You’ve had dozens, maybe hundreds, of conversations with Claude. Each one left a trail: how you prompt, what you build, where you hit walls, what workarounds you’ve invented, where your best ideas are sitting in threads you’ll never go back to.

Taken together, that trail is a remarkably detailed portrait of how you actually use AI, and it almost certainly looks different from how you think you use AI.

Today, we’d like to share the AI Mirror skill with you: a custom skill we’ve created that takes your raw Claude.ai chat export into a single, self-contained HTML dashboard that mirrors how you’ve really been using AI back to you.

It reads your entire conversation history, measures where you’re losing time and where your best thinking is trapped in old threads, and then translates all of it into a clear picture of what Claude Code could change for you.

The dashboard is built to be explored: headline numbers sit on top, and every claim opens into the full underlying analysis with a click. It reports on six areas:

• Big Picture Stats. How much work have you actually done inside Claude? Your total conversations, total words, and a plain-language read on the scale of what you’ve already built. Most people are surprised by how high these numbers actually are.

• The Chat Tax. How much of your time is going to busywork you shouldn’t have to do? The report counts how many conversations open with copy-pasted prompts, how many words you’ve retyped from scratch, and estimates the hours you’ve lost to repetitive setup.

• The Cost of Fragmentation. How much of your best thinking is stuck in conversations you’ll never reopen? The report maps how scattered your work is across separate chats on the same topic, identifies the insights and original IP trapped in threads you haven’t revisited, and shows how often you’ve had to search your own history for something you already wrote.

• Your AI Sophistication Levels. Where do you actually fall on the expertise scale? Two seven-level frameworks (one for Chat, one for Claude Code) graded from your real behavior, each with a “your next move” prompt. It also inventories the workarounds you’ve invented to cope with chat’s limits, so you can see just how much engineering you’ve been doing without a proper toolkit.

• Opportunities. What would change if you stopped working around chat’s constraints? This is the payoff section: an executive summary of what’s possible, a prioritized Claude Code roadmap of skills and pipelines worth building, and a personalized recommendation for the single highest-impact project you could start with.

• The Bigger Picture. What does all of this say about you? The closing section steps back from the numbers to show what your data reveals about your relationship with AI tools, and where migrating to an agentic workspace could help you work more efficiently, with better results, at greater scale.

• Skills and a Workflow Based On Your Previous Work in Chat. The analysis identifies the top three skills and #1 workflow that would help you do work you already more easily, more quickly, or at a higher level. Then the AI Mirror for Claude builds those skills and workflow for you.

This skill is designed to be used in Claude Code. (It will not work in the claude.ai web chat interface.)

The rest of this article contains instructions for using the AI Mirror for Claude skill in the Claude Desktop app. (If you haven’t installed Claude Desktop on your computer yet, you’ll find PDF tutorials in Appendix A, at the end of this post.)

Here’s an overview of the AI Mirror for Claude process:

1. Export your Claude Chat history.

2. Create a folder called claude-export. Copy the exported data into this folder.

3. Download the AI Mirror for Claude skill.

4. Open Claude Code (in the Desktop app) and select the claude-export folder.

5. Drag the zipped skill file into the Claude Code Desktop prompt field, then type: “Run this skill on the data file in this folder.”

Claude Code will run the analysis and open the report dashboard in a new window of your default web browser.

Keep reading for detailed tutorials on each step…

Export Your Claude Chat History

We’ve created this Chat Exporter Tool to make it easy to get your complete Claude Chat history easy. Click on the Chat Exporter icon:

The Chat Exporter Tool appears:

Click on the Go to Claude Export Page button. A popup box will appear.

Click on the Open link button.

You’ll be taken to the exact page in your Claude account settings where you can request the data export:

Click on Export data. The Export data popup window will appear:

Make sure All is selected, then click on the Export button. In the upper right corner of your screen, an Export started notification appears.

Within 15 minutes, you’ll receive an email from Anthropic containing the download link for your data:

Click on the Download Data button in the email, and your data file will begin downloading as a zipped file.

Create a new folder on your computer called claude-export and copy the data file there. Do not unzip the data file.

Download the AI Mirror for Claude Skill

Click this button to download the AI Mirror for Claude skill:

AI Mirror for Claude

You’ll be taken to the skill’s zipped file:

Click on the Download icon in the upper left corner of the screen. The file should begin downloading.

Open Claude Code (in the Desktop app) and Select the claude-export Folder

Once the download is finished, open the Claude Desktop app:

In the upper left corner, click on the </> Code tab.

In the lower half of the screen, click on the button to the right of Local to change which folder you’re working in.

A popup menu appears.

Click on Open folder…

Navigate to the claude-export folder you created earlier and select it. Then click the Open button.

The claude-export folder is now your current folder.

Drag the Skill File Into the Desktop App and Run It

Open a Finder window and open the folder that you downloaded the ai-mirror-for-claude.zip file to.

Then drag the .zip file into the Claude Desktop prompt field.

The .zip file appears in the prompt field as a gray box labeled “zip.”

Type into the prompt field: “Run this skill on the data file in this folder.”

Press Enter. The AI Mirror for Claude skill begins to analyze your data.

Claude Code will periodically ask you for permission to execute various tasks.

For tasks that must be executed once, it will ask you to Allow Once.

For tasks that must be executed multiple times, you will be given a choice to Allow Once or Always Allow. (To save on mouse clicks, choose Always Allow whenever possible.)

At some point in the process, Claude may also ask you to verify that you trust this workspace (which should be your claude-export folder):

Clikc on Trust Workspace.

Once the AI Mirror skill has surveyed your chat history, it will ask 1-2 followup questions to be sure it’s interpreting your history correctly:

Depending on how large your chat history is, it may take up to 15 minutes or so to finish the analysis and create your report.

When the skill is complete, Claude will share a link to the .html dashboard as well as a summary of highlights:

The report dashboard will open in a side window, which you can expand as needed.

Note: Claude will also create a subfolder called output in your claude-export folder, where it will save both the .html file containing your dashboard, as well as each section of the dashboard in a markdown file, which you can view in any app that allows you to view markdown files (like Obsidian).

Appendix A: Installing the Claude Desktop App

If you haven’t installed the Claude Desktop app yet, here are PDFs containing instructions for installing it on both Mac and Windows computers.

Installation Guide for Mac:

Install Claude Desktop (Mac)

Installation Guide for Windows:

Install Claude Desktop (Windows)

I still think prompts matter.

A good prompt can make a big difference. It can clarify the task, narrow the output, set the format, and stop the AI from wandering into something you did not ask for.

But at some point, I started running into a weird problem.

The better I got at prompting, the more prompting I had to do.

If I wanted useful output, I had to explain who I was, what I was building, who my audience was, what my standards were, what I had already written, what I wanted to avoid, what kind of answer I trusted, and what kind of answer felt generic.

So the prompt kept getting longer.

And longer.

And longer.

At first, that felt like progress. I was getting more specific and giving the AI more direction. I was becoming a better operator.

But after a while, it started to feel backwards.

If AI is supposed to save me time, why am I spending so much time rebuilding the same instruction every time I use it?

That is the part I missed for longer than I want to admit.

The prompt was not the whole system. The prompt was carrying too much of the system.

Subscribe now

When The Prompt Is Doing Too Many Jobs

Most AI advice still treats the prompt as the main thing.

Write a clearer prompt. Add more context. Include examples. Define the role. Specify the output. Give constraints.

That is why you see so many people sharing prompts like the prompt itself changes everything.

And look, prompts can be really useful. I am not dismissing them.

But if the whole system is built on top of one long prompt, I do not think that changes how we work in any durable way. It just makes the instruction longer.

When every useful result depends on a giant prompt, the prompt stops feeling like a useful shortcut and starts feeling like manual labor.

You are not just asking for the task anymore. You are cramming your whole work history into the request: your audience, your voice, your standards, your goals, your past examples, your current constraints, and your opinions about what good work looks like.

That can work for one request. It gets tiring when you do it every day.

And it creates another problem: if you forget one piece, the output changes.

Maybe the answer becomes generic, misses your voice, or gives you ideas you would never publish.

At that point, the issue is not only the prompt.

The AI does not have enough truth to work from.

That is the part I think gets missed in prompt advice.

A prompt is a request. It should not have to be the storage layer for everything true about your work.

What I Mean By Source Material

When I say source material, I do not only mean research files.

I mean the material that tells AI what is true about you, your work, your standards, your audience, your constraints, and your way of thinking.

For me, that can include:

1. Past newsletter posts

2. Audience notes

3. Writing guidelines

4. Performance patterns

5. Project goals

6. Examples of good output

7. Examples of bad output

8. Decision rules

9. Subscriber survey notes

10. Raw thinking from my own notes

Source material is the truth I want the AI to use as a reference.

It tells the AI: this is what I care about, this is how I think, this is what good looks like here, this is what to avoid, and this is the work we are actually doing.

Without that, the AI has to guess.

And when AI guesses, it often guesses in the most average way possible.

That is why the output can sound polished but still feel wrong.

It may be well written and useful in a generic way, but it does not sound like you, because it does not reflect your standards, and it does not understand the tiny decisions that make your work your work.

A Prompt Can Ask. Source Material Can Show.

This is the difference I keep coming back to.

A prompt can say, “Write in my voice.”

Source material can show what your voice actually sounds like across twenty examples.

A prompt can say, “Give me ideas for my audience.”

Source material can show who your audience is, what they struggle with, what they respond to, and what topics you have already covered.

A prompt can say, “Make this strategic.”

Source material can show your goals, constraints, past decisions, and the kind of tradeoffs you usually care about.

A prompt can say, “Do not make this generic.”

Source material can show the difference between something you would publish and something you would reject.

That is a different kind of instruction.

The prompt tells the AI what you want right now.

The source material tells the AI what world it is working inside.

Share

The Newsletter Ideas Example

One simple example is post ideas.

If I ask AI for ideas without much context, I usually get something that sounds fine but thin.

It might suggest topics like:

1. How to use AI for productivity.

2. The future of AI agents.

3. Best AI tools for creators.

4. How AI will change work.

None of these are terrible.

But they are not very useful either. They could belong to anyone. They do not know my readers, my archive, my current direction, or the kind of ideas I am tired of repeating.

Before, I would try to fix that with a longer prompt.

I would explain AI Maker. I would describe the audience. I would explain the current content arc. I would add examples. I would add constraints. I would say what to avoid.

Again, that helped.

But now I think the better version is to give AI better source material first.

If the AI can use my archive, audience profile, writing standards, project context, and performance patterns, the ideas change.

They become more grounded in what I have already written. They avoid topics I have already overused. They connect to the actual pain points my readers have. They fit the way I frame AI: practical systems, less repeated setup, source material before output, workflows that become more useful over time.

The question can become much simpler.

“Can you give me ten new ideas to write about?”

The output is better because the AI is no longer trying to invent my world from one prompt.

It can read the world first.

The Social Post Example

Another example is repurposing my newsletter.

I used to have a long prompt for turning one newsletter into LinkedIn posts, Substack Notes, and Twitter threads.

The prompt had to explain the platform, the tone, the structure, the ending, the things I liked, the things I hated, and the difference between a summary and a standalone social post.

If I skipped those details, the output usually drifted.

LinkedIn sounded too polished. Substack Notes sounded like tiny summaries. Twitter threads tried to compress the whole article instead of pulling out one strong idea.

Now the request can be much simpler:

“Turn this newsletter into social.”

That sentence only works because the actual instructions live somewhere else.

The AI can see the framework. It can see the examples. It can see the rules for each platform. It can see the standards I already use.

So the prompt becomes a request instead of a suitcase.

That is the shift.

Why This Changes The Work

Better prompts improve the conversation.

Better source material improves the starting point.

That matters because most of the frustration people feel with AI comes from asking it to do grounded work without giving it grounding:

• You want it to sound like you, but it has no real examples.

• You want it to understand your audience, but it has no audience notes.

• You want it to follow your standards, but the standards live only in your head.

So it does what AI often does: it gives you a plausible answer.

Plausible is not enough for real work.

Real work needs something more grounded. It needs the AI to understand the material around the task, not just the sentence you typed into the chat box.

The Source Material Audit

Here is the small test I would use.

Pick one AI task you repeat often.

Maybe it is brainstorming ideas. Maybe it is turning a draft into social posts. Maybe it is summarizing research. Maybe it is reviewing your writing. Maybe it is preparing for a meeting.

Then ask:

1. What do I keep explaining every time?

2. What examples would make the answer better?

3. What standards does the AI need to follow?

4. What past work should it be able to reference?

5. What mistakes do I keep correcting?

6. What goals or constraints would change the answer?

Those answers are source material.

You do not need to make this complicated. Start with one repeated task and one small folder of truth.

For example:

1. A short audience note.

2. Three examples of good output.

3. A few standards you care about.

4. A list of things to avoid.

5. A simple note about your goal.

That alone can change the quality of the work.

Not because the prompt becomes magical.

Because the AI finally has something real to work from.

The Next Upgrade

I am not saying prompts do not matter.

They do.

But many people have reached the point where the next improvement will not come from one more perfect sentence.

It will come from better source material.

That is what I would pay attention to now.

If AI keeps giving you generic output, do not only ask, “How can I prompt this better?”

Ask, “What truth is the AI missing?”

Because the missing truth is often the real reason the answer feels wrong.

Join Our Upcoming Cohort To Go Deeper That Starts On June 22

This post gives you a flavor of what we teach at The Agentic Academy for Knowledge Work. It’s one of many lessons that cumulatively transform how you work with AI and 10x your productivity.

To get the full experience, join the upcoming cohort:

Join Cohort

When I was 16, a book called Unleashing The Ideavirus rewired my brain. Seth Godin said one thing that stuck with me: ideas don’t just happen to spread. They’re engineered to.

That idea sent me on a 25-year obsession.

I spent thousands of hours dissecting why certain ideas break through while seemingly good ones die in obscurity. As a result, I was able to go from a failing blog to publishing articles that have been read cumulatively over 100 million times across Forbes, Fortune, TIME, the World Economic Forum, and the Harvard Business Review.

But my comprehensive approach to reverse-engineering had a cost. On average, a single article took 60+ hours. The method demanded exhaustive cross-disciplinary research and 15 drafts per article. Two tradeoffs haunted every project:

• Quality vs. quantity

• Augmentation vs. automation

I opted for quality over quantity and augmentation over automation. Every article was a chance to follow my curiosity and grow.

I accepted those tradeoffs for a decade.

This year, Claude Code + Opus 4.6 made them obsolete. Now, it’s possible to get the best of both worlds.

This article is the proof.

It was created using an AI thought leadership system that produces high-quality content quickly, at scale. The system that 16-year-old me was actually looking for, even if he didn’t know it yet.

For the first time ever, I’m now sharing that system with a small group of pioneering entrepreneurs and senior executives at $1M+ companies who want it installed for themselves, their employees, or their company. If you’re interested, reply with BLOCKBUSTER in the comments or to this email, and I’ll send you more details.

Now on to the article…

Quick Summary

I believe that the most important AI decision you make this year is to move from the chat paradigm (ChatGPT, Gemini, Grok, Claude.AI) to the agentic paradigm (Codex, OpenClaw, and Claude Code).

I’ve now talked to dozens of people about how they’re making this decision. And there are a ton of misconceptions.

For example:

1. Many think that Claude Code is just for programmers. It’s not.

2. Many feel overwhelmed by all the new tools and think Claude Code is like the others. It’s not. It’s way more important.

3. Many love the Chat paradigm, and they don’t realize the hidden taxes they’re paying and the opportunities they’re missing.

This article provides one of the most comprehensive overviews of the difference between the two paradigms, so you can make your most important AI adoption decision better.

The rise of the agentic paradigm is the new ChatGPT moment, and adopting it sooner will have a profound impact on your life.

Consume In Your Preferred Format

Slide Deck

Access Slides

Audio

The audio doesn’t just narrate the text in the article below. It actually creates a script tailored to the listening experience, so you can easily consume the ideas while driving, taking a walk, or doing chores.

Full Article

In the 1880s, the electric motor arrived in American factories. The factory owners did exactly what you’d expect. They ripped out their steam engines and dropped electric motors in the same spot. One giant motor in the center of the building, connected to the same system of shafts, belts, and pulleys that had distributed power from steam.

The result? Almost no productivity gain. For thirty years.

Economists Paul David and Chad Syverson documented this lag extensively. Factories got a better power source but kept the old architecture.

To see why that mattered, picture a steam-powered factory.

One massive engine in the basement spun a central shaft running the length of the building. Leather belts dropped from that shaft to every machine on the floor. If a machine wasn’t within reach of the belt, it didn’t run.

That single constraint dictated everything. Machines were arranged by proximity to the shaft, not by the logic of the work. Workers carried half-finished parts back and forth across the room because the layout served the power source, not the product.

Then came the electric motor. And for thirty years, factories just swapped it in — tearing out the steam engine and running the same shaft off a big electric one. Faster, cleaner, structurally identical.

The constraint was gone. The behavior wasn’t.

It took a full generation before Henry Ford realized the obvious. Electric motors didn’t need to be centralized. You could put a small motor on each machine. And once you did, you could rearrange the entire factory floor around the flow of work instead of the flow of power. The assembly line was born:

That’s when productivity exploded. Not when the technology arrived, but when the architecture changed to match it.

As the economic historian Paul David observed:

The productivity gains from electrification achieved their full flowering only after manufacturers ceased trying to adapt dynamo technology to the mechanical drive factory and began constructing factories around the new technology.

Think about what David is really saying here. The technology wasn’t the bottleneck. The mental model was. Factory owners kept trying to make a revolutionary technology fit inside a pre-revolutionary structure, and then wondered why it didn’t feel revolutionary.

This pattern didn’t just happen once. It happens every time a revolutionary technology meets an old way of working.

What’s going on here?

Why do smart people keep making this same mistake?

And what does it have to do with how you’re using AI right now?

The Pattern That Keeps Repeating

The factory story isn’t unique. The same pattern (revolutionary tool, old architecture, disappointing results) repeats across every domain I’ve studied...

Exhibit #1: Personal Computing.

The first spreadsheet users treated VisiCalc like a faster calculator. They’d compute one cell, write down the answer, clear the screen, and compute the next. It took years before people realized the power wasn’t in any single calculation. It was in the connections between cells. The same data updated everywhere simultaneously. The spreadsheet wasn’t a better calculator. It was a different category of tool entirely. But only if you stopped using it like a calculator.

Exhibit #2: Photography.

When digital cameras first appeared, professional photographers used them exactly like film cameras. One shot, careful composition, review later. They didn’t exploit the fact that digital eliminated the cost of experimentation. It took a new generation of photographers, people who’d never internalized the “film is expensive” constraint, to discover that the real advantage was shooting thousands of frames and finding the one that captured something no amount of careful composition could have planned.

Exhibit #3: Medicine.

When electronic health records replaced paper charts, most hospitals just digitized the paper. Same forms, same workflows, now on a screen. The result was that doctors spent more time on documentation, not less. It wasn’t until systems were redesigned around what digital made possible (shared records, automated alerts, pattern detection across thousands of patients) that the technology delivered on its promise.

Exhibit #4: Music Production.

When digital audio workstations replaced analog tape, the first generation of producers used them as better tape machines. Record, edit, mix. Same linear workflow. The breakthrough came when producers realized digital audio could be non-linear. You could remix, layer, and restructure infinitely without degradation. That insight created entirely new genres of music.

The pattern is always the same.

A new technology arrives. People use it to do the old thing slightly faster. They get modest improvements and a lot of new frustrations. Then someone realizes the technology enables an entirely different way of working. And that’s when the real transformation happens.

Bottom line: The value of a revolutionary tool is never in doing the old thing better. It’s in doing a new thing that the old tool made impossible.

And right now, with AI, almost everyone is still doing the old thing.

Including me. That is, until the arrival of Claude Code...

The Chat Trap: Four Walls That No Amount Of Prompting Can Fix

Most AI advice for the past three years has boiled down to one idea: get better at the conversation.

• Write clearer prompts.

• Provide more context.

• Learn the right frameworks for talking to AI.

This advice isn’t wrong. But it has a ceiling. And if you’re reading this, you’ve probably already hit it.

The ceiling isn’t about your skill level. It’s about the category of tool you’re using. Conversational AI (ChatGPT, Claude Chat, Gemini, Grok) has four structural limitations that no amount of prompting expertise can overcome. I call them The Four Walls.

For most people, these four walls are invisible. As a result, most knowledge workers are resistant to moving over to Claude Code.

Chat Wall #1: The Copy-Paste Tax

Here’s a workflow that might sound familiar.

You have a brilliant conversation with Claude about your next article. The AI produces excellent structural suggestions, a compelling hook, and three cross-domain examples you hadn’t considered. Great output.

Now what?

You copy the relevant pieces into a Google Doc. You reformat them. You open a new chat to work on a different section. You paste in the context from the first conversation so the new chat understands what you’ve already decided. You get new output. You copy that into the Doc too. You realize the AI in the second chat contradicted something from the first chat. Of course it did. They don’t know about each other.

By the end of this process, you’ve spent as much time managing the information flowing in and out of AI as you spent doing the creative work itself.

Let me put some rough numbers on this. A typical heavy Chat user might experience the following friction on any given day and not even realize it:

• Re-establishing context: 4-6 times per day, ~5 minutes each = 20-30 minutes. Pasting in your voice guide, your project requirements, reminding the AI what you already decided in a different chat.

• Copy-pasting between chats and documents: 15-20 transfers per day, ~3 minutes each = 45-60 minutes. Copying output into Google Docs, pasting context from one chat into another, reformatting along the way.

That’s over an hour per day of pure overhead. Not creative work. Not thinking. Not producing. Just shuttling information between tools that can’t talk to each other.

You didn’t eliminate busy work. You traded one kind for another.

Chat Wall #2: The Upload Wall

You have 200 research notes. Or a decade of journal entries. Or an archive of 500 podcast transcripts. You want AI to work with all of it. Find patterns. Surface connections you’ve missed.

But Chat only lets you upload about 20 files at a time. In specific formats. With size limits.

So you manually select a batch. Convert any files that are in the wrong format. Upload. Wait. Run your prompt. Then do it again for the next batch. And even after you get files in, Chat can only read them. It can’t modify the originals. Every output is a copy that you have to manually merge back.

You’ve become a human file shuttle. Selecting, converting, uploading, downloading, merging. Over and over. So that AI can access the information that’s already sitting right there on your computer.

This is the equivalent of the factory owner who had to manually carry materials from one machine to the next because the floor layout was designed around the central shaft, not the flow of work.

Chat Wall #3: The Lock-In

Your files can’t get in easily. But here’s the other side of that problem: your work can’t get out.

Every insight you’ve had, every framework you’ve refined, every perfect prompt you’ve crafted, every artifact you’ve created inside Chat lives on someone else’s server. In a proprietary format. Accessible only through their interface.

Want to switch from Claude to ChatGPT because one is better for a specific task? You start from zero. Your Claude Projects don’t transfer. Your ChatGPT custom GPTs don’t export. The six months of refined context you built inside one platform stays inside that platform.

This isn’t just inconvenient. It’s strategically dangerous. You’re building your most valuable intellectual work inside a container you don’t control. If the platform changes its pricing, degrades its quality, discontinues a feature you depend on, or gets leapfrogged by a competitor, your options are: stay and accept it, or leave and lose everything you built.

The more sophisticated your use of Chat, the deeper this lock-in gets. The power user with 50 carefully constructed Projects and a library of custom instructions is the one who can least afford to leave. The tool rewards your investment by making you more dependent on it.

Your files can’t get in. Your work can’t get out. And the longer you stay, the harder it is to leave.

Chat Wall #4: Context Rot

This is the subtlest wall, and maybe the most dangerous, because you don’t notice it happening.

Imagine you’re working on an ambitious project in Chat. A long article. A detailed strategy. A complex analysis. The conversation grows to 40, 60, 80 messages. You’re deep in it. You feel like the AI is tracking everything perfectly.

It’s not.

As conversations grow, the context window fills up. Earlier instructions get pushed toward the edges of what the model attends to. The AI starts quietly losing track of things you told it 40 messages ago. Your voice standards. Your structural requirements. Specific decisions you made early in the conversation.

It doesn’t warn you that it’s forgotten. It just drifts. Confidently producing output that contradicts what you established at the start. You catch some of these contradictions. You don’t catch all of them.

This is context rot. The silent degradation of AI quality as conversations grow longer. The more ambitious your project, the worse it gets.

The cruel irony: the most sophisticated, ambitious use of Chat is exactly the use case where Chat fails most badly. It’s as if the factory got less efficient the more machines you added to the central shaft. Which is exactly what happened.

What I’m Calling This

These four walls (the Copy-Paste Tax, the Upload Wall, the Lock-In, and Context Rot) aren’t bugs. They’re not things that will be fixed in the next model update. They’re structural features of conversational AI.

Together, they form The Chat Trap.

The Chat Trap is what happens when you get really good at a tool designed for one-off exchanges and try to use it for cumulative, system-level work. The better you get, the harder you hit the ceiling. The more context you generate, the more overhead you create to manage it. The more ambitious your projects, the more the architecture fights you.

And it’s not 1 + 1 + 1 + 1 + 1. These four walls multiply each other. Context rot makes the Copy-Paste Tax worse (you’re shuttling context that may already be degraded). The Daily Reset makes the Upload Wall worse: projects preserve your source files, but not the thinking you did with them. The Lock-In makes everything worse (the deeper you go, the harder it is to try alternatives). The compound effect is an overhead burden that grows faster than your productivity.

The conventional wisdom says: “Get better at the conversation.”

What if the answer isn’t a better conversation, but an entirely different relationship with AI?

The Crossing: From Conversation to System

Technology alone is rarely enough to create significant benefits.
Georgios Petropoulos & Erik Brynjolfsson (MIT & Stanford researchers)

Here’s what I was wrong about when I first hit this ceiling. And it took me months to admit it.

There are now two fundamentally different ways to work with AI. Not two products. Two categories.

Category 1: Chat AI. You talk to AI. It responds. Each session is self-contained. You manage the context, the files, and the workflow. The AI is a conversation partner. Brilliant, but amnesiac.

Category 2: Agentic AI. You direct AI. It takes action. It reads your files locally, searches the web, creates documents, deploys multiple sub-tasks in parallel, and builds things that persist on your computer as files you own. The conversation is the interface, but the output is infrastructure. Systems that compound over time.

This is a category shift, not a feature upgrade.

Think about the difference between a calculator and a spreadsheet. A calculator answers one question at a time. You punch in numbers, get a result, write it down, clear the screen, start over. A spreadsheet is a system. Cells reference other cells, formulas update automatically, one change cascades through the entire model. Both do math. But they’re fundamentally different things.

The calculator is a tool. The spreadsheet is infrastructure.

To make the most with AI, you 1,000% need a spreadsheet. Not a calculator.

What This Actually Looks Like: A Case Study

Let me show you what actually happens when you cross from one category to the other.

This is a real thing I built in the last few months.

I’m not a programmer. I’ve never been a programmer. I don’t know Python, JavaScript, or any programming language. I’m a writer, a teacher, and someone who’s spent 25 years reading books and thinking about mental models.

Case Study: Article Factory

What Writing Was Like in the Chat Trap: Writing a big article used to take me weeks. Research across multiple domains. Synthesizing dozens of sources. Finding cross-disciplinary connections. Structuring the argument. Writing. Editing. Quality control. Each step was a separate Chat session, each starting from scratch.

What I built: A 30+ step article production system. When I give it a topic or rough idea, it automatically follows the stages below over the course of an hour:

1. Classifies the input and identifies research directions

2. Deploys four parallel research agents across different disciplines, mediums, and platforms

3. Pulls from relevant paradigms, mental models, historical cycles, and lenses to understand everything and put it in context.

4. Brainstorms dozens of ideas and narrows them down to a shortlist via an innovation tournament

5. Consults a panel of expert personas to pressure-test the idea I selected

6. Synthesizes everything into an article architecture

7. Writes a full 4,000-6,000-word draft following my voice guide

8. Runs a 12-dimensional quality audit against my specific standards

The system produces a first draft that’s deeper than what I could write alone. It’s not that the AI is smarter than me. It’s because the system applies my 25 years of accumulated expertise simultaneously, across more dimensions than I can hold in my own head at once. My mental models, my cross-domain research habits, my quality standards, my voice. All running in parallel.

What I typed: A one-paragraph description of the idea I wanted to explore.

What I got: A production system that now exists permanently. Every article I write from now on benefits from it. And it gets better every time I refine my standards, because the system updates automatically. This article is proof that it works.

Other Things I’ve Built

The article factory is just one of dozens of systems that I’ve built. Below is a sampling of 10, to give you an idea of things I’m doing in Claude Code that would’ve been impossible or impossibly difficult in just Claude chat.

1. Text To RSS. Turns any article into a podcast episode and puts it on an RSS feed that I can listen to on Snipd. When I come across a long article, I send it to this skill so I can consume it while walking.

2. Mental Model Manual Writer. Produces a 10,000-30,000-word deep-dive manual on a single mental model. I can create multiple models at once with parallel agents. So far, I have created 300+ manuals, which are used in my content creation process.

3. AI Second Brain Creator. I created this skill that gathers all of my notes, podcast clips, books, transcripts of YouTube videos I’ve watched or subscribed to, and articles I’ve liked, then ingests them into 5K+ interconnected atomic notes forming a navigable knowledge graph that AI can use to provide personalized responses to me to help me develop articles.

4. Podcast Enricher. Goes through hundreds of podcast episodes that I’ve clipped in my library, finds the corresponding YouTube video, transcribes the full episode, downloads the full video, and creates video clips to the exact video moment I highlighted. It automatically runs every day at 9:30am EST to ingest new podcast episodes that I’ve clipped.

5. Proposal Generator. Automatically builds sales proposals for my thought leadership consulting offers using my business positioning, models from five of my favorite sales books, prior proposal templates, and transcripts from sales calls.

6. Substack Scraper. Point it at any Substack publication and it pulls down every single post, saving each as its own clean document with the original link preserved.

7. Deep Prospect Finder. Builds a searchable database by sweeping 23+ platforms, then tiers them and writes personalized outreach angles for each one.

8. Second Order Show Producer. Stitches together my daily video clips created in the Podcast Enricher (see above), then has a cloned avatar of me introduce each clip and explain the second-order consequences.

9. Title Factory. Reads the actual performance data from my past articles and 4,000 A/B tests, generates 30 high-performing titles, then creates actual cover-image variations for the best ones.

10. Idea Machine. Takes a single seed idea and methodically expands it into dozens of related angles, variations, and adjacent concepts, saving each one so I can come back and mine them later. This skill was created by analyzing the transcripts of hundreds of my classes.

This is just a sampling.

Bottom line: Rather than constructing prompts that I forget to use and having to copy and paste between, I’m now building self-improving systems that chain many skills together automatically to create high-quality outputs I can immediately use that are impossible to create in chat.

The Deeper Shifts That Are Changing

It would be easy to read those examples and think: “That’s impressive but niche. It’s about writing articles.” It’s not. The article factory is just one instance of a universal principle.

Swap “article” for whatever your expertise produces:

• A consultant’s strategic frameworks.

• A coach’s diagnostic process.

• A founder’s decision-making methodology.

• An analyst’s pattern-recognition across datasets.

• A researcher’s literature synthesis.

• A designer’s aesthetic judgment.

Every expert has a version of the same bottleneck: deep knowledge that can only produce at the speed of one person sitting in one chair having one conversation at a time.

The specific systems I built are irrelevant. The shift underneath them isn’t.

That shift has a few dimensions worth naming, because they change more than your output. They change how expertise itself works.

#1. Your Expertise Becomes A Compounding Asset

In the Chat Trap, your AI relationship resets every day. Every session starts fresh. The brilliant conversation you had yesterday? Gone. The methodology you refined last week? Re-explain it.

In The Architect’s Paradigm, every session builds on the last. Your voice guide gets refined. Your research base grows. Your workflows get smarter. Your quality standards get more precise.

For example, every time I run a system, I have checkpoints at each step where I give feedback. After I give feedback, I say, “Update the skill based on what I said.” Now, when the system operates in the future, my feedback will already be baked in.

This is fundamentally different from the chat approach where I need to open a prompt, find the correct spot in the prompt, and then edit it in order to improve the system. That’s enough friction to cause me to improve my prompts less.

This is the difference between simple interest and compound interest. In the Chat Trap, you earn interest every day but it’s withdrawn every night. In the Architect’s Paradigm, interest earns interest. Six months from now, you don’t just have “experience with AI.” You have an intellectual infrastructure. A system that encodes decades of your expertise into operational tools that work alongside you.

Peter Drucker once wrote:

The most important contribution of management in the 20th century was the fifty-fold increase in the productivity of the manual worker.

Then he added:

The most important contribution management needs to make in the 21st century is similarly to increase the productivity of knowledge work and knowledge workers.

Think about what Drucker is really saying here. Manual workers got a 50x productivity gain. Not from working harder, but from systems. Assembly lines, standard operating procedures, quality control processes. The knowledge equivalent of those systems has never existed.

Until now.

For the first time, a knowledge worker can build operational systems around their own expertise. Systems that run, produce, and compound.

#2. You Get To Do the Part That Actually Matters

Let me be honest about something. AI doesn’t write like you. Even with a detailed voice guide, even with your best work as training data, a reader who knows what to look for can tell. The cadence is too even. The word choices are slightly off. The personality is approximated, not embodied.

The Architect’s Paradigm doesn’t solve that. What it solves is the production bottleneck that kept you from doing the part only you can do.

Before, writing a deep article meant weeks of work: researching across domains, organizing sources, building structure, writing a first draft, editing, quality-checking. Most of that work isn’t where your voice lives. Your voice lives in the last 10%. The specific analogy you choose. The sentence you cut because it’s trying too hard. The moment where you break from the structure because the idea demands it. The aside that only someone with your exact experience would think to include.

The system handles the first 90%. Research, structure, a working draft that’s directionally right but not yours yet. That’s the part that used to eat weeks. Now it takes hours.

Which means you actually have time to do the 10% that matters. The part where the work becomes unmistakably yours. Before, that 10% often got rushed or skipped entirely, because you’d already spent so long on the production layer that you were tired, behind schedule, or both.

I’ve published more of my best work in the last three months than in the previous year. Not because the AI writes like me. Because I finally have time to write like me.

#3. You Own Everything You Build

Last year, everyone was saying you could run your business with Make.com and n8n. This year, they’re saying the same thing about the next hot tool. There’s always a new platform promising to change everything.

So why is this different?

Because Make.com and n8n were integration layers plumbing between other people’s software. When the tools changed, the plumbing broke. You were building on someone else’s foundation.

Everything you build in the Architect’s Paradigm lives as plain text files on your own computer. Your methodology. Your standards. Your workflows. Stored as markdown files you own. If the tool disappeared tomorrow, you’d still have every skill, every template, every piece of intellectual infrastructure you created. You could easily use another AI model with your data.

Even if a better tool comes along next year (and one probably will), what do you think it’s going to need from you?

Your methodology, clearly articulated. Your standards, written down. Your workflows, structured so an AI can execute them.

That’s exactly what you build in this paradigm. The people who do this work now won’t start over. They’ll be the ones who adopt the next tool in an afternoon, because they’ve already done the hard part: making their expertise explicit, structured, and operational.

Everyone else will still be starting from scratch. Again.

Take Action

The factory owners who reorganized their floors around distributed motors didn’t just get more productive factories. They got factories that could do things centralized power couldn’t do at all. New products. New processes. New business models that were literally impossible under the old architecture.

The same is true here. The Architect’s Paradigm doesn’t just make your current work faster. It makes work possible that was impossible before. The kind of deep, cross-domain, systematized intellectual production that no individual could sustain alone in AI chat.

The technology is here. The architecture is ready. The only question is whether you’ll keep rearranging machines around the central shaft, or redesign the factory.

Get My Thought Leadership System Personally Installed At Your Company By Me

For the first time ever, I’m taking on a handful of one-on-one pioneering clients who are entrepreneurs and senior executives at $1M+ companies who would benefit from being one of the first people in the world to have a blockbuster thought leadership system installed for themselves, their employees, and/or their company. The result of this system is the ability to consistently publish blockbuster content across many channels in your unique voice.

If you’re interested, reply with BLOCKBUSTER in the comments or to this email, and I’ll send you more details.

APPENDIX: Responses To The Top Four Objections To The Agentic Paradigm

No big transition is clean. And pretending this solves everything would insult your intelligence. Let me address the objections I hear most often, because the best ones are worth taking seriously.

#1. “I don’t have time to learn a new tool.”

This one answers itself once you see the math clearly.

Count the hours you spend each week on overhead: re-explaining context, copy-pasting between windows, shuttling files, resolving contradictions between parallel chats, reformatting output. For most serious AI users, this is 10-15 hours per week. That overhead doesn’t shrink as you get better at Chat. It grows.

The Architect’s Paradigm doesn’t add to your workload. It eliminates the overhead that Chat created. The question isn’t whether you have time to learn to work with agentic AI. It’s how much longer you want to keep paying the Conversation Tax.

#2. “What if AI gets good enough that I won’t need systems?”

The work of making your expertise explicit and structured is the work, regardless of which tool executes it. If a future AI can read your structured methodology and run it automatically, you’ll be the person it works for. If you never structured your methodology, you’ll be the person still explaining it from scratch.

The most AI-ready thing you can do today isn’t learn a specific tool. It’s make your expertise legible to machines. That’s what this paradigm demands. And it’s what every future paradigm will demand too.

#3. “I love what I do. I’m not sure I want to become an AI architect.”

This is the most important objection because it gets at the thing people are actually afraid of. That becoming “a systems person” means becoming less of a craftsperson. That the creative work you love will become colder, more mechanical, less you.

I want to make three points that changed my mind on this.

Point #1: Even the work you love is mostly work you don’t love.

Break down any knowledge work task into its actual components and you’ll find something suprising. Most of the hours, even in work you genuinely love, are spent doing things that you don’t.

Take reading a book. Over the years I’ve read 1,000+ books, because I love reading. But, here’s a harsh truth. If you’re reading for learning, most books contain maybe 5 to 10 genuinely useful ideas for you specifically. Maybe less. Yet you spend hours searching through every page to find them. The joy is the insights. The work is the search. And most of your reading time is the search, not the insights.

Take writing an article. The ideas arrive in flashes. The polish feels like craft. But in between, you’re spending hours formatting footnotes, reorganizing sources, finding that one quote you remember but can’t locate, rewording transitions, rechecking facts. Most of your writing time is production, not creation.

Take running a coaching practice. You love the sessions. You love the breakthroughs. You tolerate the scheduling, the invoicing, the follow-up emails, the content marketing you know you should be doing but avoid. Most of your coaching time is not coaching.

When you build systems around your work, you take yourself out of the parts you never liked in the first place. The parts you tolerated because they were the price of admission to the parts you loved. The system doesn’t replace your craft. It removes the tax you’ve been paying on your craft.

This is something Boris Cherny, who leads Claude Code at Anthropic, put beautifully on Lenny Rachitsky’s podcast:

I have never enjoyed coding as much as I do today because I don’t have to deal with all the minutia.
Boris Cherny, Creator Of Claude Code

Think about what Boris is really saying. He’s not saying he codes less. He’s saying he enjoys coding more. Because the thing he loved about coding was never the minutia. It was the building. And now he gets to do more of that and less of the other.

Point #2: You’re probably wrong about how this will feel.

Here’s the thing that took me a long time to accept. We’re terrible at predicting our own emotional reactions to future experiences.

The Harvard psychologist Daniel Gilbert has spent decades studying this phenomenon. He calls it affective forecasting. And his core finding, documented across dozens of studies and summarized in his book Stumbling on Happiness, is that humans are systematically bad at predicting how future events will make us feel.

We overestimate how happy a promotion will make us. We overestimate how devastated we’ll be by a breakup. We overestimate how miserable we’d be if we had to move, change jobs, start over. We predict our future emotional states using our current ones, and we’re usually wrong.

What this means for this objection is simple. When you predict “I’d hate becoming a systems person,” you’re not accurately forecasting your future experience. You’re projecting your current feelings about a version of the thing you’ve never actually done.

I went through this exactly. And, now I love writing more than I did before. Because now I spend my time on the parts I always loved, not the parts I tolerated.

Point #3: The best creative work has always run on systems.

Pixar has the Braintrust process that pressure-tests every film through structured critique. Motown had Berry Gordy’s quality control meetings where songs had to pass a committee vote before release. Every great chef has mise en place. Ingredients prepped, tools positioned, workflow mapped before the first flame.

These systems don’t replace the creative work. They protect it by handling everything that ISN’T creative.

Building a system doesn’t mean you stop doing the work you love. It means the work you love gets more of your time, not less. The system handles the parts you tolerate. You keep the parts you chose this career for.

If Gilbert’s research is right (and it has been replicated many times), you’re probably going to end up enjoying it more than you think you will.

#4. “Can’t I just wait until this is easier?”

You can. But understand what you’re giving up.

This is the compound interest problem. Two investors with the same strategy, the same returns, but one starts five years earlier. The early investor doesn’t have a 5-year head start. They have a compounding head start. The gap between those two investors widens every year.

Every week you spend in the Architect’s Paradigm, your systems get smarter. Your intellectual infrastructure grows. Your production capacity compounds.

But every week you spend in the Chat Trap, you start from scratch.

In three years, these will be fundamentally different kinds of professionals. Not because one is smarter. Because one built infrastructure that compounds and the other kept renting a tool that resets every morning.

“Why are these people making the world so hard for me to live in? Everything worked fine before.”

My mom said this to me last year, and it sent a dagger into my heart. She is 74, retired, and was, I want to emphasize this, a computer engineer.

My mom is still the most independent-hearted person I’ve ever met, and I immediately knew what she was really saying. The shift was causing her to lose what she valued most: her independence.

The same pattern kept repeating:

• Her old device would break (phone, TV)

• She would get the new version, which was “smart” by default

• The setup and usage would be overwhelming

• She’d spend hours trying to figure out something that would take someone else minutes.

• Until eventually, she’d either give up or get help.

With each new device that went “smart” and each offline process that went online, her independence eroded.

She did not see this coming. Almost nobody who gets left behind ever does. But the world was becoming more and more alien to her, and it felt like there was nothing she could do about it.

I’m writing this article because the same thing is about to happen again, on a drastically faster timeline, to a much larger group of people. And I think there’s a real chance that you’re one of them.

In fact, a huge percentage of people are already wondering the same thing about AI that my mom wondered about technology:

Why is this small group of people in Silicon Valley creating something that will completely disrupt my life, my plans for the future, my local community (in the case of data centers), and my decades of expertise that I’ve gone into debt for?

The resistance isn’t coming from where you’d expect either. This time it’s not just retirees struggling with new interfaces. Graduating seniors are booing commencement speakers for telling them to embrace AI. The people who would normally be most excited about the future are the angriest about it:

I resonate with the backlash.

Within two decades, AI may be orders of magnitude smarter, faster, and cheaper than any human worker, and there will be far more of it than there are of us. The downside scenarios are real.

But none of us get to opt out of the world we live in.

The knowledge workers who don’t embrace AI will be left behind, and no one is coming to bail them out. Those who embrace it will see their productivity shoot up to previously unfathomable levels.

But anger and excitement have one thing in common: neither one tells you what to do next…

The Most Dangerous Career Advice Right Now Is “Figure Out AI”

Everyone agrees AI is transforming the world. Almost nobody agrees on what that actually means for anyone’s career.

A clear, hopeful future has been replaced by fog. Now it’s hard to know whether AI will take all of our jobs in five years or just keep making us more productive and creative for the foreseeable future.

Therefore, it’s hard to know exactly what to do now. It’s hard to know which AI skills will pay off for years, and which will be obsolete by the time you finish learning them.

Many people are falling into one of two camps:

• Opting out of staying on the AI frontier and burying their head in the sand.

• Flailing around trying to do everything. Staying on top of the latest tools, AI models, AI harnesses, prompting techniques, and industry news. Working harder than ever, but not sure if they’re making real progress.

This article is about clarity.

It provides you with the one AI skill and the one category of tools that are virtually guaranteed to deliver the biggest return for knowledge workers who apply them.

This clarity is critical because once you know what won’t change, you know what to invest in now and can be confident it will pay off.

Not only that, based on Harvard research and early results of people who are making the switch (more on this later), I can confidently say that the skill you’ll need to learn is one that you’ll actually enjoy doing.

And if you haven’t started yet, that doesn’t mean you’re behind. You’re in Stage 2 of a 5-stage pattern. The window is still open.

Very few have felt the true magnitude and speed of what’s happening, because we’re all inside it. It’s so ever-present that it’s invisible. To really see it, you have to step outside of it.

This article will help you take that step outside.

The Multi-Century Pattern That Reshaped Civilization Twice Is Running A Third Time

In 1700, saying “I can’t read” carried no stigma. By 1900, it was a serious liability.

In 1990, “I don’t use computers” was a defensible position. By 2015, it ended careers.

Today, “I’m not a programmer” is normal. By 2030, it will sound the way “I can’t read” sounded in 1900.

I’m not predicting this flippantly.

We’re inside the third run of a historical pattern that has already reshaped civilization two and a half times:

1. Once for reading and writing (text literacy)

2. Once for counting and calculating (numerical literacy)

3. Finally, for using and authoring software (software literacy)

Just as reading literacy spread for centuries before writing literacy did, software usage (digital literacy) spread for decades before software authorship became mainstream.

Today, knowledge workers use dozens of software apps on their phones, in their browsers, and on their desktops. Someone who can’t use software is essentially unemployable as a knowledge worker.

Starting in November 2025, when AI agents became able to reliably create working code, we entered the second stage of the third literacy—Software Authorship —in which domain experts turn what they know into running systems using plain English, with AI doing the technical work.

On the surface, Software Authorship doesn’t sound like a civilizational shift on the scale of reading or arithmetic. For 50 years, software has been a niche specialty: built by highly paid engineers, used passively by everyone else. Calling it the next universal literacy feels like a stretch at first.

It isn’t.

On the surface, Software Authorship doesn’t seem likely to have much impact on the average knowledge worker’s day-to-day work and career trajectory.

It will.

On the surface, Software Authorship feels like the type of shift that will take decades to run its course.

It will likely take 5-10 years. Maybe less.

The shift will create a new generation of economic winners and losers:

• Winners: On the one hand, the most advanced AI users creating software will be 100x, then 1,000x, and then 10,000x more productive than the average knowledge worker who lightly uses AI in chat. This is already happening, which I explain later in the article.

• Losers: On the other hand, many people will be left behind. Way more and way faster than in any previous technological shift.

I see the tsunami coming, and 99% of people don’t recognize what’s about to happen. As a lifelong educator, my mission is to help people make this shift as smoothly as possible.

On a personal level, I feel motivated by both the opportunity and the risk:

• Over the past few months, I’ve seen my productivity skyrocket higher than it’s been at any other point in my career.

• I fear being left behind on a visceral level because my mom isn’t the only person I’ve watched be left behind in the past, and it’s brutal…

What Happens To The People Who Sit This One Out

A 2005 comment from a dear mentor still sticks with me:

“I used to be good at computers in the 80s. I shouldn’t have let my skill slip. Don’t make the same mistake I made.”

Earlier in his career, my mentor decided to be more productive by delegating technical tasks to his employees rather than learning them himself. Until one day, he woke up and realized four harsh truths:

1. He was completely dependent on others for basic tasks.

2. He had become the kind of person who got the warm handshake at the front of the room and the knowing look behind his back among employees.

3. He was losing contracts to others because he wasn’t keeping up with the times.

4. He was so far behind that he couldn’t catch up.

I remember one moment in my early 20s when he asked me for help with a very basic tech task. It felt so obvious, I couldn’t help but smirk. He paused, closely examined my face, and then immediately ended the interaction. He never asked me for tech help again.

Looking back, I see his vulnerability in asking for help and his shame at my response. I wish I could’ve responded differently.

He never did catch up.

My mom and my mentor weren’t stupid or lazy. On the contrary, during their careers, they were each ambitious and successful. They just didn’t develop a key universal literacy when they had the chance. And they didn’t realize what they’d lost until it was too late.

It’s the boiling frog problem. The water heats one degree at a time. The frog never feels the moment when it should jump out, until the moment when it can’t. AI is doing the same thing to most knowledge workers. Each new headline is interesting but not alarming. Each week, it still feels okay to start later. The water just gets a little hotter.

Realizing this, I started doing the one thing my mom and my mentor didn’t. In March 2023, I made the decision to focus on studying and writing about AI full-time.

Then last winter, my news feeds blew up…

I Spent Twenty Years Convinced I Wasn’t A Programmer. I Was Wrong.

I saw the most luminary programmers stop writing code all at once:

Given the latest lift in LLM coding capability, like many others I rapidly went from about 80% manual+autocomplete coding and 20% agents in November to 80% agent coding and 20% edits+touchups in December. i.e. I really am mostly programming in English now.
—Andrej Karpathy (former head of AI at Tesla), 40,000 likes

Pretty much 100% of our code is written by Claude Code + Opus 4.5. For me personally it has been 100% for two+ months now, I don’t even make small edits by hand. I shipped 22 PRs yesterday and 27 the day before, each one 100% written by Claude.
—Boris Cherny (head of Claude Code) on behalf of Anthropic’s team, 7,000 likes

programming always sucked. it was a requisite pain for ~everyone who wanted to manipulate computers into doing useful things and im glad it’s over. it’s amazing how quickly I’ve moved on and don’t miss even slightly. im resentful that computers didn’t always work this way. I 100%, I don’t write code anymore.
—roon (prominent OpenAI engineer), 6,800 likes

The era of humans writing code is over. Disturbing for those of us who identify as SWEs [software engineers], but no less true. That’s not to say SWEs don’t have work to do, but writing syntax directly is not it.
—Ryan Dahl (creator of Node.js), 20,000 likes

I also saw headlines like this:

Before, AI would get you 80% of the way there, but you still had to manually fix the last 20% of bugs. Suddenly, people were saying that AI was doing all of their coding.

I knew I should try building software.

I was even excited by the idea.

At the same time, it also filled me with dread because I had already failed before, over and over.

My mom was a computer programmer. She encouraged me to follow that path. So in my teens, I bought the website design books and learned HTML and Adobe Photoshop.

Emboldened by my progress, I bought more advanced programming books and took a computer science course at school. But that’s when I hit a wall. Every time I pushed past the basics, the same cycle started:

• I’d write something

• It wouldn’t work

• I’d spend the next three hours figuring out why

Fix one bug, start building again, hit another.

Most of my time went to debugging. It wasn’t fun.

There was no dramatic moment where I quit. I just gradually stopped trying and made a quiet decision about myself: I’m not a programmer.

After college, I tried the other route to creating software: hiring coders. I found an overseas development team and spent $40,000 over a year, working nights and weekends, building an app that let people track their goals.

I quickly learned what it feels like to depend entirely on someone else to build what’s in your head:

• Because of the 12-hour time zone difference, one miscommunication cost a full day.

• I couldn’t tell whether a fix should take an hour or a week.

• I had no way to tell whether the work was high quality.

It was like bringing your car to a mechanic when you don’t know how a car works. You hand over the keys, you pay the invoice, you hope you weren’t lied to, and you hope your car works when you get it back.

Ultimately, the app failed.

Eventually, I lost interest because the whole process killed everything that made the idea exciting.

By the time AI coding tools arrived, I’d seen the data: virtually every successful software company has a technical founder.

At the same time, I said to myself, “I’m not a programmer. I’m a writer.” I’d made peace with it.

So, when Claude Code launched in 2025, my first reaction was: even if AI writes 95% of the code, I don’t want to spend all my time fixing the other 5%. Even with a shorter learning curve, not worth it.

Then came December 2025, when I saw everyone saying that AI could do 100% of coding. Even though I was interested, I didn’t make time for it. I was busy.

What finally broke through was a friend who sat me down in January and said, “I think you can do this. Let me just show you.”

In one call, he walked me through the basics and suggested a few things to try. For the first time in twenty years, I felt like the “programming door” might not be completely closed.

I tried it. And it wasn’t what I expected.

For my first real project, I decided to create a mental model manual.

I had spent four years creating these manuals by hand for my Mental Model Club. Each one took roughly 50 hours of research, writing, and editing. To start, I downloaded all of the old manuals onto my computer. Then, I asked Claude Code to analyze the structure of each manual. Finally, I asked it to produce a new one in the same structure.

Within a few minutes, I had created a manual on the Second Order Effects mental model. On my very first attempt, it was shockingly close to what took me a month to create manually.

That’s when the lightbulb hit me.

Over the next week, I created 300 more manuals with AI. Same depth as when I did it by hand. But way faster. The numbers didn’t lie:

• Before AI: 192 weeks to create 48 manuals

• After AI: 1 week to create 300 manuals

That’s an astounding 1,200x multiplier. And it had only cost me $50 (on top of my $200/month subscription).

Then I asked Claude Code to build something more complex. And it did. But the things I had to fix were never the code. My time was spent on:

• Deciding what to build

• Planning it out

• Iterating with the AI

• Judging what “done” looked like

For the first time, building software required my expertise, not someone else’s.

And because I wasn’t trapped in debugging hell anymore, something unexpected happened: I was having fun. Not in a forced way. Genuine fun.

I described what I wanted in plain English, and Claude built it. I didn’t write a single line of code. I didn’t read a single line of code.

The systems I’d dreamed about for years, but never had a way to build, were suddenly real. And building them turned out to be the most direct path to everything I’d wanted to do with my work.

After I built one tool, I built another. Then five. Then 20. Then dozens more. In just a few months.

Every one of those tools encodes my expertise in ways that no software company would ever productize, because the knowledge is mine. A 27-step news analysis pipeline built on years of mental models I’ve developed. A writing voice system that captures my exact style. An AI-powered research system with over 12,000 notes searchable by meaning, not just keywords.

None of it required me to be a programmer. It just required my domain expertise and my AI prompting expertise.

After months of spending most of my day programming, I noticed two things that surprised me.

First, I realized I was no longer just a thought leader who happened to program part-time. I was actually a programmer. For example, to generate articles like this one, I spent most of my time developing software to streamline the process. Within a few months, I went from not knowing how to code to identifying as a software engineer. It has been the fastest identity shift I’ve ever gone through in my life.

Second, I realized that I actually love programming now…

The Harvard Research That Explains Why I Was Wrong About How I’d Feel About Programming

There’s a Harvard psychologist named Daniel Gilbert who studies something called affective forecasting: our ability to predict how we’ll feel about experiences we haven’t had yet. His finding, across decades of research, is that we’re terrible at it.

We consistently overestimate how much we’ll hate many things we’ve never tried.

Gilbert’s lab has shown this across romantic breakups, tenure denials, election losses, and dozens of other events people are sure they’ll never recover from. They almost always recover faster than they predicted.

This TED Talk clip summarizes the research:

And, according to a follow-up study, the single best predictor of how you’ll actually feel?

Asking people who’ve already done it.

In the study, Gilbert and his collaborators asked undergraduates to predict how much they would enjoy a 5-minute speed date and a peer evaluation.

• One group got detailed information about the event itself.

• The other group got just one stranger’s reaction to the same experience.

The strangers’ reactions won.

People who relied on a single secondhand report predicted their own feelings more accurately than people who studied the situation in detail and then predicted how they would feel.

And the kicker: when participants were given the choice, they preferred the detailed information. They actively rejected the strategy that worked.

This research is relevant right now because most of the people who move to coding with AI actually enjoy it. Boris Cherny, the creator of Claude Code, says that this is roughly what he sees at Anthropic among people who make the shift. And Anthropic is at the leading edge of this wave.

Furthermore, Lenny Ratchitsky, who interviewed Cherny, found something similar when he did three polls on X that collectively got 1,500+ responses:

Below are the specific poll results:

If Gilbert’s research holds and the trend continues, most people who switch to AI programming will enjoy it.

So the emotional barrier is probably lower than you think.

But there’s a second barrier most people haven’t questioned yet: the assumption that building software requires a different kind of thinking than you’re already using.

It doesn’t.

Your Job Is Already Software. You Just Don’t See It Yet.

Let me show you what I mean.

Strip away the job titles, and every knowledge worker is doing the same three-step loop all day:

• Input. Take in information.

• Transformation. Make sense of it, process it, create something.

• Output. Export the result.

For example:

• A lawyer takes in the details of the case, drafts the argument, and files the brief.

• A marketer takes in funnel data, develops the angle, and ships the campaign.

• An accountant records transactions, prepares the reconciliation, and sends the report.

• A designer takes in references, develops the direction, and ships the layout.

Every one of those workflows IS fundamentally like software. Information in. Transformation in the middle. Information out. The shape of every knowledge worker’s job is the shape of a software workflow.

If you’re a consultant or a coach or a strategist and you just felt your jaw tighten at the idea that your job is “fundamentally like software,” I get it. I had the same reaction. My work felt too human, too intuitive, too judgment-dependent to be described that way.

But only the shape of the work is software. The soul of the work is domain expertise.

Looking at the full sweep of knowledge work, four distinct eras emerge, each one inverting the relationship between human and software a little further:

1. Era 1: Human With Mechanical Tools (before ~1980). Work happens entirely in the human’s head and hands with mechanical tools. Paper, pens, ledgers, typewriters. Software doesn’t exist as a workplace tool.

2. Era 2: Human With Software (~1980 to present). The human is the agent. Software is the tool. The human does the work, and the software helps along the way. The lawyer types in Word. The accountant works in Excel. The marketer logs into HubSpot. (This is where most of us still are. We’ve spent our entire careers getting very, very good at being the human in “Human With Software.”)

3. Era 3: Software With Human (2024 to present). The inversion happens. Software does the work. The human directs and judges. The founder doesn’t write the cold outreach. She designs a system that writes thousands of personalized messages while she sleeps. She isn’t using software the way her predecessors used Outlook. The software is doing the work. She’s directing it.

4. Era 4: Software-Only (~2030+). Software does the work autonomously without human direction in real time. Already true in narrow domains: algorithmic trading, automated support for routine cases, dynamic pricing engines, ad bidding. Likely to expand as Era 3 systems mature.

We are standing at the line between Era 2 and Era 3 right now. The professionals who have already crossed it are building 20x leverage. For example, serial entrepreneur Garry Tan is literally coding 400x faster than before AI:

Source: Y Combinator Podcast

Although Tan is an outlier because he’s an early adopter and world-class software engineer, he’s not alone. Legendary entrepreneur and investor Marc Andreessen reports that programmers in his portfolio company are 20x more productive with AI. This experience also aligns with my daily experience and that of friends who are all in on the tools.

I now believe that, in a year or two, people still operating with an Era 2 mindset (humans with software) will watch their work get done around them by people with a fraction of their experience.

The skill you need to move from Era 2 to Era 3 (software with humans) is Software Authorship: the ability to turn what you know into running software, using English as the interface and AI as the implementation. Not writing code. Just describing what you need, precisely enough that AI can code for you.

Boris Cherny is the head of Claude Code at Anthropic, the tool I use to build my own software. He stopped writing code entirely in November 2025 after AI became good enough to write it for him. But the most important thing he’s said isn’t about coding is about who builds the best software:

“The best person to write accounting software, I think maybe even today, is not an engineer. It’s a really good accountant because they know the domain really well. And coding is the easy part. It’s knowing the domain that’s the hard part.”

The technical part is now the easy part. The knowledge you’ve spent your career developing is the hard part. And “hard” here means “valuable.” It means “irreplaceable.” It means that the person who knows a domain most deeply builds the best tools, because the tools are made of domain knowledge now, not code.

This is why I’m saying that almost every knowledge job has software-shaped holes in it that better software would fill:

• The reports your team runs that take hours to compile.

• The data hand-offs between systems your IT department keeps promising to fix.

• The dashboards you wish you had.

• The custom tools that would make your job 40% easier, if only somebody would build them.

The only reason these tools don’t exist is because the supply of programmers has always been tiny relative to the demand for software.

That supply constraint just broke. Because now it’s possible for anyone to build.

The Five-Stage Pattern That Has Reshaped Civilization Twice Is Running Again

Everyone is asking, “Will AI replace me?”

That’s the wrong question for this moment.

The real question is bigger and older: what happens when a skill that only specialists have becomes something everyone can do?

That question has been answered exactly twice in human history. Both times, the answer unfolded as a five-stage pattern that reshaped economies, professions, and daily life. I call that pattern the Literacy Arc.

My mom and my mentor didn’t see the Literacy Arc until it was too late. The rest of this article lays out the full pattern so you can see where you are right now and respond better: the five stages, the two forces driving them, the historical precedent, and the specific window that will be open for the next few years.

READ THE REST OF THE ARTICLE

Paid subscribers get:

#1. The full article

The article is roughly 12,500+ words and includes 10+ videos and charts to make the ideas more concrete and visceral.

#2. A multimedia version

This includes slides and an audio podcast you can listen to while walking or doing chores. Scroll to the bottom of this post to access them.

#3. Everything else that comes with the paid membership to this newsletter:

• Blockbuster Live. This is a monthly 90-minute masterclass where I share my #1 AI lesson learned for the month.

• Augmented Awakening. This is a monthly 90-minute session with world-class coach Anand Rao to help you use AI to augment your personal growth.

• Weekly Blockbuster Article (often with a prompt). I share long-form articles about the deepest, most important AI mental models and paradigms that I see no one else writing about. These articles typically come with slides and an audio podcast.

• $2,000+ In Other Perks. This includes books, courses, prompts, mental model manuals, and much more…

To summarize, you get two live classes per month, $2,000 in perks, and a weekly blockbuster article for just $20/month or $150/year.

Subscribe now

Now let me show you the full pattern…

[insert paywall here?]

The Historical Pattern Behind What Cherny Described

I shared Cherny’s insight earlier: coding is the easy part, and the best person to build accounting software is a great accountant, not an engineer. But before he said that, he laid out a historical analogy that puts the whole shift in perspective:

“Before the printing press, essentially 10% of the European population was literate. They knew how to read and write. They were often employed by kings and lords that were not literate. And their job was to read and write. This is not something that everyone knew how to do.

The printing press was invented. And in the 50 years after the first printing press, there was more literature published in Europe than in the 1,000 years before. And over the same period, the cost of a book went down like 100x.

And then it took a couple hundred years because learning to read and write is hard. You need education systems and government and everyone can’t be working on farms and so on. But over the next few hundred years, literacy went up.

And so now we can all read and write and you don’t need a degree in reading and writing to know how to read and write. Although still there are professional writers and that is the thing that you can do.

So I think the thing that’s about to happen, and it’s going to be much faster than 50 years, is software will be a thing that is fully democratized that anyone can do.”

When I first heard Cherny lay this out, I felt two things at once:

1. Relief that someone had finally named the pattern I’d been sensing.

2. A chill at how fast he said the timeline would compress.

Both feelings turned out to be warranted…

There Is A Pattern Behind Every Universal Literacy. It Has 5 Stages.

What Cherny described follows the Literacy Arc that I introduced above. It evolves across five stages:

1. Stage 1: The Spark

2. Stage 2: The Premium

3. Stage 3: The Inversion

4. Stage 4: The Lock-Out

5. Stage 5: The Forgetting

Stage 1: The Spark

A technology arrives that makes a once-rare skill possible at scale:

• Text literacy: Gutenberg’s press in the 1450s.

• Numerical literacy: The switch from Roman numerals to the modern 0-9 digits, plus the first cheap printed math books that appeared in the late 1400s.

• Software literacy:

  • Use half: The personal computer in the 1980s.

  • Authorship half: Late 2025, when AI models crossed the threshold of “good enough” code generation.

The Spark is mostly felt as a curiosity. The technology exists. The skill is possible. Almost no one has it yet.

Stage 2: The Premium

A specialist class learns the skill and earns a premium for it. Non-adopters experience no penalty.

• Text literacy: Roughly 1450 to 1800. Notaries, secretaries, and clerks commanded a premium. Most workers were illiterate without consequence.

• Numerical literacy: Roughly 1500 to 1800. Merchants and bookkeepers commanded a premium. Most workers traded in barter or memorized prices.

• Software literacy:

  • Use half: Roughly 1980 to 2005. IT staff commanded a premium. Senior executives had their assistants handle email for a surprisingly long time.

  • Authorship half: Roughly 1980 to the present. Programmers still command a hefty premium compared to other knowledge work professions.

Stage 2 is the long, comfortable middle.

This is where my mentor was for most of his career: successful, respected, delegating the technical work to others.

This is also the stage where most non-adopters tell themselves one of three stories:

• They will learn the skill “later.”

• They can delegate it forever.

• The skill isn’t as important as others say it is.

Stage 3: The Inversion

The premium collapses as the skill spreads. At the same time, daily life starts requiring the skill for the first time.

• Text literacy: Roughly 1800 to 1850. Industrial economies started requiring written contracts, posted laws, and printed job descriptions.

• Numerical literacy: Roughly 1800 to 1900. Standardized currencies and consumer pricing made counting required for ordinary commerce.

• Software literacy:

  • Use half: Roughly 2005 to 2015. Banking, taxes, and job

    applications all moved online. “I don’t use computers” went from a

    defensible position to a quiet cost.

  • Authorship half: Began in November 2025 when Anthropic released Opus 4.5, the first coding model that could reliably produce working code.

This is the dangerous stage.

The non-adopter feels the first friction without yet recognizing what it is. Forms they can’t fill out. Conversations they can’t follow. The looks from their juniors. Contracts that don’t come.

Later in his career, my mentor had entered Stage 3 of the Literacy Arc. He could feel the change in the room. He did not yet have a name for it.

If you’re feeling a flash of recognition right now, that’s good. You still have time to act on it. The people who get locked out are the ones who feel the friction but explain it away.

Stage 4: The Lock-Out

The skill becomes a universal expectation. Non-adopters are structurally excluded.

• Text literacy: Reached this stage by roughly 1850 in Northern Europe

  and the United States. The illiterate could not sign contracts, read

  posted laws, or hold most paying jobs.

• Numerical literacy: Reached it by roughly 1900. The innumerate could not price goods, manage accounts, or hold professional positions.

• Software literacy:

  • Use half: Reached it by 2015. The senior professional who never

    adapted could not bank online, file taxes, or apply for jobs.

  • Authorship half: Hasn’t arrived yet, but is approaching faster than any previous literacy arc.

The Lock-Out is no longer a missed premium. It is a permanent closing of the door. When digital literacy was in The Premium, the IT desk walked you through using email. By the Lock-Out, the same desk rolled its eyes when you asked how to share your screen.

If that’s uncomfortable to read, good. It means you’re paying attention. The people who got locked out in the previous rounds weren’t careless. They were competent people who didn’t recognize the transition while it was still possible to act. The fact that you’re here, reading this, means you’re not them. Not yet.

Stage 5: The Forgetting

The literacy becomes invisible. Lacking it stops being a circumstance and becomes an identity. Someone who says, ‘I don’t know how to use Zoom,’ isn’t telling you about their education anymore. They’re telling you something about themselves.

This is what happened to my mom. She didn’t lose a skill. She lost her place in the world. And by the time she could name what had happened, the world had moved on without her.

Reading is in The Forgetting now. Arithmetic mostly is. Digital literacy will arrive there around 2030. Software authorship will follow soon after.

Where You Are In The Pattern Determines What Happens Next

By understanding each of the five stages and what it feels like to be inside them, we can better sense-make and respond:

The five stages also help us understand the importance of acting sooner rather than later:

Seeing this moment as part of a cycle that repeats itself throughout history, rather than as a completely novel event, allows us to better contextualize the rapidly changing, alien world of AI.

Unfortunately, the Literacy Arc is invisible and counterintuitive for most people because the previous instances happened before we were born.

Fortunately, by studying analogous historical cycles, we can begin to build the same intuition we’d have built if we lived through them.

Famous investor Ray Dalio captures how studying these cycles helps us better invest our time and money in the following quote:

The times ahead will be radically different from those we’ve experienced in our lifetimes, though similar to many times in history.

[…]

To anticipate and handle situations that I had never faced before I needed to study as many analogous historical cases as possible to understand the mechanics of how they transpired. That gave me principles for dealing with them well.

Dalio’s approach is particularly important now.

Normally, a major cycle would give us decades to track it, get used to it, and see its consequences unfold. But AI is coming in like a “supersonic tsunami,” and our institutions and culture haven’t adapted yet:

• Institutions: We can’t major in AI at school. Most companies have limited AI training, if anything.

• Culture: Our peers, bosses, and colleagues are often just as confused.

We’re left to learn on our own. And the consequences of how we choose to educate ourselves are very real. There is no one coming to save us if we adopt too slowly.

To help us handle the AI software authorship “situation,” I’m going to take a page out of Dalio’s playbook and take us back to the 1400s to better understand the explosion of reading literacy…

First, the production of the artifacts (books, newspapers, magazines) had to become cheap

Every Literacy Arc runs on two forces:

• Force #1 - Cheap Supply: a technology collapses the cost of producing the relevant artifact (books, numerical documents, software).

• Force #2 - Universal Demand: daily life becomes saturated with that artifact until the skill goes from useful to necessary.

Both forces have to be present. When they combine, the Literacy Arc advances through all five stages. When only one is present, it stalls.

I’m going to walk through both forces in detail using reading literacy, because once you see how this pattern played out across 400 years, you’ll recognize the same pattern compressing over the next five.

Before Gutenberg, a single undecorated Bible took a scribe over a year to produce and required around 200 calfskins. Monasteries kept their own herds because buying skins on the open market would have bankrupted most scriptoria.

As a result, one Bible could cost over $100,000 in today’s dollars. Only churches and kings could afford a library.

Fortunately for us, book costs collapsed in three waves:

• Wave #1 (Paper): Paper mills spread across Europe in the 13th-15th centuries, replacing animal-skin parchment.

• Wave #2 (Printing Press). Gutenberg’s press arrived in the 1450s, lowering the cost from nobles-only to merchant-level affordability.

• Wave #3 (Industrial Revolution). The Industrial Revolution brought steam presses and wood-pulp paper, making books cheap enough for everyone.

Today, you can get a hardcover bible for less than $10 with the literal click of a button. Universal literacy followed…

Second, the demand for reading had to develop

Cheap supply by itself never produces universal literacy. Even as the cost collapsed across three waves, it took roughly 400 years from the first wave for universal reading/writing literacy to arrive. The reason is that universal demand also had to develop.

Demand developed because each cost collapse didn’t just produce more books. It also produced an entire new universe of printed artifacts that hadn’t existed before:

• Newspapers became viable.

• Printed legal forms standardized contracts.

• Shop signs and product labels became cheap to produce in volume.

• Posted notices, almanacs, broadsides, printed schedules, and receipts entered daily life.

Each new artifact created new daily situations where reading was useful… and eventually, necessary. The daily situations created the demand. And the demand did the teaching:

• Parents taught children at home.

• Apprentices learned from masters.

• People taught themselves from cheap printed primers.

Bottom Line: Low Price + High Demand = Universal Literacy

The low-demand skills that the new technology replaced, like calligraphy, became decorative arts. The cases where demand was high but production remained expensive (medical care, for most of history) led to professions being reserved for elites. Skills that become low-priced and high-demand became universal.

Arithmetic became universal the same way and in the same order.

• Commerce (enabled by the printing press) made numerical artifacts cheap: standardized currencies, printed account books, price tags.

• Trade created the demand and the literacy education. Merchants learned arithmetic through apprenticeships and self-study. The first printed math textbook in the West, the 1478 Treviso Arithmetic, was written in plain Venetian for traders to teach themselves, not for classroom use. People were numerate centuries before schools taught arithmetic.

Universal literacies start as consumption-focused (reading) and then become creation-focused (writing).

And the same is happening with Software Authorship right now:

• Digital literacy: personal computers turned everyone into a software user.

• Software authorship: AI coding tools are turning everyone into software creators.

Understanding this historical pattern of how literacies evolve gives me great conviction that software authorship will become a universal literacy.

To see how this just became possible, look at how both forces are interacting today:

• Force #1: The cost of coding just collapsed

• Force #2: The latent demand for software was released

Force #1. The Cost Of Coding Just Collapsed

While personal computers made it drastically cheaper to create software, coding was still expensive given that there are so few programmers, and learning to code takes lots of time and intelligence.

For 30 years, custom software cost between $10,000 and $100,000 just to start building. That price tag meant only large organizations could afford it. Solo professionals, small teams, and individuals had to live with whatever generic software they could buy.

In late 2025, that cost dropped by 97% with vibe coding tools like Replit, Bolt, and Lovable. As these tools became easier and more reliable, the number of non-coders who were coding suddenly skyrocketed:

• Lovable went from 2.3 million users to nearly 8 million in 2025, with over 100,000 new projects built on the platform every day.

• Replit has reached 50 million users.

• Claude Code, the tool Cherny runs, became the first AI coding product in history to reach $1 billion in annualized revenue. By early 2026, it was reportedly on track to make $2.5 billion in a year.

• Within months of launching, so many non-coders were using Claude Code for general knowledge work that Anthropic launched a separate product in January 2026: Claude Cowork.

The speed of adoption tells its own story. Most of the world’s software is stored on a platform called GitHub. Every time a programmer saves a new change, it gets recorded as a “commit.”

• In September 2025, less than 1% of GitHub commits were being authored by Claude Code.

• By February 2026, the number was 4%.

• By May 2026, some estimates suggest the number exceeds 10%. And that’s just for one AI tool. Add in all the others, and the share of code being written by AI is dramatically higher.

Bottom line:

It is cheap, fast, and easy for people to create software now.

Force #2. The Latent Demand For Software Was Released

Remember those software-shaped holes in your workday? The reports, the hand-offs, the dashboards, the custom tools nobody ever builds? Those holes been there for decades. And there’s a reason nobody built software to fill them.

The programmers we did have were busy charging a lot and building products that scale to millions of users, not single-purpose tools for an auditor in Boston.

There’s no point in asking for software that nobody is going to build. So the demand for custom tools sat as latent friction inside every knowledge worker’s day, dismissed as “that’s just how things work.”

In 2025, the supply constraint broke.

Every knowledge worker is about to discover that their job was mostly software-shaped all along, and that the supply to fill that shape will be domain experts augmented by AI coding.

A reasonable objection at this point: if everyone can create software, won’t the advantage disappear as fast as it arrived?

Not necessarily. Every time the cost of producing software has dropped before, the demand for software grew faster than the supply.

Marc Andreessen lived through the last episode of this. In the late 1980s, “expert systems” arrived with predictions that programmers would soon be obsolete. On The Lex Fridman Podcast, Andreessen described what actually happened then, and what he expects from this round:

“Of anything in industrial society, code has the highest elasticity, which is to say, the easier it is to make it, the more it gets made. Effectively, there’s unlimited demand for code. In other words, there’s always some other idea for a thing that you can do, a feature that you can add, or a thing that you can optimize.

Overwhelmingly, the amount of code that exists in the world is a fraction of even the ideas we have today, and then we come up with new ideas all the time.

In the late ‘80s, early ‘90s when sort of automated coding systems started to come out. Expert systems were a big deal in those days. And there was a famous book called ‘The Decline and Fall of the American Programmer,’ that predicted that these new coding systems were gonna mean we wouldn’t have programmers in the future. And of course, the number of programming jobs exploded by like a factor of 100.

My guess is we’ll have more coding jobs probably by like an order of magnitude 10 years from now that will be different. They’ll involve orchestrating AI. We will be creating so much more software that the whole industry will just explode in size.”
— Marc Andreessen, Co-Founder, Netscape and a16z

Bottom line:

Force #2 has been active for a long time now. The demand was already there. It just had no supply to answer it. Now it does.

As a result of both forces, we’re now just beginning to see a new class of knowledge economy winners—domain experts with no coding background…

The New Winners Of The Knowledge Economy

Fast forward to February 2026.

The first-place winner of Anthropic’s coding hackathon didn’t know how to code. That was Michael Brown, a personal injury and traffic lawyer, who said afterward:

“It’s crazy to me that I ended up winning this contest, and I didn’t write a single line of code. I didn’t even read a line of code.”

What’s even crazier was who he was competing against. 13,000 people applied. 500 were accepted. 498 of those accepted were professional software developers with years of experience shipping products.

Brown built an AI-powered permit assistant for California ADUs, the backyard cottages and converted garages that the state has been pushing as a housing solution. As a result of his domain experience, he knew why these permit applications get rejected on first submission more than 90% of the time.

At the hackathon, he built a solution that would fix that. He spent six days explaining to Claude Code, in plain English:

• How permit law worked

• Which sentences or phrases a clerk would flag as problematic

• Which language the office expected to see

• What a successful application looked like.

He didn’t win first place because he was a better coder than the software developers he was competing against.

Brown won because he understood exactly what his app needed to do and what result it needed to produce, and he could describe it in enough detail that Claude Code could translate his specifications into code that worked.

Think about that. That’s like a behavioral psychologist who’s never played a hand of poker walking into the World Series of Poker Main Event and taking the bracelet.

Lovely Mcinerney saw the same thing from her accounting desk. She has Big 4 audit experience and runs quarterly closes for an investment fund, reconciling every asset, every depreciation schedule, and every dollar the fund has touched. For years, the knowledge she needed to do fixed-asset work either lived in her head or was buried in spreadsheets only she could follow.

She started encoding that knowledge into plain-English files and handing them to Claude Code. Her summary of what’s actually hard:

“The folder structure is easy. The hard part is deciding what goes inside.”

That sentence is the entire thesis of this article in one line. The technical part is now easy. Your domain knowledge is what matters.

Daniel Roth has been a journalist his entire career. He runs a 400-person editorial team at LinkedIn and has never learned to code. In the past year, he shipped Audio2, an app which turns podcast moments into shareable video clips, and Commutely, a real-time subway tracker for New York commuters. Both built by a career editor who can’t read the code inside them.

Bottom line:

Brown, Mcinerney, and Roth aren’t alone. There are now millions of people like them.

The production constraint is mostly gone now, and the progress toward making AI coding more reliable, secure, and scalable is happening faster than any other area of AI knowledge work.

The transition from a few million people to the broader knowledge economy may happen much more quickly than you think…

What Took Reading 400 Years Will Take Software Authorship 5 Years

Text literacy took roughly 400 years from Gutenberg to universal schooling.

Arithmetic took about 400 years to move from merchant self-study to standard curriculum.

Software authorship will take fewer than ten.

I’ve felt this compression myself. The gap between the AI tools I started with in 2023 and what I’m using now is so large, it’s hard to describe to someone who wasn’t there for both.

On SWE-bench, the standard benchmark for real programming tasks, top AI systems went from solving roughly 2% of problems at the benchmark’s launch in October 2023 to nearly 90% on the verified subset today. The benchmark saturated so completely that OpenAI stopped reporting scores on it in 2026 and researchers had to release a harder replacement, SWE-bench Pro, where top models currently solve fewer than half of the problems.

Three reasons explain why:

• AI companies understand coding

• AI is uniquely good at code

• AI that codes better can be used to build better AI

Reason #1: AI companies understand coding

The primary role inside AI research labs is coders. This means that all of the research labs deeply understand the challenges that coders face and how to alleviate them. Dario Amodei, co-founder and CEO of Anthropic, makes this point on The Lex Fridman Podcast:

“Programming is a skill that’s very close to the actual building of the AI. So the farther a skill is from the people who are building the AI, the longer it’s going to take to get disrupted by the AI. But programming is the bread and butter of a large fraction of the employees who work at Anthropic and at the other companies.”
— Dario Amodei, Co-Founder and CEO of Anthropic (2025)

In the following interview on The Dwarkesh Podcast, Amodei gives a concrete example of how this helped them:

Around the beginning of 2025, he told his engineers to start using Claude to accelerate their own research. The tool they built for internal use, originally called Claude CLI, saw such fast adoption that they launched it externally as Claude Code.

The engineers who build Claude now use Claude Code to write the code that makes Claude better. And when Claude gets better, Claude Code gets better, so the engineers can build even faster.

Reason #2: AI Is Uniquely Good At Code

Coding sits at the extreme easy-to-verify end of the problem spectrum. Noam Brown, who leads multi-agent research at OpenAI, has a framework for understanding which problems AI will crack fastest.

Problems have two components: generation and verification.

• Certain problems are hard or subjective to verify (which poem is better?)

• Other problems are easy to verify (is this math solution correct?)

When a problem is easy to verify, AI progress moves incredibly quickly, because the AI can iterate to the correct answer. Brown explains:

“You might call it a generator-verifier gap where it’s really hard to generate a correct solution, but it’s much easier to recognize when you have one.”
— Noam Brown, OpenAI, 2025

Writing software is hard. But checking whether software works is easy: you run it. It either does what you asked or it doesn’t. The feedback is instant, objective, and automatic. And AI is capable of doing the verification itself, generating millions of attempts, checking each one instantly, and learning from the failures without a human in the loop.

Compare that to a legal brief, where quality is subjective and requires a senior partner to evaluate. Or a medical diagnosis, where you might not know if you were right for weeks. Or a financial forecast, where the answer doesn’t arrive for a quarter.

This is also why your domain expertise is so valuable. The skills AI masters fastest are the ones where the output can be checked automatically. The skills that require human judgment to evaluate are, by definition, the ones that will require human judgment the longest. If your work is hard to verify, your expertise is hard to replace.

Code sits at the intersection of both advantages:

• The AI builders know the domain intimately

• The outputs can be verified automatically

This is why AI coding ability has improved faster than AI ability in almost any other domain.

Reason #3: AI That Codes Better Can Be Used To Build Better AI

The printing press could not improve the printing press. But an AI model that writes better code literally builds better AI models.

This self-reinforcing cycle did not exist in any previous wave of literacy. Text literacy and arithmetic literacy both depended on human institutions that improved across generations. But software authorship depends on a technology that improves itself quarterly.

Tom Davidson, an AI researcher at Open Philanthropy, has mapped this dynamic in detail: once you train an AI system as capable as a top human researcher, you can immediately run millions of copies in parallel, doing the work that small teams of experts currently do.

The result:

While the first two waves unfolded across four centuries, software authorship will be widespread within this decade. The clock speed for this wave of universal literacy is fundamentally different.

The Five Key Second-Order Implications For Knowledge Workers

In 2011, when Marc Andreessen wrote in The Wall Street Journal that “software is eating the world,” he meant that software companies were about to displace incumbents across every major industry, from bookstores to taxi fleets to newspapers. At the time, Exxon Mobil was still the most valuable company on earth.

15 years later, the thesis played out exactly as he predicted:

9 of the 10 most valuable companies in the world are now software and technology companies, led by NVIDIA at over $5 trillion. As I’m writing this, the others include Apple, Microsoft, Alphabet, Amazon, Meta, Tesla, TSMC, SpaceX, and Broadcom.

But in another interview on The Lex Fridman Podcast, Andreessen asked a question his original thesis couldn’t answer:

Where are the hyperproductive people? If AI tools are so powerful, why hasn’t productivity exploded?

Software ate the world, but only the technical minority got to hold the fork. The other 99% of knowledge workers could use software but couldn’t create it. The tools were powerful, but the interface was locked.

Now the interface is English, and the implications change everything about how knowledge work is structured.

Five implications in particular stand out:

Implication #1. Domain expertise becomes the scarce resource

For 50 years, the limiting factor in building software was technical skill. That equation is inverting. The deepest expert in every field suddenly has something no software engineer can replicate:

• Visioning. Envisioning what to build out of the universe of things you could possibly build.

• Specification. Articulating how you want the system to work, precisely enough that an AI can act on the description.

• Iteration. Guiding the AI’s output through successive rounds, closing the gap between “almost right” and “right.”

• Judgment. Evaluating whether the output fits the intent. Cherny said it plainly: “Coding is the easy part. It’s knowing the domain that’s the hard part.”

Brown proved it at the hackathon. He didn’t beat all those developers because he was better at code. He beat them because he understood permit law at a level no software engineer could match. His system was powerful because Brown’s domain knowledge was encoded at the right level of detail and nuance.

Here is what this means for everything you’ve already built: every year you spent learning your domain, every hard case, every pattern you internalized, every judgment call you learned to make without thinking, all of it just became more valuable, not less. The investment wasn’t wasted. It was preparation for a moment where the only thing standing between an idea and a running system is the depth of the knowledge behind it.

Will AI eventually develop that depth of domain expertise on its own? Possibly.

But that’s exactly why the window described in this article matters.

Implication #2. “I’m not a programmer” stops being a valid professional identity.

For generations, knowledge workers have sorted themselves into “technical” and “non-technical” as though these were permanent traits. Résumés announce these identities. Job descriptions enforce them. Entire career paths were built around which side of the line you fell on.

For twenty years, I organized my entire career around “I’m not a programmer.” It explained a whole category of things I didn’t do and wasn’t going to try. Letting go of that identity was harder than learning any tool.

I’m not the only person crossing that line, now that the interface to building software has shifted from code to English. Brown, Mcinerney, and Roth all wrote zero lines of code. What they wrote were clear descriptions of what they needed, in the language of their own domain.

What most people miss: the line between “technical” and “non-technical” has never been fixed. It’s been moving your entire career. And depending on your age, you may already have crossed it as many as three times.

In the 1990s, using a computer at work was a technical skill. Senior professionals had assistants who handled email, printed documents, and managed digital calendars. “I don’t use computers” was a reasonable professional position. By 2005, it was a career-limiting one. The skill stopped being called “technical” and started being called “work.”

In the late 1990s and early 2000s, finding information was a technical skill. Professionals relied on librarians, research departments, and junior staff to look things up. Then search engines made it trivial. By 2010, a professional who “doesn’t Google things” wasn’t principled. They were handicapped.

In the 2010s, working with data was a technical skill. Analysis lived with BI teams and dedicated analysts. Then tools like Google Analytics and Tableau put basic data work within reach of anyone. By 2020, every marketer, product manager, and executive was expected to be “data-driven.” The ones who said “I’m not a numbers person” couldn’t justify their own decisions.

Each time, the same thing happened:

• A tool made a specialist skill accessible.

• The skill migrated from “technical” to “expected.”

• The people who clung to the old identity found themselves dependent on specialists for things their peers now did for themselves.

And each cycle moved faster than the last.

The Literacy Arc tells you exactly where this ends. The shift from illiteracy being normal to being unacceptable took reading centuries. For software authorship, it will take a decade or less. Possibly much less.

Implication #3. Software becomes personal, not just institutional.

Today, software is something your company buys. Salesforce. SAP. Workday. Big platforms built for millions of users, none of whom get exactly what they need. Knowledge workers spend years learning to navigate the limitations of tools that were never designed for their specific workflow.

Software authorship changes the unit of production. Instead of one tool for 10 million users, you get 10 million tools for one user each.

• The lawyer builds a permit assistant that fits his exact jurisdiction.

• The accountant builds a depreciation tracker that matches her exact fund structure.

• The editor builds a clip tool that fits his exact editorial process.

This is not a marginal improvement to existing software. It is a different category of software entirely.

Before universal literacy, personal writing didn’t exist. If you needed a message sent, you hired a scribe.

After universal literacy, you just wrote. And the diversity of personal writing turned out to be enormous: letters, diaries, grocery lists, notes to self, to-do lists, journal entries, thank-you cards. Nobody thinks of these as “writing” in the professional sense. They’re life infrastructure.

Likewise, personal software isn’t “custom software.” It’s externalized expertise. In the same way that a personal letter externalizes what you think in a way no published book can, personal software externalizes what you know and how you work in a way no SaaS product can.

I mentioned my 87 tools earlier. Here’s what’s actually inside three of them:

• A 27-step news analysis pipeline that runs every AI news event through my specific model of how technology reshapes society: 2,000+ mental models, 300+ paradigms, 20 first principles, 11 analytical frameworks, 200+ effect chains, four master scenarios with probability weights. No other analyst has this model. It took me years to build the thinking. The software took weeks.

• A writing voice system that encodes my exact style: my signature phrases, my structural moves, a library of my actual published sentences that get transplanted into new articles. It’s 1,200 lines of my craft, externalized.

• An AI Second Brain with over 12,000 notes, searchable by meaning, not just keywords. When I start a new article, the system surfaces my own prior thinking, connections I’d forgotten, tensions between ideas I haven’t resolved yet.

These are my unique expertise, made into running systems, built by me, even though I’ve never written a line of code.

Implication #4. Knowledge workers become capital creators, not just labor sellers.

Marc Andreessen made an observation on The Lex Fridman Podcast that reframes the economics of what we’re describing:

Software is “our modern philosopher’s stone” because it “transmutes labor into capital.”

Someone sits at a keyboard, types, and a capital asset comes out the other side. An asset that works when they don’t. An asset that can compound for decades.

Until now, that transmutation was reserved for software engineers and the entrepreneurs who could afford to hire them. Everyone else sold hours.

Software authorship extends the transmutation to every knowledge worker. Brown’s permit assistant isn’t equivalent to billable hours. It’s an asset that processes applications whether he’s at his desk or not.

Mcinerney’s depreciation tracker isn’t a spreadsheet she maintains. It’s a system that runs her fund’s quarterly close.

Roth’s clip tool isn’t a favor he asked engineering for. It’s infrastructure that his 400-person team uses daily.

The ceiling for knowledge workers has often been: work more hours, charge more per hour.

Software authorship changes the ceiling to:

How many capital assets can you create from what you know?

The first time I built a tool that ran a process I used to do by hand, and it worked while I was asleep, something shifted in how I understood my own career.

I wasn’t selling hours anymore. I was building things.

That shift felt less like a productivity hack and more like a change in what my work actually was.

Implication #5. The productivity multiplier is real, and it compounds.

As briefly shared earlier, Garry Tan is the CEO of Y Combinator, the most influential startup accelerator in the world. In 2008, he co-founded a blogging platform called Posterous. It took a team of ten, $10 million in venture funding, and two years to build it.

In early 2026, Tan rebuilt the whole thing. Alone. In 90 hours. While having a full-time job and kids. Using AI. Only weeks after learning Claude Code.

The codebase is over 70,000 lines. He has never looked at a single one.

He described the experience this way:

“I took modafinil just to stay awake longer to be able to turn the momentary crystalline structures I had in my brain into lines of code before sleep or human distraction turned it to grains of sand. I love coding, but I love coding with AI even more. I speak, it listens, and we create. I see the structure and it is built.”
— Garry Tan, TechCrunch, March 2026

Tan tracked the numbers. His 2026 output rate is 810 times his previous coding pace: 11,417 logical lines per day versus 14 lines.

This is not an isolated case. Marc Andreessen is seeing the same thing on the investment side. On The Monitoring The Situation Podcast, he told this story about one of his partners at Andreessen Horowitz:

“We have one partner who’s built an entire AI system for everything that he does at work, and he is absolutely excited about it, and it works great, and he loves it. And it’s like his partner in all of his work now.

And I asked him, I said, have you looked at the code? And he’s like, ‘Hell no.’

He’s not a programmer by background. And yet all of a sudden he’s hyperproductive.”

“At our leading-edge companies, estimates are the leading-edge programmers are like 20x more productive than they were a year ago. Like, it’s the most dramatic increase in programmer productivity in, like, ever. And coding is the first domain in which this has happened. Now people want to project forward and say this is going to happen in every area of knowledge work. And I think you can predict a similar outcome.”
— Marc Andreessen, Co-Founder of a16z, MTS Podcast, 2026

This same pattern is showing up inside large companies, not just startups. Tobi Lutke, the CEO of Shopify, sent an internal memo in April 2025 describing what he was seeing across the company:

“I’ve seen many of these people approach implausible tasks, ones we wouldn’t even have chosen to tackle before, with reflexive and brilliant usage of AI to get 100X the work done.”
— Tobi Lutke, CEO of Shopify, Internal Memo, April 2025

Lutke didn’t frame it as an opportunity. He framed it as a mandate: teams must now demonstrate why they cannot get what they want done using AI before requesting additional headcount.

At Anthropic, the average increase among all developers since Claude Code was released in 2025 is 250%, with no decrease in quality, according to Boris Cherny. To put this in perspective, when he worked at Facebook, before AI models, the average increase in productivity per year was 1-3%:

Source: Boris Cherny on the Big Technology Podcast

Before Anthropic, I used to work at a big tech company [Facebook]. One of my responsibilities was the health of all of the code across Meta’s apps. So this is Facebook, Instagram, WhatsApp. And ne of the reasons that we care about the health of the code, and this is essentially things like code quality is engineers are more productive. And there was a big team of people that worked on productivity. Before models like Claude, you would work for a really long time, and you would see maybe a 1-3% improvement in productivity per engineer over the course of a year, something like that. And that was a pretty big improvement. And it was a very hard one. You essentially had to try a lot of ideas, and eventually you find something that improves productivity.

What happened with Claude is now many companies, including Anthropic, and all of our biggest customers, are reporting gains on the order of hundreds of percentage points. And I think the last number that we reported is the amount of code written per engineer at Anthropic has grown something like two hundred and fifty percent since we introduced Claude Code. And this is while keeping code quality and reliability and all of these things kinda stable. So without those things regressing, the volume of code has grown a lot.

This kind of productivity impact I think is just very new.

Reid Hoffman’s #1 AI advisor runs 54 parallel agents and checks in just once a day on many of them:

Cherny takes things a step further and runs hundreds of parallel agents at once overnight:

Source: Boris Cherny on the Big Technology Podcast

The question is not whether this multiplier is real. The data is already in. People at the frontier are experiencing it. The question is whether you’re on the compounding curve or watching it from outside.

The Software Authorship Window Closes Around 2030-2035. You Have 5 Years.

The Literacy Arc has a window between Stage 2 and Stage 4: the period when the skill is available but not yet universal. For reading, that window lasted 400 years. For arithmetic, 400 years. During those windows, the people who could read, write, and calculate captured asymmetric advantage.

They ran the businesses. They held the offices. They wrote the laws.

The software authorship window opened in late 2025. It will close in the next 5-10 years.

The CEO of OpenAI, Sam Altman, is already seeing this play out. At Stripe Sessions in 2026, he said:

“For a long time, I think the most important ingredient that I looked for, YC looked for, that this part of our industry looked for on a founding team was technical talent. And that’s still very important, but now people who just really deeply understand their users and can’t code at all, I want to fund those people. And that’s a big turnaround.”
— Sam Altman, CEO of OpenAI, Stripe Sessions, 2026

He called it “the revenge of the idea guys.” For decades, Silicon Valley dismissed people who had great ideas but couldn’t build them. Altman is saying the dismissal no longer holds. And he’s not just saying it. He’s deploying capital behind it.

Altman is also funding the technology that will eventually close the window he’s telling people to climb through. The better AI gets at coding (and everything else that once belonged solely to humans), the more quickly Stage 4 approaches.

No one knows exactly when AI will be able to do everything a human knowledge worker can do. But the people closest to the technology have guesses, and they’re worth hearing, because they aren’t speculating from the outside.

• They are building the very systems that enable Software Authorship.

• They are working with models that are six months to a year ahead of what the public has access to.

• They understand the trajectory of the technology because they are creating it.

Yes, they are biased. They have every incentive to overstate the pace. But they also have something almost no one else has: direct contact with where these systems are right now and where they are heading next.

Here are the three most prominent AI leaders in the world, in order from the longest to the shortest predicted timeline.

Demis Hassabis, the CEO of Google DeepMind and a Nobel Prize winner in chemistry, has been working toward artificial general intelligence since he co-founded DeepMind in 2010. At the time, almost nobody was working in AI, and most people in tech thought the field was a dead end. His co-founder, Shane Legg, wrote blog posts back then predicting when AGI would arrive. Those posts are still on the internet.

“We’ve been very consistent how we define AGI as basically a system that exhibits all the cognitive capabilities the human mind has.

[…]

I’ve got a probability distribution around the timings, but I would say there’s a very good chance of it being within the next five years. So that’s not long at all.

[…]

We used to do this extrapolation of compute and algorithmic progress, and basically we predicted around 20 years it would take from when we started out, and I think we’re pretty much on track.”
— Demis Hassabis, CEO of Google DeepMind

Hassabis points out that DeepMind has been on track with its original 20-year prediction since 2010. This isn’t a sudden burst of optimism. It’s a forecast that has held for over 15 years. If he’s right, AGI is only four years off.

Dario Amodei, the CEO of Anthropic and the company behind Claude, puts the timeline closer.

“There’s a lot of problems that are basically, like: ‘we can do this when we have the country of geniuses in a data center.’ ...If you made me guess, it’s like one to two years, maybe one to three years, it’s really hard to tell. I have a strong view, 99%, 95% that, like, all this will happen in 10 years. I think that’s just a super safe bet. And then I have a hunch, this is more like a 50/50 thing, that it’s going to be more like 1 to 2, maybe more like 1 to 3.”
— Dario Amodei, CEO of Anthropic

Amodei’s framing is unusually precise for this kind of prediction. He is nearly certain (95-99%) that AI will match all human cognitive capabilities within ten years. His personal hunch, at 50/50 odds, is that it happens within one to three years.

Elon Musk, who runs xAI (the company behind Grok) in addition to Tesla, SpaceX, and several other companies, gives the shortest timeline of all.

“I’d be surprised, by the end of this year, if digital human emulation has not been solved. Can you do anything that a human with access to a computer could do?”
— Elon Musk, CEO of xAI, Tesla, and SpaceX

Musk believes AI could match any human performing knowledge work by the end of 2026.

So the range, from three of the people arguably most responsible for building AGI, is sometime between the end of this year and four years from now.

I don’t know who is right. Nobody does. But if Amodei and Musk are closer to the mark, the breakthrough that pushes AI across that line could happen very, very soon.

That said, here is what I want you to take away from this section. It isn’t the specific date.

Even if AI never gets a single percentage point better than it is today, the case for software authorship is already overwhelming. AI is good enough right now for anyone to build custom software that transforms their productivity.

If AI never gets any better than it is today, it’s still good enough to drive this third universal literacy arc we’re in to completion.

The window is closing, no matter what.

But the people who are building with AI today are already capturing the same kind of premium that scribes captured in 1500 and merchants captured in 1600: the advantage that comes from having a skill that most of your peers haven’t developed yet.

If you join them, you’ll also enjoy:

• An exponential increase in productivity and creativity.

• The ability to spend more of your time on the work only you can do.

• The capacity to turn your knowledge into systems that run whether you’re at your desk or not.

That premium is available to you right now. And the advantage compounds.

An accountant who builds her first tool this year and learns from the experience will build her fifth tool next year and her twentieth the year after that. She will develop the visioning, specification, iteration, and judgment skills that can only be built through practice.

The question isn’t whether AI will eventually do all of this without you. The question is whether you’ll have built your advantage before it does.

You’ve Already Done the Hard Part

A friend sat me down and said, “I think you can do this. Let me just show you.”

That single conversation changed the trajectory of my career.

He didn’t teach me to code. He didn’t walk me through a tutorial or assign homework.

Instead, he showed me that the barrier I’d been running into for twenty years was gone. That the thing standing between me and building software had been quietly removed, and I hadn’t noticed.

What my friend saw, and what I couldn’t see yet, was that I already had everything that mattered.

• I had the domain expertise.

• I had the judgment.

• I had twenty years of knowing what was important in my field and why.

The only thing I was missing was the knowledge that the barrier was gone. He gave me that one piece. Everything else, I already had.

I’ve spent this entire article trying to do the same thing for you.

Not to scare you. Not to pressure you into a career change you don’t want. To show you the door is open, and that what’s on the other side isn’t what you think.

You don’t need to learn to code. You need to find out what happens when you describe what you already know to a system that can build it.

The window is open. Your expertise is the thing that matters most. And the tools speak your language now.

I think you can do this. Let me just show you.

MORE COMING SOON

This article is the first part in a series to help you make the shift to becoming a software author. The goal of this article was to help you make the commitment. To help you understand the “why to.” Without that, no “how-to” advice will help.

In the coming articles, I will help you understand:

• How to leverage your domain expertise with four new skills that will pay you back forever.

• How to most quickly learn agentic tools like Claude Code and Codex.

MULTIMEDIA VERSION

Access Slides >>

Download Audio >>

The Agentic Academy for Knowledge Work is a 10-week cohort that migrates you from using AI as a chatbot to running an agentic system built around your own work, with white-glove support every step of the way. Starting in Claude Code and structured so it transfers to Codex.

Created by two of the leading writers and teachers in AI knowledge work, it is the first and only cohort of its kind. Because it is not a course you watch. It’s a concierge-style program where we guide you, hands-on, through building a custom AI work system tailored to your specific work, your specific domain, and your specific goals.

The full cohort will start on June 22.

This page is a detailed look at what’s inside the program:

• The week-by-week curriculum

• A comprehensive skill suite

• 24/7 community support

• Weekly tech calls to help you get unstuck

And also…

• Who it’s for

• What you walk away with

• Pricing

• How to join

You don’t need to know how to code. Claude Code speaks English. You don’t need to be technical. You describe what you want, and AI reads your files, applies your standards, and produces ready-to-use work.

What’s Inside: The 10-Week System

Every week has a practical outcome. Every session is a build session, not just a lecture. You are not learning disconnected features. You are building one layer at a time until Claude can help with a real workflow that saves you time and allows you to do things that were impossible before:

We start in Claude Code because it is the clearest place to learn the core agentic primitives:

• Project folders

• Source files

• Rules

• Skills

• Workflows

• Permissions

• Review and evaluation.

We also show how that same structure adapts to Codex. The point is not to make you dependent on one product. The point is to teach you how the agentic paradigm actually works.

Every session is recorded and available on-demand within 24 hours. You keep access forever.

What You Walk Away With

By the end of the 10 weeks, agentic work is no longer an abstract idea. You have a working version of it built around your own project:

1. One real AI system you can keep using

You leave with one piece of your actual work turned into a process Claude can help run again.

• It takes a real input from your work and turns it into a finished output.

• It uses your project files, standards, examples, and decision rules.

• It connects to the apps where your work already happens, like Granola, Google Docs, Gmail, Notion, Asana, or Slack.

2. AI that understands how your work works

Your agent stops behaving like a blank chat box because you have given it the material it needs to help properly:

• It understands one real project well enough to help without a fresh explanation every session.

• It knows which files, examples, notes, and references to use for different tasks.

• Your repeated corrections turn into reusable rules and examples, so the same mistakes are less likely to come back.

3. The ability to build the next process yourself

The first process is the proof. The bigger win is knowing how to build the next one:

• You know how to turn repeated work into skills instead of starting from a blank prompt.

• You can orchestrate subagents for larger jobs instead of forcing everything into one overloaded chat.

• You know how to let Claude help while protecting private files, reviewing risky changes, and keeping sensitive data from being shared publicly.

And underneath all of it, the one idea that makes everything portable: an agent’s output is shaped by what it can see. Once you can control that, you can pick up any new tool and put it to work.

What This Changes In Practice

Before the cohort, AI work usually depends on you holding the whole system together. You remember which context to paste, which examples matter, which standards to mention, which app has the source material, and how to clean up the output when the AI misses something obvious.

After the cohort, more of that structure lives outside your head. Claude knows where the project lives, which files matter, what good output looks like, what mistakes to avoid, and which process to run for the task in front of you.

That can look like:

• A client call turns into a prep brief, follow-up, and deliverable draft that pulls from the client’s notes, your previous deliverables, and your delivery standards.

• A messy content idea turns into a researched draft, voice-aware edit, and platform-specific version without rebuilding your style guide every time.

• A week of meetings and scattered updates turns into a status report that surfaces decisions, risks, owners, and next actions for the right audience.

Claude still needs your judgment. The first change is that you are no longer spending the first half of every task getting AI back up to speed. The second work is that AI actually does much of the work for you.

Our Approach

We designed the program backward from what would maximize results for all students:

1. Built-in support from start to finish

You do not have to invent the setup yourself. Chat X-Ray helps turn your AI history into a starting project. The Skill Library gives you ready-to-use skills for your type of work. The Skill Upgrader and project audits help improve what you build during the cohort and after it ends.

2. You do not get stuck alone

You are not left alone with a tutorial and a folder full of files. The live sessions, WhatsApp community, and instructor support are there when you are choosing a project, structuring your files, fixing a stuck process, or deciding what to build next.

3. You stay in control of what Claude can touch

You learn how to give Claude useful access without handing it unlimited control. We help you set up permissions, review steps, and simple checks so you can protect private files, review risky changes, and reduce the chance of sensitive data being shared publicly.

What You Get When You Join

Everything below is part of the cohort. The live sessions teach the system. The skills, templates, and support help you build it around your own work.

1. The Live Cohort

• 10 live build sessions: 90 minutes each, Mondays at 11am ET, with two instructors helping you build on your real project.

• All session recordings: available within 24 hours and yours to keep forever.

Learn more about the curriculum

2. Community + Support

• WhatsApp support: Join a community with a 24-hour response guarantee, so you are not stuck waiting a week when something breaks.

• Weekly tech support call: provide a weekly Q&A call on Thursday at 11:00am-12:00pm EST to answer all of your questions that are causing you to get stuck.

Learn more about the community & support

3. Concierge Skill Suite

We built and tested the first of its kind skill suite to make the migration to the agentic paradigm a whiteglove experience:

• Chat X-Ray: a personalized analysis of your ChatGPT or Claude history that shows how you use AI, where your time is going, and which project is the best place to start.

• The Chat Exporter exports your complete conversation history from Claude, ChatGPT, and Gemini, converting raw platform data into clean, organized files that Claude Code can work with.

• The Context Exporter goes beyond chat to pull your professional data from Substack, LinkedIn, X, Zoom, Google Meet, Microsoft Teams, and other platforms where your work lives, giving Claude Code access to the full picture of your professional output.

• The Chat X-Ray is a comprehensive, AI-powered analysis that reads your entire chat history and produces a personalized report showing you exactly how you use AI, where your time is going, what’s costing you, and what changes when you move into Claude Code.

• The Skills Library Builder takes the patterns the Chat X-Ray identifies and automatically builds them into reusable Claude Code skills, so the prompts you’ve been re-typing for months become one-command tools you never have to explain again.

• The Pipeline Builder identifies workflows that were fragmented across dozens of separate chat conversations and consolidates them into single, repeatable pipelines that run in Claude Code with full continuity.

• The Knowledge Repository Builder organizes all of your exported data, frameworks, voice guides, methodologies, and institutional knowledge into a structured system that Claude can search, reference, and draw on automatically.

• The Skill Finder: analyzes your project folder and the skills you have already built, then suggests 5 to 10 new skills worth exploring based on the work you are actually doing.

• The Skill Upgrader analyzes any skill you’ve built, identifies the mental models and thinking patterns that would make it perform better, and rewrites the skill with those improvements integrated, so your tools get smarter every time you use them.

• Starter project folder: the first version of the project structure Claude will use during the cohort.

• 1 to 3 starter skills: generated from patterns in your AI history, so you do not arrive at Week 1 with an empty folder.

• Claude Code setup and Codex transfer pattern: the starting tool and the portable structure you can carry into Codex.

Learn more about the skill suite

4. The Agentic Archetype Kit

Each student builds the same underlying system, but the examples, skills, and workflows change based on the kind of work you actually do.

• A starter project structure for your archetype: content creation, consulting and coaching, entrepreneurship, or knowledge work.

• A 10-skill Skill Library curated for your type of work, installed in Week 5 and ready to use immediately.

• A starter workflow that matches your archetype, so you can see what a finished process looks like before building your own.

• A capstone option designed around the kind of work you want the system to help with.

5. Build templates

You do not have to create every structure from a blank page. The cohort gives you the working pieces you need to build, test, and improve your setup.

• CLAUDE.md and AGENTS.md starter template

• Reference routing template

• Correction log template

• Skill customization template

• Custom skill template

• Workflow mapping template

• Sub-agents template

• Tool selection guide

• Capstone submission template

• Maintenance plan template

6. The improvement tools

The system is meant to keep getting better after the first version works.

• The Automation Recommender: reviews your project, hooks, plugins, MCPs, and subagents, then suggests the next automations to add, from new hooks to useful tool connections to subagents that could take on part of the work.

• The CLAUDE.md and AGENTS.md Improver: analyzes your project files and improves the instruction files Claude Code and Codex use to understand your work. It is built for knowledge workers, not developers, so the suggestions stay tied to your actual projects, standards, and recurring tasks.

• Quality checks and simple evals: help you compare output before and after changes, so you can see whether the system is saving time, improving quality, or both.

• AI-specific mental models: the thinking patterns that help you design better skills, clearer workflows, and safer delegation.

Who It’s For, And What Changes For You

The system is the same for everyone. The skills, examples, and workflows change based on what your work actually is.

1. Content creators (newsletter writers, bloggers, podcasters, YouTubers): Claude knows your voice, audience, content pillars, and publishing standards. A brain dump becomes a researched draft becomes a voice-checked edit becomes platform-specific social content, in one workflow. Your content operation runs like a studio instead of a solo scramble.

2. Consultants and coaches: Claude knows your clients, frameworks, and deliverable standards. Client notes become prep briefs. Meeting transcripts become insight summaries with running profiles that remember last month. Each client no longer needs a fresh introduction every conversation.

3. Operators and knowledge workers: Claude knows your projects, stakeholders, reporting formats, and decision criteria. Meeting notes become status reports. Scattered updates become action plans. The weekly report that takes two hours becomes a workflow that takes twenty minutes.

4. Entrepreneurs and business builders: Claude knows your product, market, customers, and priorities. Customer interviews become insight summaries. Strategy sessions become execution plans. The five hats you wear every day finally get their own infrastructure.

Who This Isn’t For

We would rather you find this out now than after you pay:

• If you want a tool that works the second you open it with nothing to set up, this will frustrate you. The first few weeks are foundation-building before it pays you back.

• If you are hunting for a pack of prompts to copy, this is the wrong room. The whole point is to move you past that.

• If you don’t have time after the class to apply what you learn. Watching alone will not do it.

• If you don’t have proven workflows you already use without AI. Without a body of past work for the agent to learn from, you will only see a fraction of the value.

If you are still reading and nodding, you are exactly who we built this for.

Summary

• Format: 10 weekly live build sessions, 90 minutes each.

• Schedule: Mondays, 11am ET. Starts June 15, final session August 17.

• Tools: Claude Code first (Claude Code Desktop), structured to transfer to Codex. Obsidian as your local file home.

• Community: WhatsApp, 24-hour response guarantee.

• Recordings: every session recorded, yours to keep forever.

• Requirements: an Anthropic account with Claude Code Desktop access, and one real project with a body of past files to build around. Setup instructions sent before Week 1.

The 30-day guarantee

Try the first 30 days of the full program. Attend the sessions. Do the work. Build on your real project. If you have put in the effort and the system isn’t delivering, email us within 30 days of the start date, and we will refund your full tuition. No hoops. No argument.

Frequently Asked Questions

No. The cohort is built for people who have never written a line of code and don’t intend to start. Claude Code speaks English. The skills that matter are the ones you already have: knowing your domain and knowing what good output looks like.

The same goes for being technical. You don’t need to be the person in your community that people turn to when the computer stops working.

Because you have adapted to the constraints so well that you have stopped noticing the cost. The Chat X-Ray skill we provide makes those costs visible: early users found they were spending 12 to 26 hours over a 20-month period on pure overhead like re-explaining context and re-pasting prompts. That drops to near zero in Claude Code. For more on the invisible ceiling of chat, read The Chat Trap: Why the Smartest AI Users Are Working the Hardest.

You can learn individual features from YouTube. What you can’t get there: a personalized analysis of your work patterns, a curriculum that builds each concept on the last, live help when you are stuck, and instructors who look at your actual project and tell you what to fix.

Claude Code is the clearest place for a non-technical professional to experience agentic work: files, structure, permissions, workflows, and review. The mental models and workflows transfer to other tools. We show how the same structure adapts to Codex.

Live is better, since the sessions are build sessions, not lectures. Every session is recorded and available within 24 hours, and you keep access forever.

Plan for 2-4 hours per week, building on your real project.

Every session is recorded, the WhatsApp community gives async support, and because everyone builds a different project there is no single pace to fall behind from. The Ongoing plan also includes unlimited retakes.

Bring one real project you spend at least a few hours a week on, with past files to draw on. We recommend sticking with one so each week builds on the last, but we will help you pivot early if it turns out to be a poor fit.

Yes. The Ongoing plan is built for teams of two or more, and the Custom plan for whole teams of five or more.

Ten weeks. Your real work. A complete AI system by the end.

Featured Testimonials

“Michael’s strategies helped us get articles published on Fortune, Business Insider, and Inc. We even had one article earn over $100,000 in revenue! Seminal works, and there is nothing like it on the market.”

—Russ Ruffino, Clients On Demand

“Over my career, I’ve sold over $100 million dollars in online courses to over one million paying students. I’ve written several books, and I know the best copywriters in the world. Michael is the smartest article writer I’ve ever met. He has mastered the unique art of creating free, high-quality educational content that is also high-converting. If you are a coach, consultant, or thought leader who wants hundreds or even thousands of leads from every article you write, there is no other program like this in the world.”

—Eben Pagan

Read More Testimonials

Pricing

Three ways to join, depending on how far you want to take it:

Discount For Founding Members Of Michael And Wyndo Newsletter

Founding-member discount for AI Maker and Blockbuster Blueprint : $500 off any plan. This is the very first cohort. Founding members lock in the lowest price this will ever be, and the price goes up after this run.

Join Now

Option #1: 10-Week Cohort

Join 10-Week Cohort

Option #2: Ongoing

Join Ongoing Program >>

Option #3: Enterprise

Fill Out Form To Setup Call

I want to tell you about something that’s been bugging me for months.

I was using AI every single day. ChatGPT, Claude, NotebookLM—the whole stack. And I genuinely felt productive. Ideas were flowing. Content was getting drafted faster than ever. I was convinced I’d figured this thing out.

But every session ended the same way.

Command+C. Command+V

My workflow looked like this:

• Brainstorm newsletter ideas in ChatGPT → copy to Notion.

• Research topics in NotebookLM → manually pull insights into Docs.

• Draft content in Claude → copy to Substack editor.

• Generate social posts → copy to a scheduling tool.

• Analyze what’s working → screenshot and paste into my tracking spreadsheet.

I was doing this dance every day without questioning it.

Then one afternoon, I caught myself mid-paste and just... stopped.

I looked at what I was actually spending my time on. I wasn’t thinking or creating. I was just deciding where things should go, how to format them, what context to add by hand, and how to connect this output to everything else I’m working on.

I was a human API—the glue between AI outputs and the places where my real work lived. AI was supposed to free me from mechanical busywork. Instead, it created a different kind.

For months, I’d been stuck on the wrong question:

“Which AI tool should I use?”

Then I asked a different one:

“Where does my AI actually live?”

My AI lived in a chat window. My work lived in Google Docs, Notion, Slack, email, and five other places. I was the bridge between them—every single time.

I started calling this the “copy-paste tax.” And I’d been paying it for months without noticing.

If any of that sounds familiar, I think you’ll relate to what I’m about to share.

The truth was I didn’t realize how stuck I was until I experienced what comes next.

What it actually feels like when AI doesn’t just help you think—but can execute inside the places where your work lives.

When you go from being the middleman copying outputs between tools to directing AI that can read, write, and act across your entire workflow.

I’ve been calling this an “agentic AI workspace.” Not just AI that answers questions—AI that can act on your behalf, in the places where your work actually happens.

And once I experienced it, I couldn’t go back.

Let me show you what I mean.

Subscribe now

The Three Levels of AI Integration

When I started mapping out how people actually use AI, I noticed a pattern. There are roughly three levels—and most people are stuck at the first one without realizing there’s anything beyond it.

Level 1: The Chat Window

This is where most of us start. You open ChatGPT, Claude, or Gemini. You ask a question. You get an answer. Then you take that answer and manually put it somewhere useful.

The AI is helpful. But it has no idea where your work lives, what you did yesterday, or what you’re trying to accomplish this week. Every conversation starts from scratch. Every output requires you to do the last mile.

This is fine for quick questions. But if you’re trying to run a business, manage projects, or produce content consistently—you hit a ceiling fast.

Level 2: AI With Access

This is the shift that changed things for me.

Instead of AI that lives in a chat window, you start using AI that can actually read and write inside your work environment.

There are a few ways this is happening right now:

1. Google’s approach: Gemini is already woven into Gmail, Docs, Sheets, Drive, YouTube, and NotebookLM. If you live in Google Workspace, your AI can already summarize email threads, draft responses, analyze spreadsheets, and pull research—without you ever leaving the tools you’re already in.

2. Claude’s approach: MCP (Model Context Protocol) lets you connect Claude to your email, calendar, Google Drive, project management tools, and local files. Claude Code takes it further—it can read, write, and execute across entire file systems and repositories.

To be honest, this is where OpenAI has lost me for a while. They don’t have strong multi‑app integration like the other two. I know they have the app store, but those don’t really change how I work.

The difference from Level 1 is simple but significant: AI that has edit access to where your work actually happens. You stop copying outputs out of a chat window. The AI puts things where they need to go.

Level 3: The Agentic AI Workspace

This is where it gets interesting—and where I’ve been spending most of my time with Claude Code.

Level 2 gives AI access to your tools. Level 3 is when your AI can chain actions together across those tools without you orchestrating each step.

Here’s the litmus test I use:

“Can your AI read your meeting notes, draft a follow-up email, update your project tracker, and flag what needs your attention—without you copy-pasting between any of those steps?”

If you’re still the one connecting the dots between tools, you’re at Level 2. If AI can move through your workflow the way you would—reading context from one place, acting on it in another—that’s Level 3.

And this is where things are moving fast.

Notion just entered this space. Last week, they launched custom AI agents that can run autonomously inside your workspace — on a schedule or triggered by events, without you prompting them. You describe a job in plain language, and the agent handles it: triaging tasks, compiling status updates, answering team questions from your internal docs, pulling feedback from Slack, routing work across projects. It connects to Slack, email, calendar, Figma, Linear, and more. For teams that already live in Notion, this is a serious option — your project management tool just became an agentic workspace overnight.

Claude Cowork is another path worth knowing about. Anthropic recently released it as a less technical alternative to Claude Code. Same idea — AI that can read and write your local files, break down multi-step tasks, create spreadsheets, draft reports — but without needing a terminal. If Claude Code feels too developer-oriented for you, Cowork gives you the same agentic capability through the regular Claude desktop app. I wrote an in‑depth guide on Claude Cowork if you want to go deeper.

We’ve now reached a level of AI integration where it can act across your entire work environment, not just answer questions in a chat. The question is whether you’re taking advantage of it.

I’m not going to pretend I’ve fully arrived here. I’m somewhere between Level 2 and 3, and it’s an ongoing process. But even partial progress has been transformative (thanks to OpenClaw!)

Here’s what that looks like in practice.

What Changed For Me

I’ll give you two examples from my own life — because this shift didn’t happen all at once. Instead, it happened in two phases, and each one showed me something different about what an agentic AI workspace actually means.

My Newsletter: From Scattered Tools to One Repository

For months, my newsletter operation was scattered across five or six tools. Ideas in Notion. Drafts in Google Docs. Research in NotebookLM. Social posts created separately. Performance data in a spreadsheet I’d check once a week and forget about.

None of these tools talked to each other. I was the one holding all the context in my head, manually connecting the dots every time I sat down to work.

Then I made one decision that changed everything: I moved my entire newsletter operation into a single repository.

Everything. Published posts, drafts in progress, content ideas, social media archives, writing guidelines, performance data, daily brain dumps, research notes. All of it, in one place, organized in folders.

And then I pointed Claude Code at it.

I’ve written before about how I turned Claude Code into my personal AI operating system for writing and research—and I genuinely mean it when I say it’s been the single biggest shift in how I work with AI.

Here’s why.

Claude Code doesn’t work like a chatbot to answer your questions. It’s an agent. It operates inside my file system. It can read every file in my repository, understand the relationships between them, and write outputs directly where they belong. No copy-pasting. No switching tabs. No reformatting.

• So instead of “Claude, help me brainstorm ideas” followed by pasting into a separate tool—I say: “Read my top-performing posts and my writing guidelines, then draft social variants that match my voice.” And it writes them directly into the right folder.

• Instead of manually checking what content is working—I say: “Look at my performance data and tell me what patterns you see.” It already has the context.

But the part that really surprised me was when I connected Claude Code to NotebookLM through MCP. Now I can do deep research without ever leaving my working environment. I’ll feed sources into NotebookLM, and Claude Code can pull those research insights directly into my drafts and notes—no more alt-tabbing between a research tool and a writing tool, trying to hold everything in my head.

That combination—Claude Code for execution, NotebookLM for research, both accessible from the same environment—is what finally made the “agentic workspace” idea feel real for me. My research feeds my writing. My writing feeds my social content. My performance data feeds my next topic decision. And none of it requires me to be the one carrying context between steps.

If you want the full technical walkthrough, I wrote a complete Claude Code implementation guide.

My Productivity System: From Organized to Opinionated

The newsletter was the first domino. But the second one changed how I think about productivity tools entirely.

I wrote about this last week — how I replaced Notion, Todoist, and Google Sheets with Obsidian + Claude Code. So I won’t repeat the full setup here. But the reason it matters for this conversation is what it revealed about the difference between a system that stores information and one that thinks about it.

With my old stack, I’d spend 45 minutes every Monday clicking through four apps trying to piece together what actually matters this week. I had all the information. But I was still the one connecting the dots.

When I moved everything into an Obsidian vault and connected Claude Code to it, something unexpected happened. Claude started catching things I couldn’t see from inside my own plans:

• One week, it flagged that I’d deferred the same project for three consecutive weeks — something I hadn’t noticed because each individual deferral felt reasonable. It pulled a line from my own weekly review where I’d written “working on multiple things at the same time ruined my focus” — and used that to restructure the next week’s plan around sequential focus blocks instead of parallel task-switching.

• Another time, it caught that a blocker I’d logged on Monday — a payment issue that needed a colleague’s help — was about to collide with an announcement I’d scheduled for Wednesday. I’d stopped thinking about the blocker. Claude hadn’t.

These aren’t dramatic moments. But stacked together, they add up to something I never had with traditional productivity apps: a system that has opinions about my plan. Not just storage. Not just organization. Actual pattern recognition across weeks of my own data.

That’s the shift I keep coming back to. The difference between an app that holds your plan and an AI copilot that reads your plan, compares it to your past behavior, and tells you what you’re avoiding.

Why Both Examples Matter

The newsletter repo showed me that AI can operate inside creative work. The Obsidian vault showed me it can operate inside strategic work too. Same principle — consolidate into plain files, give Claude Code access, stop being the one carrying context between steps.

And it doesn’t stop at the desk. I use OpenClaw that lives inside my Telegram. It has access to both my newsletter repository and my Obsidian vault. So if I’m on my phone and need to check my sprint progress, capture an idea for next week’s post, or ask what’s on my plate today — I just message OpenClaw from Telegram. The agentic workspace follows me, even when I’m away from my laptop.

But the part that feels closest to what an agentic workspace should actually be is Claude Code’s auto memory. Every time a new pattern comes up during a session — a preference I express, a workflow I repeat, a decision I make — Claude Code saves it to its memory automatically. I don’t have to tell it to remember. It just does. And the next session starts smarter because of it.

That’s the compounding effect I keep coming back to. It’s not just that every new file I add makes the system better. The AI itself is learning how I work, session after session — what I care about, how I like things structured, which patterns matter. That never happened with any tool I’ve used before.

Where To Start

I’m not going to give you a 10-step action plan. If this post resonated, you probably already know which part of your workflow has the most copy-paste friction. The one where you spend more time moving outputs between tools than actually doing the work.

Start there. Just that one workflow.

There are two ways to find your starting point.

The first:

“Where does most of your work already live?”

That’s the consolidation path — move your work into a place where AI can reach it.

But there’s another way in that I think is just as valid:

“Where do you already feel the least friction when working with AI?”

Maybe you’ve noticed that certain tasks with AI feel effortless — the back-and-forth is natural, the output is useful, you’re not fighting the tool. Start there. Don’t reorganize your entire life around AI.

Find the interaction that already works and ask:

“What would it look like if this AI could actually act on what we just figured out together, instead of me copy-pasting the result somewhere else?”

That’s how my newsletter repo started. I wasn’t trying to build an agentic AI workspace. I was just tired of the friction in one specific workflow. The system grew from that.

If you’re deep in Google Workspace — Gmail, Docs, Sheets, Drive — Gemini is the shortest path. The integration is already there. You don’t need to set anything up. Start asking Gemini to act across your Google tools instead of using it as a standalone chat.

If you work with files, notes, or anything that benefits from persistent context — look at Claude Code or Claude Cowork. Claude Code is what I use, but if the terminal feels intimidating, Cowork gives you the same file-level access through the regular Claude desktop app. Either way, point it at a folder where your real work lives. Even something as simple as a project folder with a few key documents changes the dynamic.

If your team already lives in Notion — you might not need to switch anything. Notion’s new custom agents can automate workflows, triage tasks, and connect to Slack, email, and other tools directly inside your existing workspace. Worth exploring before you rebuild from scratch.

If you’re managing projects across multiple apps and feeling the friction I described — consider whether consolidating into plain markdown files (Obsidian or even just a folder of .md files) and connecting Claude Code to it would simplify things. It did for me. It might not for you. But the principle holds: the fewer walls between your AI and your actual work, the less time you spend being the middleman.

You don’t need to transform everything at once. I didn’t. The newsletter repo came first. Obsidian came months later. Each step taught me something about how I actually work — not just how I thought I worked.

Share

The Pursuit

I called this post “In Pursuit of Agentic AI Workspace” for a reason. I haven’t arrived. I’m still figuring out where the edges are, what works and what doesn’t, which parts of my workflow genuinely benefit from AI integration and which ones I’m overcomplicating.

But the direction is clear to me now.

The copy-paste era of AI — where you ask a chatbot for help and then manually move everything into your real work — is a phase. A necessary starting point, but not the destination.

What comes next is AI that lives where your work lives. That reads your context, acts on your behalf, and gets smarter the more you use it — because your environment gives it something to build on.

The shift isn’t really about tools; we need to reframe it as a different question.

Not “which AI should I use?” but “where should my AI live?”

Once you answer that, everything else follows.

I’m still in pursuit. But I’m a lot closer than I was six months ago — and if you’re feeling that same copy-paste friction I described at the start, I think you might be ready to start too.

See you in the next one.