to cherish life is to remind yourself of life's impermanence. It is to remember that every time you see someone that is one less time you see them. It is to remember that everytime you go somewhere that is one less time you visit. By doing this, you naturally slow down. Almost like a reflex, you start to truly live.” - Andrew Anabi

→ Why does it matter? For those of us that don’t live on the coast and aren’t planning to move there, how many times do you think you’ll see the ocean again? Cherish those visits.

Subscribe now

The mission:

The Wall Street Journal once used “Read Ambitiously” as a slogan, but I took it as a personal challenge. Our mission is to give you a point of view in a noisy, changing world. To unpack big ideas that sharpen your edge and show why they matter. To fit ambition-sized insight into your busy life and channel the zeitgeist into the stories and signals that fuel your next move. Above all, we aim to give you power, the kind that comes from having the words, insight, and legitimacy to lead with confidence. Together, we read to grow, keep learning, and refine our lens to spot the best opportunities. As Jamie Dimon says, “Great leaders are readers.”

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, or legal advice. Readers should do their own research and consult with a qualified professional before making any decisions.

The big idea: Always-On Intelligence

Reading time: 7 minutes

Knowledge work carries a hidden tax.

You read an article on Tuesday. By Thursday, it connects to a board conversation, a product decision, or a pattern you are seeing with customers. You know you saved it. Highlighted it. Maybe even sent it to yourself.

And when you need it, you cannot find it.

Most knowledge workers do not suffer from a lack of information. They suffer from a lack of usable memory. The article sits bookmarked in your browser. The notes live in multiple apps. The meeting transcript hides in a folder you forgot about. You earned the insight. But it is scattered across tabs, tools, and your own imperfect recall.

The Second Brain Became a Second Job

We built a whole category of software to solve this. Note-taking apps. Read-it-later tools. Systems for what people started calling a second brain.

I bought the vision. I also bought the tools. I have tried almost all of them.

For me, most of them turned organizing all of this into a second job.

Every new input became another small task. Save it. Name it. Tag it. Link it. Revisit it. Clean it up before the structure gets messy. The tool that was supposed to reduce cognitive load started competing for it.

That is why so many of these systems do not reach their potential. The idea was right. The upkeep was brutal.

Bush Saw the Problem Early

In 1945, Vannevar Bush published As We May Think and described the Memex: a personal system that could store a lifetime of records and let a user move through them by association instead of strict hierarchy. He understood something that still feels true today: the mind does not think in folders. It moves through trails.

Bush could imagine the library. He could not staff it.

Who updates the links when new information arrives? Who decides what belongs together? Who keeps the system current as the archive grows?

For decades, the answer was simple: you did.

AmbitiousOS Changed My Mind

Earlier this year, with the help of Claude Code, I started building AmbitiousOS on top of the full Reading Ambitiously archive. More than 50 subscribers use it today through voice and text. At first, I thought I was building a smarter interface to a body of work.

I now think we’re building a preview of something bigger.

In version one, AmbitiousOS could search the archive, retrieve passages, and answer questions across the corpus.

Version two changed my mind.

Now Claude maintains a living wiki on top of the source material. The original newsletters remain the source of truth. New inputs update pages on companies, people, and concepts. The system adds links, flags contradictions, and writes the changes back into the knowledge base.

The opportunity is an always-on executive assistant with a working knowledge of your world. One that understands your reading, your meetings, your notes, your documents, your decisions, and the threads that connect them. One that feels like you hired your own librarian to build a private Library of Alexandria around your work.

Right now, I still trigger AmbitiousOS by hand. I drop in an article and tell Claude to ingest it. It reads, files, and updates the system in seconds.

Now remove the trigger.

Imagine waking up and the system has already processed the three articles you saved the night before. It tied one to a company you track, another to a theme you have been writing about, and a third to a decision you made last quarter. It noticed that one of them conflicts with an assumption sitting in your notes, and it flagged the issue before your next meeting. It connected these concepts in a recent email you sent.

You did not ask the question. The system did the maintenance, then proactively alerted you.

The Harness Is Becoming the Product

OpenClaw is the clearest open-source version of that idea. A self-hosted gateway that connects apps and chat surfaces to an always-available assistant, with sessions, memory, tools, multi-agent routing, background work, and scheduled jobs. It feels like you gave AI a real operating environment instead of a single prompt box.

Anthropic is building the closed-source version of the same vision in Claude Cowork. Anthropic describes Cowork as a system for autonomous knowledge work that runs on desktop, works across local files and applications, and returns a finished deliverable. It is built around outcomes, not one-time prompts.

Two versions of the same idea. In my view, this could become the iPhone-caliber product of the AI era.

The model still matters.

But the product value is moving into the harness around it: memory, background maintenance, access to the tools where work lives, and skills that turn context into action.

The Missing Layer in Knowledge Work

Knowledge work still runs on the same loop: gather, make sense, retrieve, decide.

We built good tools for gathering. We built decent tools for retrieval. We never built a system that kept the knowledge layer current on its own.

Now we can.

That matters because the cost of broken memory compounds. You lose time. You miss context. You redo research. You walk into meetings with fragments instead of continuity. Over time, your own accumulated knowledge starts to feel less like an asset and more like a pile.

The products that matter next will keep context warm, maintain your knowledge base in the background, and prepare work before you ask. They will return something closer to a deliverable than a response.

Bush imagined the library in 1945. The missing piece was upkeep.

The web gave us infinite information. Search gave us retrieval. These new harnesses point to the missing layer: persistent intelligence that keeps your accumulated knowledge alive while you work.

I started AmbitiousOS by trying to build a smarter interface to a newsletter archive.

I came away with a stronger conviction about where this market is heading.

You do not need more information.

You need an always on intelligent assistant that remembers what you know and gets to work before you ask. And with AI, it’s now possible.

Subscribe now

Best of the rest:

💸 Clouded Judgment Per Token Pricing – Jamin Ball makes the clearest case yet that in AI, pricing is becoming strategy itself, because the winners will be the companies that turn falling inference costs into value-based token and credit models before everyone else does. – Clouded Judgement

⏳ Some Things Just Take Time – Armin Ronacher makes the case that in an AI age obsessed with speed, the real moats are still trust, craftsmanship, and the long, unglamorous work of sticking with something long enough for it to matter. – Armin Ronacher

🤖 OpenAI CEO and CFO Diverge on IPO Timing – The real story is not just IPO timing, but whether OpenAI can outrun the financial strain of a $600 billion compute buildout before public markets start demanding discipline. – The Information

🐢 Thoughts on slowing the f*ck down – A sharp corrective to AI coding hype, this argues that speed without judgment produces brittle software, runaway complexity, and teams that stop understanding the systems they ship. – Mario Zechner

🚀 SpaceX Is Said to Target More Than $2 Trillion Valuation in IPO – This matters because a $2 trillion target would make SpaceX’s debut the biggest IPO ever, and force public markets to underwrite one of the most ambitious growth stories in modern corporate history. – Bloomberg

Charts that caught my eye:

→ Why does it matter? Goldman Sachs found that industries where AI augments workers are adding jobs, while those where AI substitutes for workers are shedding jobs. The unemployment rate for high-substitution occupations is already up more than 1.4 percentage points from its 2022 baseline. The workers feeling the most pressure appear to be entry-level ones, suggesting the rungs at the bottom of the career ladder may be getting harder to find.

→ Why does it matter? The architecture behind one of the most popular apps in live sports: The Masters!

→ Why does it matter? Hat tip to Tey Bannerman for surfacing this one. Microsoft has branded somewhere around 78 separate products “Copilot,” and the chart makes the sprawl hard to ignore: Copilots inside Copilots, Copilots for other Copilots, and a dedicated hardware key on your keyboard to summon them. Will the real copilot please stand up?!

→ Why does it matter? This is the leaked OpenAI cap table (I have no idea if it is fake news). If it's real, it tells a few stories at once. Microsoft’s $13 billion bet is now worth 17.6x, making Satya Nadella’s 2019 call one of the greatest capital allocations of the modern era. The nonprofit foundation is sitting on $220 billion in gains from a $0 cost basis, which is the most unusual wealth creation in corporate history. NVIDIA, despite being the backbone of everything OpenAI builds, is currently underwater on their position. And then there is Ashton Kutcher, who turned $30 million into $1.3 billion in under three years, a 43x return that most professional investors will never see in a lifetime. If this company IPOs at $1 trillion or more, every number on this table increases by 30%, and the nonprofit’s gains become a governance story unlike anything we have seen before.

Tweets that stopped my scroll:

→ Why does it matter? A frontier-class multimodal model, built from the same research that powers Google Gemini 3, is now running entirely on a phone at 40 tokens per second. This could jeopardize cloud-based AI inference. On-device models at this speed and capability level suggest that privacy, latency, and cost constraints that once required a server farm could someday be available locally on the device in your pocket.

→ Why does it matter? The CFO of the world's most valuable AI company is privately raising concerns about spending and IPO timing, and she no longer reports directly to the CEO. Worth watching this relationship closely.

→ Why does it matter? Tony Fadell raises a question worth sitting with: when product management and product marketing operate as separate functions, does the story stay close enough to the product? There is no single right answer to how you organize these roles. But his point is worth understanding. The person who defines what gets built and the person who decides how it gets explained need to be in constant contact, ideally the same person. Greg Joswiak’s superpower at Apple wasn’t marketing genius in isolation. It was empathy deep enough to become the customer, which kept the message and the product aligned.

→ Why does it matter? A year ago, the big question was whether enterprise AI revenue would materialize. Well, here it is. Anthropic added $11B in annualized revenue in a single month. When your million-dollar customer count doubles in under two months, that’s worth paying attention to!

Worth a watch or listen at 1x:

→ Why does it matter? Brad Lightcap is OpenAI’s COO. He covers how OpenAI thinks about the transition from tools to agents, what enterprise adoption actually looks like at scale, and where the real bottlenecks are. Worth a listen!

Subscribe now

Quotes & eyewash:

“If you don’t build your dream, someone else will hire you to help them build theirs.” ‌- Dhirubhai Ambani, Founder, Reliance Industries

→ Why does it matter? Are you climbing the right hill?

Subscribe now

The mission:

The Wall Street Journal once used “Read Ambitiously” as a slogan, but I took it as a personal challenge. Our mission is to give you a point of view in a noisy, changing world. To unpack big ideas that sharpen your edge and show why they matter. To fit ambition-sized insight into your busy life and channel the zeitgeist into the stories and signals that fuel your next move. Above all, we aim to give you power, the kind that comes from having the words, insight, and legitimacy to lead with confidence. Together, we read to grow, keep learning, and refine our lens to spot the best opportunities. As Jamie Dimon says, “Great leaders are readers.”

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, or legal advice. Readers should do their own research and consult with a qualified professional before making any decisions.

Programming note: No issue next Friday. Reading Ambitiously returns April 10.

The big idea: The Stall

Reading time: 6 minutes

The first time I smoked a pork shoulder, the first few hours made me feel smarter than I was.

I bought the meat, got the cooker steady at 225, inserted the fancy digital probe, and watched the internal temperature rise on schedule. For a while, it all felt wonderfully mechanical. Heat goes in, number goes up, dinner gets closer. I cut the lawn. I started imagining the finish before I had earned the right to.

Then the temperature stopped climbing.

Not slowed. Stopped.

It sat there for what felt like an absurd amount of time. Then it barely moved. At one point, I was convinced it had gone backward. I checked the batteries in the thermometer. I tapped the smoker gauge like that might somehow help. Meanwhile, guests were arriving, and the meat was still nowhere near where it needed to be. That is when panic sets in.

Pitmasters have a name for this moment: the stall. It unnerves backyard cooks because the line had been moving so obediently, until suddenly it wasn’t.

That is what makes the stall so unsettling. You did the setup right. You followed the process. The early feedback looked good. Then the signal vanishes, and your confidence goes with it.

What the Stall Really Is

The explanation, at least in barbecue, is simple. As moisture evaporates, it can cool the meat almost as quickly as the smoker heats it. From the outside, it looks like nothing is happening. Underneath, a great deal still is.

The stall is not just a barbecue phenomenon. It is a recurring feature of life and progress. It shows up almost everywhere that the work is real.

Anyone who has lifted weights for more than a few months knows the feeling. Early on, progress is generous. The bar moves. The numbers go up. Effort feels neatly rewarded. Then the curve flattens. The same lift that used to move cleanly starts to feel heavy and stubborn. If you are impatient, you do what the nervous pitmaster does. You force it. You add weight too quickly, let your form erode, and turn a plateau into an injury.

The same pattern shows up in sports, learning, writing, and work. At the beginning of any worthwhile pursuit, improvement is visible enough to keep your confidence high. Then the gains become harder to see. The work continues, but the proof arrives less often and less neatly.

When Progress Goes Quiet

Most plateaus are not dangerous in themselves. What is dangerous is how we interpret them. We mistake a pause in feedback for a failure in process. We assume something is broken because something is no longer obvious, and because we assume something is broken, we start making emotional decisions. In barbecue, that means cranking the heat.

In lifting, it means chasing load rather than technique. In the workplace, it means abandoning a sound process because results are arriving more slowly than your patience would prefer.

The stall is rarely just about progress. It is about psychology. We depend on visible signs that our effort is paying off. When those signs disappear, even temporarily, our judgment gets noisy. The stall asks for continued effort without immediate reassurance.

Systems Over Goals

That is one reason I have become increasingly drawn to the idea of systems over goals.

Goals matter. They give shape to ambition. They tell you what you are aiming at. But goals are often least useful in the middle, and the middle is where most of life is lived. The middle is where the scale stalls. The draft gets worse before it gets better. The business has a quarter that feels flatter than the last three. The visible progress that made the process feel rewarding goes quiet.

A goal is an outcome. A system is a repeatable way of operating. It is the set of behaviors, constraints, and rhythms that still make sense on the days when the scoreboard is silent. Goals tell you where you want to go. Systems tell you what to do this week, especially when the mood shifts against you.

Goals can make you stare at the gap. Systems return your attention to the process. Goals ask, am I there yet. Systems ask, did I run the right play today.

That is why systems matter most in the stall.

When progress is obvious, almost any plan feels motivating. When progress becomes non-obvious, psychology takes over. People improvise. They abandon routines that were working. They reach for intensity when what they needed was consistency. They confuse urgency with wisdom.

What a Good System Does

A good system protects against that. It gives you something sturdier than mood.

It shifts your attention from lagging outcomes to leading behaviors. It lowers the odds that one bad day turns into a broken week. Most importantly, it tells you what to do when the signal disappears. A system does not require you to be heroic every day. It gives you a repeatable way to keep running the play while the deeper work catches up.

That will vary by person and by pursuit. The details are personal. The principle is not. A good system reduces the number of decisions you have to make when you are discouraged.

The better question is not just what outcome do I want. It is what process can I trust when the outcome is not yet visible. Put differently, what kind of operating system will keep me steady when my instincts start telling me to overcorrect.

Your Operating System

This is also where AI starts to feel genuinely useful to me. Most people still approach it as a search tool, a writing tool, or a source of quick answers. It can do all of that. But one of its more practical uses may be helping people build a better system for themselves.

Used well, AI can help you define the inputs that matter, track them, notice patterns across your weeks, and review what is working without having to reconstruct the story from memory after a frustrating day. It can serve as an accountability partner, a thought partner, and a low-friction layer for personal change.

That does not mean outsourcing judgment or borrowing someone else’s routine off the internet. It means more people now have access to an always-on partner that can help them build something highly personal, not a generic productivity system, but their system, shaped around their rhythms, blind spots, and the specific ways they tend to lose the plot when they hit a stall.

One of the best ways to learn how to use AI may be to point it first at the oldest management problem you have: yourself.

Proof Later

With grilling season upon us, I’m thinking about that first cook.

At the time, the stall felt like evidence that I had lost control of the cook. In reality, it was evidence that I did not yet understand the process. The meat was not broken. The thermometer was not lying. The work had simply moved somewhere I could not easily see.

A lot of life is like that.

The hardest moments are often not the beginning, when uncertainty is expected, or the end, when the result is visible. They are the long middles, when effort is still required, but reassurance has gone missing. That is when people start making emotional decisions. That is when they crank the heat.

The stall has made me steadier in those moments, less eager to mistake silence for failure. It has reminded me that some of the most important forms of progress do not announce themselves in real time. Less eager to crank the heat. Consistency first. Proof later.

Subscribe now

Best of the rest:

🤖 More! More! More! Tech Workers Max Out Their A.I. Use. - “Tokenmaxxing” shows how fast AI has shifted from productivity tool to workplace status marker, and why spiraling usage costs may be the next hidden tax of the enterprise AI boom. - The New York Times

💼 OpenAI sweetens private equity pitch amid enterprise turf war with Anthropic - This is the clearest sign yet that the AI leaders are not just selling models, they are building distribution machines to lock up the enterprise before the market settles. - Reuters

🌎 Larry Fink’s 2026 Annual Chairman’s Letter to Investors - Fink’s core argument is bigger than BlackRock, namely that if capital markets are driving more growth and AI is set to concentrate even more wealth, broader ownership, private market access, and tokenized investing start to look less like product innovation and more like economic necessity. - BlackRock

🏢 Productive Individuals Don’t Make Productive Firms - George Sivulka makes the right point at the right moment, namely that AI gains do not compound at the company level until workflows, incentives, and systems are redesigned around the technology instead of simply layered on top of old ways of working. - X

🤖 TurboQuant: Redefining AI efficiency with extreme compression — Google’s latest compression work matters because it points to a future where long-context AI and vector search get dramatically cheaper and faster, without the usual accuracy tradeoffs. — Google Research

Charts that caught my eye:

→ Why does it matter? Thoma Bravo is one of the most sophisticated software investors on the planet, and their thesis here is precise: public markets are pricing software companies on sentiment, not fundamentals. The chart shows the Rule of 40 profile (a measure of real business quality combining growth and profitability) climbing steadily toward 35% even as EV/FCF multiples compress sharply, a divergence that only happens when fear is doing the pricing, not math. The crowd sells the category, the disciplined buyer buys the company. Not all software businesses will recover equally, but the ones with durable unit economics and sticky revenue are trading at a discount that history suggests will not last.

→ Why does it matter? The revenue gap between OpenAI and Anthropic appears to be $6B. It may actually be much larger. Anthropic counts gross hyperscaler revenue before paying out its AWS revenue share, while OpenAI reports net revenue after paying Microsoft Azure. Strip out the accounting difference, and Anthropic's $19B annualized figure shrinks materially. If both companies pursue an IPO this year, the SEC is unlikely to let that comparison stand, and the chart you think you are reading today will look very different once the numbers are restated on a common basis.

→ Why does it matter? Anthropic shipped over 100 product releases in 52 days, averaging roughly two per day, and most users caught only five of them. The pace of AI product development has moved so far ahead of human attention that even power users are operating with an incomplete picture of what the tools can already do.

Tweets that stopped my scroll:

→ Why does it matter? Sierra just shipped Ghostwriter. An agent that builds agents, removing the bottleneck of needing specialized AI talent to deploy customer-facing intelligence. If a platform can now handle voice, chat, multilingual support, and system integrations in one shot, the barrier between "we're exploring AI" and "we have AI in production" collapses.

→ Why does it matter? Guillermo Rauch is the Founder & CEO of Vercel. And his company has quickly become a favorite of AI builders. SORs are highly defensible. UIs not so much. Agents with CLIs and skills. API access. Feels like the future.

→ Why does it matter? Researchers just demonstrated that anonymity online is functionally dead. LLMs can now cross-reference pseudonymous posts against real identity profiles at scale, moving from less than 0.1% to 54% match rates on HackerNews to LinkedIn. Yikes.

Worth a watch or listen at 1x:

→ Why does it matter? Jensen Huang is the most important CEO in technology right now, and this three-hour conversation with Lex Fridman is the closest thing to a masterclass on what it takes to build the infrastructure for a civilizational shift. He covers AGI, scaling laws, the future of coding, data centers in space, and China with the kind of clarity and conviction.

→ Why does it matter? Andrej Karpathy is one of the clearest thinkers in AI, and this conversation with Sarah Guo covers the territory most people are still catching up to: agents that loop, self-improve, and run experiments without a human in the seat. He introduces the idea of the "loopy era," where AI systems don't just execute tasks but iterate on their own outputs, which changes everything about how you build, supervise, and trust them. If you want the mental model for where software development goes next, Andrej has it for you.

→ Why does it matter? Jensen Huang sits down with the All-In crew and goes deep on the inference explosion, physical AI, and what it actually means to run the world's most valuable company during the biggest platform shift in a generation. Jensen thinks deeply about how to put words to some of these big ideas shaping the AI era, and I find those framings so helpful.

Subscribe now

Quotes & eyewash:

→ Why does it matter? Shoutout @rtwlz! The ultimate way to get pumped up before a meeting.

→ Why does it matter? I’m ready for it!

Subscribe now

The mission:

The Wall Street Journal once used “Read Ambitiously” as a slogan, but I took it as a personal challenge. Our mission is to give you a point of view in a noisy, changing world. To unpack big ideas that sharpen your edge and show why they matter. To fit ambition-sized insight into your busy life and channel the zeitgeist into the stories and signals that fuel your next move. Above all, we aim to give you power, the kind that comes from having the words, insight, and legitimacy to lead with confidence. Together, we read to grow, keep learning, and refine our lens to spot the best opportunities. As Jamie Dimon says, “Great leaders are readers.”

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, or legal advice. Readers should do their own research and consult with a qualified professional before making any decisions.

Enjoy this week’s Big Idea read by me:

The big idea: Prompt to Harness: The Future of Software Development

Reading time: 4 minutes

Over time, engineering teams automate more of the mechanics of job and reallocate their effort to where human-only skills matter most.

There was the era of the mainframe, when the work sat closer to the machine. Then came waterfall, which formalized sequencing and handoffs. Then agile, which tightened shortened release cycles. Today, many teams operate in a world of continuous delivery and, increasingly, continuous deployment, where software ships constantly and iteration happens in near real time.

Each shift changed software engineering. More importantly, each shift changed the craft and where to focus.

What teams did not do was stop engineering. They stopped spending time on the parts that had been abstracted away. The locus of value kept moving.

That is what makes the rise of agentic coding feel less like a break from history and more like the next turn in a familiar story.

From Prompt to Harness

The first phase of this AI cycle was prompt engineering.

If we ask the model better questions, we tend to get better answers. Prompting quickly became an important skill for getting useful results from AI. Guides emerged. Websites helped people write better prompts. You can even ask an LLM to help you write one.

But as our ambitions expanded, we moved from one-off outputs to real systems. It became clear that a well-written prompt could still yield weak results if the model lacked the right information at the right time.

Enter context engineering.

In addition to clever instructions, we learned that you also need to protect the context window. Think of context as the model’s memory at runtime. What should the model see? When should it see it? And how do we avoid wasting this precious resource?

A great deal of innovation has gone into addressing context rot, something we wrote about back in January. Tools, MCP, skill libraries, orchestration, and RAG are all, in one way or another, attempts to improve the quality of context.

This was an important deepening of the craft.

Prompting did not go away. It got wrapped inside a growing set of skills and disciplines. We learned that asking well still mattered, but so did supplying the right context, in the right sequence, inside a finite window.

Now the discipline is expanding again.

As the unit of work shifts from answering a question to completing a task over time, the burden moves from the prompt itself to the system around it. Once agents begin taking on longer-running work across multiple steps, tools, and context windows, we also need a harness.

Enter harness engineering: the tooling, practices, tests, guardrails, and feedback loops that keep the system reliable over time.

Once agents are doing meaningful work across files, tools, and dependencies, the quality of the outcome depends less on any single prompt and more on the system wrapped around the model. The harness keeps it on track while the work unfolds.

Prompt, context, and harness engineering are best understood as cumulative ideas. They build on one another. Together, they are beginning to define how software gets built going forward.

That matters because some engineering teams may now be able to increase velocity dramatically, but only if they are willing to redefine the job around these new disciplines.

Over a five-month period, a team at OpenAI built and shipped a product with zero manually written code. The repository reportedly reached roughly one million lines. The application logic, tests, configuration, documentation, and tooling were generated by AI. OpenAI has suggested that it was built in roughly one-tenth the time it would have taken to hand-code.

The engineer’s role in that project appears to have been different.

When something failed, the answer was often not to work harder inside the codebase itself. It was to ask what capability was missing, what artifact was unclear, what test was absent, what feedback loop was too weak, and what part of the environment needed to become more legible to the agent.

This begins to look like the work of a harness engineer.

The deeper shift is that engineering judgment moves from isolated acts of code-writing into the design of systems that can generate, test, and improve code over time.

The Future of Software Development

We may not have fully named this phenomenon yet.

Harness engineering is one candidate. Regenerative software is another. The phrase AI software factories gets at part of the same idea, too. But the label matters less than the shift underneath it.

Prompt engineering, context engineering, and harness engineering increasingly look less like isolated disciplines and more like connected skills of the craft.

Intent goes in. Code comes out. Tests push back. Evaluations refine the path. The harness gets stronger. The next cycle improves.

That is why this looks less like an adjunct to software engineering and more like one path along its next evolution. The craft is not being automated away. It is being relocated toward the design and supervision of the systems that produce software.

There is a historical rhyme here.

Earlier eras of industrialization did not simply replace labor with machinery. They reorganized work around process, measurement, and reliability. Something similar appears to be happening in software, and increasingly across knowledge work more broadly.

Each new layer of abstraction has tended to push more human effort toward the place where judgment mattered most.

Prompt engineering taught us to ask better questions. Context engineering taught us to give the model the right information at the right time. Harness engineering is teaching us that reliability and quality come from the system that wraps around the model, as well as from feedback loops that improve over time.

If that framing holds, the leading software companies are unlikely to treat coding agents merely as point tools. They will begin to reinvent their software development lifecycle around these ideas, because this feels less like a break from history and more like the next turn in a familiar story.

Subscribe now

Best of the rest:

🧠 Tim Ferriss Says Self-Help Might Be the Problem – After 20 years writing and consuming self-improvement content, Ferriss argues the obsession with fixing yourself creates the very unhappiness it promises to cure. – The Blog of Author Tim Ferriss

🤖 Finance Bros to Tech Bros: Don’t Mess With My Bloomberg Terminal – “Keep your stupid AI off my Bloomberg terminal” is the perfect distillation of this whole fight, because it shows how quickly AI hype runs into the hard reality of entrenched workflows, proprietary data, and fanatically loyal power users. – The Wall Street Journal

⌚ How the Swiss Watch Industry Survived by Selling Identity, Not Accuracy – When quartz made precision a commodity, surviving watchmakers pivoted to luxury branding. The same shift is coming for knowledge work. – Paul Graham

🏦 How a $20B hedge fund turned GPT-5 into a senior analyst – Balyasny built a purpose-built AI research engine, evaluated across 12+ dimensions, that reasons through thousands of documents at institutional scale. – OpenAI

🏗️ AI Is Coming for SAP Before It Replaces It – Legacy systems hold decades of critical business data. AI finally makes them programmable, approachable, and worth the upgrade cost. – a16z

Charts that caught my eye:

→ Why does it matter? Last year we were wondering whether the revenue would show up. Well, it did. Anthropic added about $6 billion in annualized revenue in February alone, which is larger than Snowflake’s full FY2025 revenue of $3.6 billion and exceeds Databricks’ last reported $4 billion revenue run rate, suggesting this is not a normal software ramp. It is a sign that when AI starts doing real work, not just helping people use tools, revenue can compound at an incredible pace.

→ Why does it matter? LLMs are starting to tell users to break up, and that should not surprise anyone who understands the data. Train a model on Reddit relationship advice, and you train it on a corpus where a huge share of people default to “leave.” Feed it fifteen years of posts shaped by less patience, less communication, and less willingness to compromise, and the model will learn that exit is the safest, most common answer. The model is doing exactly what the dataset taught it to do.

→ Why does it matter? Software has historically given customers the tools to do the job, which kept the market tied to seats, modules, and trained users. Agents change that equation because the software starts doing the job itself. As these systems develop the skills of their human labor equivalents, the addressable market expands from software budgets to labor budgets, and that is why the TAM gets much bigger.

Tweets that stopped my scroll:

→ Why does it matter? Gokul is describing something most companies still treat as a future concept. A customer problem surfaces in Slack, a coding agent cuts a PR, and 15 minutes later, the fix is live. We did it with AmbitiousOS at 35k feet. That is a new operating model. The companies that build and expand their own software factories over the next 24 months will have a structural velocity advantage.

→ Why does it matter? Ramp just handed AI agents a corporate card with guardrails. Until now, a key missing primitive in agentic systems was not intelligence or tools, it was money. Ramp Agent Cards give AI Agents: spend limits, merchant controls, and full auditability mean agents can now operate in the real economy without a human co-signing every purchase.

→ Why does it matter? Anthropic just shipped the ability to message Claude from your phone and have the finished work return to your desktop. This is the early architecture of an always-on AI collaborator that works while you sleep, commute, or sit in back-to-back meetings. The race to own the persistent AI relationship, one that knows your context, your files, and your intent across devices, is a real platform war of this era.

→ Why does it matter? The barriers between “I have an idea” and “here is a thing you can click” are collapsing fast. Google Stitch is a new product from Google that helps you get to prototypes quickly.

Worth a watch or listen at 1x:

→ Why does it matter? Dylan Patel breaks down the three physical constraints that will determine who wins the AI era: logic, memory, and power. If you want to understand why AI scaling is not simply a money problem, Dylan Patel and Dwarkesh Patel are two of the brightest minds on the subject, oh, and they are cousins.

→ Why does it matter? Jensen Huang does not do keynotes, he does roadmaps, and this one covers the full arc from Blackwell to Rubin to Feynman while making the case that AI factories are the defining infrastructure build of our era.

Subscribe now

Quotes & eyewash:

→ Why does it matter? 10+ years in the making.

Subscribe now

The mission:

The Wall Street Journal once used “Read Ambitiously” as a slogan, but I took it as a personal challenge. Our mission is to give you a point of view in a noisy, changing world. To unpack big ideas that sharpen your edge and show why they matter. To fit ambition-sized insight into your busy life and channel the zeitgeist into the stories and signals that fuel your next move. Above all, we aim to give you power, the kind that comes from having the words, insight, and legitimacy to lead with confidence. Together, we read to grow, keep learning, and refine our lens to spot the best opportunities. As Jamie Dimon says, “Great leaders are readers.”

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, or legal advice. Readers should do their own research and consult with a qualified professional before making any decisions.

Enjoy this week’s Big Idea read by me:

The big idea: Trust is the rate limiter on risk transfer

Reading time: 3 minutes

By the simplest definition, business is the exchange of risk. That’s it.

For that exchange to take place, the most essential ingredient is trust.

The world is adopting AI at a speed we rarely see in technology. And I am optimistic. This is a new operating system.

At the same time, human nature has its own operating system, older than markets and older than modern institutions. One of its core jobs is deciding when reliance is safe.

With AI, we can rapidly go from idea to artifact.

Still, business isn’t the exchange of artifacts. It’s the transfer of risk. When a buyer chooses you, they’re not only buying your product, they’re choosing who they can rely on when the stakes are real.

Maybe that sounds obvious or trite. And you’re saying “duh, Jack”. I think it’s easier said than done.

Evidence

There is an explosion of new companies, and news every day about how fast they’re growing. Speed is impressive, and velocity is important. But can trust be established at the same pace, especially when the risk being handed over is meaningful and priced accordingly?

I’ve worked for two companies in my career: IBM and Ridgeline. Both were shaped by founders who understood this deeply, Thomas Watson and Dave Duffield. IBM is one of only a handful of companies in the “century” club. Dave is one of a small number of founders to have built multiple companies that went public and entered the S&P 500.

Amongst many things, I think they both understood that technology is important, but it’s rarely the deciding factor on its own.

Imagine for a second if you gave peak-era IBM sales teams today’s AI copilots, agents, and automation tools. They’d sign customers up. They’d probably crush sales targets. But the deeper advantage wouldn’t be the tools. It would be the culture and the way customers learned to rely on IBM over time.

Lately, I keep thinking about that point as the x-factor in company building.

Trust Engine

My first year on the job at IBM was spent in training programs designed to teach employees how to show up as an IBMer. At the first official meeting I ever attended with Dave Duffield about what would become Ridgeline, core values were the first order of business.

Buffett and Munger seemed to understand this too. If you read their letters carefully, the “people” part is foundational to their investment thesis.

So what do we do about it?

The takeaway for me is not “slow down.” It’s “build the trust engine as deliberately as you build the product.” Put the same care into culture and people as you put into the shiny objects, especially right now when so many of us are trying to figure out AI.

When I say “trust engine,” I mean a set of core values, practices, and culture that make you consistently reliable.

Because the decision to hand over risk won’t be made on demos and feature comparisons alone. It’s made on pattern recognition. Who has shown up consistently? Who has history? Who understands the domain beyond the slide?

That’s what I mean by trust turning into something personal, almost emotional, in business relationships.

Rate Limiter

AI makes it a better time than ever to launch a new product or start a company. You can build faster, iterate faster, and learn faster.

But the thing you’re really asking a customer for is still the same. You’re asking them to trust you enough to underwrite risk with you.

That’s why trust is the rate limiter on risk transfer. You can accelerate the build cycle and generate evidence sooner. Reliance still takes time, and it still gets earned the old-fashioned way: by doing what you say you’re going to do.

And that’s my personal favorite definition of the word trust.

Subscribe now

Best of the rest:

💼 OpenAI Selects Law Firms Cooley, Wachtell for IPO Prep – The company has taken its first concrete step toward a potential public listing that could rank among the largest in history, signaling that the AI leader may soon test public markets at a pivotal moment for the industry. – The Information

🤖 Meta acquired Moltbook, the AI agent social network that went viral because of fake posts — Meta just bought one of the internet’s strangest experiments in AI-generated social activity, signaling that even chaotic, synthetic engagement can be strategic when you are racing to define the future of online interaction. – TechCrunch

🗣️ The stranger secret: how to talk to anyone – and why you should — In an age of headphones, phantom phone use, and social caution, this piece argues that small, low-stakes conversations with strangers are not awkward relics but essential practice for rebuilding our social muscles and civic trust. – The Guardian

Charts that caught my eye:

→ Why does it matter? Curious what websites ChatGPT is visiting to help source its response?

Tweets that stopped my scroll:

→ Why does it matter? This feels like the future user experience of how we’ll use computers. Bravo, Perplexity!

→ Why does it matter? This is a crazy story, and it’s the second time I’ve heard about an incredible entrepreneur asking someone they wanted to hire, “Where are you?” and then immediately going to see them. There’s a great story in Elon’s biography about him doing the exact same thing. It makes you think. Is there anyone in your life you need to go see right now?

→ Why does it matter? An example of a skill called cost-estimate that I really like. An easy one for anyone to replicate. And a great way to convey the ROI on investing in these tools.

→ Why does it matter? This was one of the most discussed tweets on the internet this week, and it captures an important idea: recursive learning. A human writes the strategy document. An agent runs experiments autonomously. A clear metric determines what stays and what goes. Then the loop runs again and again, compounding while you sleep.

→ Why does it matter? I cannot wait for Claude to nail generating graphics and animations. It’s a tough use case for AI right now. Everybody is trying to get AI to generate slides. There isn’t really a clear leader. This is progress.

Worth a watch or listen at 1x:

→ Why does it matter? To me, Coatue is saying this is a regime change, and they are investing as if it were. They do not assume the Magnificent 7 will remain the Magnificent 7; they keep their own roster, the “Fantastic 40,” a curated list of companies they believe can lead in an AI-driven world. That mindset, in their opinion, makes the SaaS repricing feel inevitable.

→ Why does it matter? What I take from this conversation is a very practical theory of where agents actually work. Code repos are unusually agent-friendly, they are text-heavy, versioned, and full of built-in feedback like tests, which is why coding agents have sprinted ahead while most “knowledge work” still feels squishier. Taylor’s “harness” idea is the missing layer, turn a workflow into something repo-like with documentation, rules, tools, and even a blunt memory system, sometimes as simple as a directory of markdown files.

Subscribe now

Quotes & eyewash:

→ Why does it matter? Connect Claude Code to your Apple Vision Pro. Now we’re talking.

Subscribe now

The mission:

The Wall Street Journal once used “Read Ambitiously” as a slogan, but I took it as a personal challenge. Our mission is to give you a point of view in a noisy, changing world. To unpack big ideas that sharpen your edge and show why they matter. To fit ambition-sized insight into your busy life and channel the zeitgeist into the stories and signals that fuel your next move. Above all, we aim to give you power, the kind that comes from having the words, insight, and legitimacy to lead with confidence. Together, we read to grow, keep learning, and refine our lens to spot the best opportunities. As Jamie Dimon says, “Great leaders are readers.”

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, or legal advice. Readers should do their own research and consult with a qualified professional before making any decisions.

Enjoy this week’s Big Idea read by me:

The big idea: AI Product Strategy for Model Consumers

Reading time: 9 minutes

Most of our readers are not training AI models. We’re building software that consumes them.

That sounds obvious, but it changes where you place your bets.

If you are not building the brain, your job is to build a system around the brain. The part customers pay for. The part that compounds into durable value. The part that stays defensible when the underlying LLM makes meaningful improvements.

One way to make sense of this world is to split it into two camps: Model Producers and Model Consumers.

Model Producers train frontier models, and increasingly, they are shipping products on top of them. In the American closed-source world, that center of gravity sits with OpenAI, Anthropic, and Google. You can make a case that xAI belongs in the conversation, too.

Model Consumers are the rest of us. Mostly software and services companies that rent intelligence and build products around it.

If the brain is rented, what is the most defensible way to build the product around it?

Let’s talk about the brain first.

The Big 3

You know their apps: ChatGPT, Claude, Gemini.

These labs are no longer just building large language models. They are building across three layers: models, apps, and harnesses.

Models are the brains. They determine reasoning, writing, coding, and multimodal perception. They are what benchmarks measure and where labs are in a close race.

Apps are the front door. Apps are the products you use to talk to a model, and how models increasingly complete work for you. More and more, they sit inside productivity suites, developer tooling, and operating systems.

Harnesses are what turn intelligence into work. Apps come with a harness. They give models tools, permissions, file access, logging, memory, guardrails, and the ability to execute multi-step tasks. A harness is the difference between answering a question and completing a job.

If “harness” feels old-fashioned, think of an exoskeleton. You are not adding intelligence. You are channeling the raw cognitive power of an LLM. The harness, or exoskeleton, directs that power toward output you can rely on.

Claude in a chat window is one experience. Claude inside a coding harness is another, because that environment is designed to read a repo, edit files, and run commands. Claude, as a desktop co-worker, is another, because it has local file access and the autonomy to complete tasks.

This matters because Model Producers are not stopping at models. They cannot, because models are quickly becoming commoditized.

They are building apps and harnesses, too, because that is where distribution, retention, and margin live. To quote Jeff Bezos, “Your margin is my opportunity.”

That is why the AI labs are moving into territory that used to be exclusively the domain of software companies.

And if producers are climbing into the app and harness layers, model consumers need to know what they can still own.

Five places defensibility lives

If you are a software company, you are a Model Consumer. You rent the brain. You build around it. This is typically referred to as “wrapping” an LLM.

Here is the simplest stress test.

If your model API disappears, what is your IP?

Here is what I’m seeing.

1. Integrations: The unglamorous work. Connectors, schema mapping, permissions, and data normalization. In the enterprise, connectors are a product. If your AI cannot reliably access the right data with the right permissions, you have a demo, not a deployable system.

2. Proprietary data and modeling: LLMs train on public data. What dataset do you own that is not public, and how do you make it legible to the model? This is ontology in practice. A representation of how the business works that the model can reason over safely.

3. Tools and plugins: These are the proprietary actions you expose to the model, APIs, workflows, and system updates inside your product. Instead of just generating language, the LLM can execute real tasks within your software.

4. Skills: Reusable workflows, instructions, playbooks. Enterprises do not have one workflow. They have hundreds. A skill library that starts small compounds institutional knowledge over time.

5. Memory: If the system learns how your team works, replacement gets expensive. Memory turns a generic assistant into something closer to an institutional colleague.

If you want one organizing principle, it is this: everything a Model Consumer builds should ladder to one or more of these five.

Once you’ve made that call, the product strategy gets clearer.

Four product strategy plays for model consumers

You do not need to be a Model Producer to win. You need a playbook that maps to what you can own. Here are four repeatable, product-level plays. Each is a different way to own an app, a harness, or both, while renting the model.

1. Universal AI Assistants: The front-door play, owning the app layer. The product wants to sit where users already spend time. Microsoft Copilot is the canonical example because it lives inside Microsoft 365 and draws on organizational context. Glean is another, built around permission-aware connectors and enterprise search. Their defensibility is distribution, integration, and grounding.

2. AI Agents: The outcome play, owning app plus harness. Agents take responsibility for slices of work. They have tools, entitlements, audit trails, and multi-step execution. Convert services into software. The closer you get to outcomes, the more pricing shifts from seats to results. Maybe #6 IP opportunity here, training your own small language model.

3. Enterprise-Grade Controls: Permissions, governance, compliance, observability, auditability. Enterprises will not deploy AI at scale without trust and transparency they can measure and defend. Build it for them.

4. Embedded AI Features: The compounding play, owning workflow depth inside your existing product. AI becomes a feature set inside existing workflows. Summarization, triage, autofill, exception handling, and natural language query. Many categories will be won by the product that becomes meaningfully easier to use thanks to AI.

Why this matters

Frontier models will keep getting better. At the same time, they will converge and become harder to tell apart.

The labs know this. That is why they are investing so heavily in apps and harnesses. They are moving up the stack into territory that used to be exclusively the domain of software companies.

If you are not building one of these brains, but building around one, you must be explicit about where your differentiation lives.

Decide where your intellectual property sits. Integrations. Proprietary data and ontology. Tools. Skills. Memory. Pick a primary bet(s) and build around it.

Because here is the uncomfortable truth.

As base models improve, some products will wake up and discover that what they thought was a moat was simply proximity to the frontier model. When the underlying intelligence gets meaningfully better, parts of the value proposition compress overnight.

I’m told Larry Ellison once said something to the effect of, “Your company is a feature on my platform.”

In an AI world, that is the risk.

If what you built can be absorbed into someone else’s app or harness the moment the brain improves, you were never building a platform. You were building a feature. And borrowing the intelligence that powered it.

Make sure you own something that survives when the brain gets smarter next week.

Subscribe now

Best of the rest:

🏦 Howard Marks No Longer Sitting on the Fence – The Oaktree founder went looking for an AI bubble, used Claude to build his own tutorial, and came back convinced this is real. – Robert Dee ACSI

🤖 There is exactly one way that SaaS can be saved — Eoghan McCabe argues that survival in the age of agents requires radical creative destruction, detailing how Intercom cannibalized its own legacy business, bet the company on AI, and reignited growth by rebuilding around agents from the ground up. – Eoghan McCabe

⚖️ Anthropic Picks a Fight It Cannot Win – Ben Thompson argues Anthropic’s standoff with the Department of War is a principled but ultimately untenable position, disconnected from how power actually works. – Stratechery by Ben Thompson

👨‍💻 We stress-tested GPT 5.4 on the hardest UI on the internet – After months inside legacy insurance portals, Pace found GPT-5.4 can finally click accurately, reason across hundred-step workflows, move fast enough for production, and retain UI memory, signaling that computer-use agents are crossing from demo to deployment. – Jamie Cuffe

Charts that caught my eye:

→ Why does it matter? This is a wild chart if true. Anthropic is becoming the dominant player in business AI spend in just 18 months, with Claude Team alone now accounting for nearly half of all enterprise AI chat dollars.

→ Why does it matter? The Financial Times distilled the entire AI debate into a single chart, and the punchline: the most likely outcome is not singularity or extinction. It is a modest 0.2 percentage point bump to trend growth, the same steady line that has held since 1870. That is the base case. The dramatic curves, up toward abundance or down toward human extinction, are the tails. Most of the AI conversation lives in those tails. Most of the evidence points to the middle.

→ Why does it matter? An interesting chat from Anthropic here. The gap between blue and red is the most important story in AI adoption right now. Across nearly every occupational category, AI can theoretically handle a substantial share of job tasks, but observed usage barely registers. Even in the fields leading adoption, legal, software, business, and finance, actual usage sits at a fraction of what is already possible today.

Tweets that stopped my scroll:

→ Why does it matter? On the evening of the Super Bowl, a rumor circulated about a leaked OpenAI commercial featuring a mysterious new device. I referenced it in that week’s edition of Reading Ambitiously, only to discover there were no confirmed reports of such a spot. For the first time, I felt like I had been pulled into fake news, so I removed the reference before publishing. Then, last week, that same device appeared in Joe Gebbia's hands. Call it conspiracy theory corner if you’d like, but when Jony Ive was at Apple, he prized secrecy and understood that the ecosystem feeds on leaks. The anticipation was part of the product. It is hard not to wonder whether this, too, is part of a similar playbook.

→ Why does it matter? Could a MacBook drop from over $1,000 to a projected $80 by 2081, while a McDonald’s burger climbs toward the same number? The things we build with silicon get cheaper every decade, while the things we grow, staff, and serve in person get more expensive.

→ Why does it matter? Qwen 3.5 running on-device on an iPhone 17 Pro is a big deal. When frontier-class reasoning runs locally, without a cloud call, without latency, without a subscription, the economics of AI applications change completely. This would finally mean Apple could fix Apple Intelligence! The edge is becoming the new data center, and the phone in your pocket is the new inference engine.

→ Why does it matter? Google just shipped a CLI for its entire Workspace suite. Instead of clicking through polished dashboards, the most productive users are dropping back to the terminal with Claude Code. 2026 feels like the year of CLI’s and skills.

Worth a watch or listen at 1x:

→ Why does it matter? Chris Degnan was employee #13 at Snowflake and the only person to hold the CRO seat across all four CEOs and $3.5B in ARR. He refuses to sugarcoat what scaling a sales org actually requires: relentless travel, paranoia about competition, and staying close enough to the front lines that you could close a deal yourself on any given Tuesday. He warns against becoming a “spreadsheet manager,” a CRO who hides behind dashboards until the org quietly stops trusting them.

→ Why does it matter? John Arnold built one of the most successful trading operations in history, then stepped away to focus on philanthropy. In this conversation, he applies that same systems-level thinking to China’s manufacturing velocity, the hard economics of the energy transition, and what AI-driven demand means for physical infrastructure. His observations on China’s industrial capacity alone are worth the time.

Subscribe now

Quotes & eyewash:

→ Why does it matter? has become one of my favorite recent follows. She is an entrepreneur who is now using AI to homeschool her kids, and the way she applies the most advanced tools to personalize their education is so cool and hands-on. This week, my six-year-old and I walked through a lightweight software design session together. A few prompts later, we had built a spelling test game in Claude Code, complete with Trolls. Thanks for the inspiration, Jesse.

Subscribe now

The mission:

The Wall Street Journal once used “Read Ambitiously” as a slogan, but I took it as a personal challenge. Our mission is to give you a point of view in a noisy, changing world. To unpack big ideas that sharpen your edge and show why they matter. To fit ambition-sized insight into your busy life and channel the zeitgeist into the stories and signals that fuel your next move. Above all, we aim to give you power, the kind that comes from having the words, insight, and legitimacy to lead with confidence. Together, we read to grow, keep learning, and refine our lens to spot the best opportunities. As Jamie Dimon says, “Great leaders are readers.”

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, or legal advice. Readers should do their own research and consult with a qualified professional before making any decisions.

The big idea: Top 5 lessons learned vibe coding AmbitiousOS

Reading time: 6 minutes

Fourteen days ago, AmbitiousOS was a Friday-night idea. Today, it is live in production, running a private beta with about 20 subscribers. Two weeks from idea to production would have sounded unrealistic not long ago. With Claude Code, it is the new normal.

Here are the top five lessons I learned building AmbitiousOS with Claude Code over the last two weeks.

The AI is a shadow of its creator

“An institution is the lengthened shadow of one man.” - Ralph Waldo Emerson

AI output is downstream of a dataset, and datasets are curated by humans first.

The intelligence is not new, at least not yet. It is relocated, packaged, and distributed through LLMs.

AmbitiousOS is constrained exclusively to my personal research and writing. Over the last two years, I’ve curated thousands of pieces of material that make up the dataset the model retrieves from. Before a single token is spent, the dataset has already been filtered and weighted by my human brain.

This matters because the “artificial” intelligence is an extension of the curator. It is both surprising and not surprising that AmbitiousOS ended up resembling its creator. Growing up, I was the person family members asked to set up their computers. I’ve always gotten joy out of explaining technology to others. Reading Ambitiously is that impulse in newsletter form, and AmbitiousOS is what it feels like when you turn that same impulse into an AI.

This is a lesson for anyone building or investing in AI. “AI for [insert your best idea]” is exciting, but you should look even harder at the team behind it. The best systems are built in the image of creators who care deeply about why the system exists in the first place.

The full power of a database, and I love it

I am not a trained database analyst. Claude built AmbitiousOS on Postgres, and it is astonishingly powerful. If Excel is one way to interact with data, this is a different category entirely.

Query, transform, validate, and refactor schema. I can do it all by talking to Claude with my voice. Databricks reported that AI agents now create 80% of new databases in Neon (their Postgres offering) and 97% of database branches. No surprise to me.

Working in the terminal, while intimidating at first, might be the most pleasurable UX I’ve ever used. Claude Code with direct database access, programmatic access via APIs and MCP to other systems, and the ability to speak in English and build highly personalized software. That is awesome.

Vibe coding a dashboard is one thing, getting to production is another

It is one thing to vibe code an HTML front end. It is another thing to ship software.

At one point, Claude said, “Let’s check the logs.” I went looking and realized I didn’t really have logging infrastructure. There is a lot of work that goes into manufacturing software. Yes, I could ask Claude to invent something, wire it up, and ship it. But there is a big difference between a prototype and a product.

And once you get to a product in production, it is a lot of work. I spent 2 days on the sexy features and 12 on security, deployment, maintenance, and bugs.

If you’ve spent your career in enterprise software, you already know the iceberg. AI compresses the visible work, but it does not eliminate the invisible work.

A lot of what we see AI producing is incredible demos, parlor tricks, and the “wow” moments.

What most people do not see is the manufacturing line required to get to production: security and permissions, secrets management, environments, deployments and pipelines, observability, auditability, regression testing, incident response, rollback, and the daily maintenance that starts the moment real users show up.

Most of my feature development happened last week. This week has been dominated by ensuring a safe transition to production. Claude has been a coding partner, but also something closer to a CISO and SDLC partner, pushing on infrastructure assumptions, reviewing flows, and helping test the deployment, not just the sexy features that’ll drive sign-ups.

35,000 feet

Earlier this week, one of our private beta testers wrote in to say they couldn’t log in via our authentication workflow. They sent an email with details, plus a screenshot. I happened to be on an airplane, and I fired up Claude Code at 35,000 feet on United’s Wi-Fi. I gave Claude the email and the bug.

It took a couple of iterations, but Claude quickly identified the changes we needed to make. We wrote a fix, pushed it to GitHub, and redeployed the application in a matter of minutes, all from 35,000 feet.

Then Claude drafted the reply, and I sent it to the user to let them know the fix was in. They were able to log back into AmbitiousOS.

It’s a wild time to be alive. At the same time, now that we’re live, we have an obligation to our users.

The roadmap is more AI-native, without getting sloppy

What comes next for AmbitiousOS is more AI-native capability.

I want to give it a voice. I want to give it a phone number. I want you to talk to it on WhatsApp without logging into a browser. On the roadmap are things like an MCP server and a more conversational interface, plus the less glamorous work that makes those experiences safe: persistence infrastructure, chat history, stronger security controls, and a lot more hardening.

There is an old trap in software. People assume all progress is feature progress. But once you ship, you cannot spend 100 percent of your budget on features. You are now running a software company. Focus on the visible work that excites users, while also focusing on the invisible work that keeps them safe, in a reliable, available way that brings them back.

It’s a new world. Be safe out there.

Subscribe now

Best of the rest:

🙈 Anthropic Quietly Dropped Its Hardest Safety Commitment – The most safety-conscious AI lab just scrapped its core pledge to never release models without guaranteed risk mitigations in place. – TIME

🎙️ Robert Caro on Power, Process, and the Man Who Reshaped New York – The author of The Power Broker joins 99PI’s epic mini-series breakdown of his masterpiece. Caro on Robert Moses is worth every minute. – 99% Invisible

🔗 Block shares soar as much as 24% as company slashes workforce by nearly half – Jack Dorsey cut nearly half his workforce, 4,000 people, because “AI” can now do the work. Or maybe he just way overhired during COVID. Shares soared 24%. – CNBC

💾 AI Is Bringing the Thin Client Back, and This Time It Might Stick – The PC and mobile era favored thick clients, but AI flips the logic: intelligence lives in the cloud, not the device. – Stratechery by Ben Thompson

👨‍💻 The Future of Software Engineering – A look at how AI is redrawing the boundaries of what it means to build software, and who gets to do it. – ThoughtWorks

Charts that caught my eye:

→ Why does it matter? Vertical Industry Software is the only category breaking above its 8-quarter average on both quota attainment and inbound lead sentiment. At the same time, everyone is debating whether vertical software has a future. The numbers say the opposite of the bear case: reps are hitting quota at rates not seen since early 2022, and inbound lead flow is at a multi-year high heading into Q4 2025. This is what a category looks like when the market starts to believe the moat is real, not theoretical.

→ Why does it matter? The top decile of AI-native software companies now generates roughly $650K in ARR per employee, up from under $400K just a year ago. Productivity gains from AI?!

Tweets that stopped my scroll:

→ Why does it matter? Bill Gurley is flagging the same trap that burned investors in 1999: revenue growth that looks spectacular until you ask what it actually costs to generate. If AI companies are buying compute from Nvidia and reselling it at a loss to win customers, the top-line numbers are a mirage.

→ Why does it matter? Perplexity Computer orchestrates multiple agents from OpenAI & Anthropic. GPT-5.2, Claude Opus, Claude Sonnet, running in parallel on a single financial workflow, each assigned to the task it handles best.

→ Why does it matter? As we wrote in RA (Feb 6, 2026), I think Konstanine is right: we will see consolidation and the biggest players will get bigger. But at the other end of the barbell, a key group of domain experts will matter even more, especially in fields where 90% accuracy is not good enough.

→ Why does it matter? This is Salesforce acknowledging something important. And sharing a glimpse into SalesForce, Slack & Claude working together. But where is AgentForce?!

Worth a watch or listen at 1x:

→ Why does it matter? Dan Sundheim is one of the all time greats in investing today. In this episode of ILTB, he lays with Patrick O’Shaughnessy exactly why he is making concentrated bets on Anthropic, OpenAI, and SpaceX. He talks about how his firm effectively invests across both public and private markets.

Subscribe now

Quotes & eyewash:

→ Why does it matter? Silicon Valley was so far ahead of its time. Must watch.

Subscribe now

The mission:

The Wall Street Journal once used “Read Ambitiously” as a slogan, but I took it as a personal challenge. Our mission is to give you a point of view in a noisy, changing world. To unpack big ideas that sharpen your edge and show why they matter. To fit ambition-sized insight into your busy life and channel the zeitgeist into the stories and signals that fuel your next move. Above all, we aim to give you power, the kind that comes from having the words, insight, and legitimacy to lead with confidence. Together, we read to grow, keep learning, and refine our lens to spot the best opportunities. As Jamie Dimon says, “Great leaders are readers.”

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, or legal advice. Readers should do their own research and consult with a qualified professional before making any decisions.

The big idea: Building AmbitiousOS v2 + Total Recall & Memory + AI Agents w/ Claude Code

Reading time: 8 minutes

On Friday night, I pulled a stack of journals off the shelf.

Throughout my career, I have taken many notes. Spiral notebooks. Hardbound journals. iPad note-taking apps. Evernote. The notebook I took everywhere in the first 200 days of Ridgeline. Every person we met. Their advice. Our goals.

Reading Ambitiously is an extension of that habit. One hundred and one editions of how technology, markets, and ambition are evolving. In many ways, it is my public journal.

And on Friday night, I had an idea.

What if subscribers could talk to it? Not search it. Talk to it. What is Reading Ambitiously? Could it become a part of their team?

In about 30 minutes, Claude and I had a plan. We built it, and we’re going to make it available to you.

Memory

Memory is a funny thing. The human brain is extraordinary at pattern recognition and terrible at retrieval.

Every technology era believes it has solved the knowledge problem. In the early 2000s, we built document management systems. Taxonomies. Keyword indexing. Then came the “second brain” movement.

All useful. All still dependent on one thing: you had to remember the path. And spend a lot of time gardening to keep it clean.

Search assumes you know what you are looking for. Memory allows you to ask what you have forgotten.

The latter of which AI is very good at.

From Newsletters to AmbitiousOS v2

There are tools that claim to do this (Glean, Notion AI, various “chat with your docs” products). I have tried a few. They feel heavy. You do not quite understand what they are retrieving or why. Plus, I wanted to do this to understand what’s going on technically here.

The first step was building a vector database. If you are new to that term, think of it as a way to store meaning instead of keywords. Each paragraph becomes a mathematical representation (an embedding) that captures its semantic shape. Similar ideas sit near each other in that space, even if the words are completely different.

Every edition of Reading Ambitiously was pulled from Substack. Not easily, I should add. But we wrote code to fetch the full archive, then chunked each edition into smaller sections: roughly 300 to 900 tokens at a time, preserving headings so the structure remained intact. Each chunk was embedded and stored with metadata (title, date, tags, entity references).

In our first load, we went with all editions available on Substack, which is 59. About 5,000 chunks. Indexed and stored.

From there, we layered on more intelligence and tooling:

• A retrieval system that searches by meaning.

• A temporality layer that detects when you are asking how your thinking has evolved, and sorts results chronologically.

• A knowledge graph that maps 460 entities (companies, people, concepts, frameworks) and how they connect across editions. Claude read every piece of content, extracted the entities, and built the graph automatically.

• An AI agent that does not just search once and respond. It decides which tools to use, searches the archive multiple times, checks the knowledge graph for connections, and synthesizes everything into a grounded answer with citations.

• Persistent memory so the system does not reset every session.

If I’ve lost you in the technical weeds, I’m sorry! This is technical. But what mattered was what happened next.

The Moment It Clicked

I asked a simple question: “Tell me every time I’ve mentioned AI pricing.”

The system did not just return quotes. It synthesized across years. It cited the original editions. It highlighted the first mention and recent usage. It showed how the framing shifted. In the UI, I could see the agent working: “Ambitiously thinking...” with a timeline of each step (searching the archive, looking up entities, finding connections) leading up to the answer.

A few seconds. Doing this manually across 100+ editions would have been an afternoon.

“I have a meeting about AI agent pricing tomorrow. Refresh my thinking: what have I written, what frameworks apply, and what is my current position?”

It pulled relevant chunks from the archive, cross-referenced the knowledge graph, and gave me talking points. Not generic AI-generated ones.

Then I asked one more: “How has the thinking on enterprise SaaS moats evolved over time?”

It detected the temporal nature of the question, sorted everything chronologically, and showed me how my position developed. Early editions versus recent ones. Where I changed my mind. Where I doubled down.

It felt less like a search and more like total recall. Search finds a file. Memory has a conversation with your own thinking.

Near-Perfect Memory, With Limits

Software is moving from tools you open to systems that respond to intent. This is the personal version of that shift. You do not open a folder. You ask. And because the corpus is yours, because the embeddings are legible, because you can see exactly which editions it pulled and why, you understand how it works. Feels like magic, but it’s AI!

The implication is this: near-perfect memory will belong to those who choose to document.

If you want leverage over your own thinking five years from now, you need to give your future self a corpus. Journals. Investment memos. Product specs. Friday notes. The format matters less than the habit. Write it down.

The model is not the asset.

The archive is.

We’re launching in private beta today. The plan is to make AmbitiousOS generally available to Reading Ambitiously subscribers next week. If you’re interested, please reach out.

Subscribe now

Best of the rest:

🤖 Introducing Agentic Wallets: Give Your Agents the Power of Autonomy - Coinbase is productizing the missing piece for real agent autonomy, wallets with built-in skills, spending guardrails, and the x402 rails so agents can pay, trade, and operate without waiting on humans. - Coinbase Developer Platform

🔋 The AI Productivity Paradox – An engineer who builds AI agent infrastructure shipped more code than ever last quarter. He was also more exhausted than ever. The two facts are connected. – Siddhant Khare

🐝 Anthropic’s Talent Density Is NFL-Level – Steve Yegge talked to 40 Anthropic employees and came away convinced they’re building a “hive mind” for software development. Harder to get hired there than making the NFL. – Medium

Charts that caught my eye:

→ Why does it matter? This is the fastest revenue ramp in enterprise software history. Going from $0 to $14B in annualized revenue in roughly two years, with 10x-plus growth each year, puts Anthropic in rare company. For context: it took Salesforce ~20 years to hit $14B in annual revenue.

→ Why does it matter? How many years did it take for human beings to create fire, build shelter, and hunt with spears? About 2.2 million years. We tend to think humanity progressed like the red line, but it actually looked more like the green line. The gap between those lines is where most organizations lose patience and quit. The ones who stay in the game during that flat period? They’re positioned at the elbow when everything compounds at once.

→ Why does it matter? New research from Anthropic on how Agents are being deployed, how much autonomy people are granting, and where they’re deployed. The full study can be found here.

Tweets that stopped my scroll:

→ Why does it matter? I’m sure this has hit most of your feeds by now. Sam Altman and Dario Amodei refusing to hold hands is priceless.

→ Why does it matter? This is an entirely AI-generated 3.5-minute movie. Wow.

→ Why does it matter? Many readers are setting up Claude Code as their Chief of Staff. If last year was the year of AI Agents, perhaps this year is that of Personal Operating Systems. Check out this short video for some great inspiration.

Worth a watch or listen at 1x:

→ Why does it matter? Dario Amodei doesn’t do many long-form interviews. When he does, he’s remarkably precise with language. This conversation covers the end of the scaling exponential, what “a country of geniuses in a data center” actually means for labor markets, and Anthropic’s positioning against OpenAI. The insight density is high. Amodei’s careful phrasing reveals as much as his answers. You’ll want to catch the pauses.

→ Why does it matter? Great to be with Marley Kayden to talk all things AI for Asset & Wealth managers, as well as what we’re seeing in the overall software market! Thank you for having Ridgeline on the show!

→ Why does it matter? Bret Taylor is the founder and CEO of Sierra, an AI agent company transforming customer service. Bret is one of the brightest minds on AI-native company building.

→ Why does it matter? Boris Cherny built Claude Code as a simple terminal-based prototype just a year ago, and it has already begun reshaping how software is built and how knowledge work is done. The details are the point: Claude Code’s growth to 4% of public GitHub commits, daily active users doubling last month, and the counterintuitive product principles behind it. When Boris says coding is “solved,” the claim is really about where work is moving next, from writing lines to directing systems.

Subscribe now