Y Combinator · YC24 Talk

Personal AGI Is How You Stay Under Your Own Power

Garry Tan argues that building a personal knowledge system — not renting a chatbot subscription — is how individuals reclaim intellectual leverage in the age of AI agents.

Y Combinator · 42 min talk · Aug 6, 2026 · brief by The Daily Read

42 min

Talk runtime

8–10×

Productivity floor

220K

GBrain pages

7,000

YCS26 founders

Tap a timestamp pill below to jump the video to that moment.

The problem

Your skills compound in someone else’s repo

The talk opens not with a pitch deck but with a philosopher. Garry Tan chose Baruch Spinoza to anchor his YC24 address because Spinoza’s problem is yours. In 1673, the Elector of Heidelberg offered the cursed heretic a full professorship — salary, legitimacy, a chair, and “freedom of philosophizing” — provided he kept quiet about the parts that unsettled the established religion. A thousand guilders a year to show up and stop building his own thing. Spinoza turned it down and went back to grinding lenses.

The modern equivalent, Tan argues, is everywhere. Your judgment compounds in your employer’s codebase. Your workflows embed in their tooling. Your pattern recognition — the thing that actually makes you good — resets when you close the tab. “Every comfortable arrangement where your judgment compounds in someone else’s repo,” he said, “is a thousand guilders a year.”

What Tan is diagnosing is structural. The factory broke the craftsperson’s ownership of their tools — the loom belonged to the mill. Knowledge workers assumed they were safe because their tools lived in their heads, where no one could confiscate them. But AI agents change that equation: cognition can now be externalized and versioned. The only question is who controls the version.

The chatbot industry has muddied the water here. Tan draws a careful line between what he means by personal AGI and what is already being sold under that banner.

“I do not mean a chatbot you pay $20 a month to. I do not mean a slightly better autocomplete.”

“I call it personal AGI, not artificial general intelligence for everyone all at once. General intelligence for one person — you. This was a dream of a great many people. Vannevar Bush called it the Memex, a machine that would be an extension of yourself and your brain. And I want to be precise about what I mean because the words ‘personal AI’ have already been captured by marketing departments. I do not mean a chatbot you pay $20 a month to. I do not mean a slightly better autocomplete. I do not mean an assistant that knows your calendar and nothing else. That’s just a subscription you rent. It’s a corporate AGI you don’t own. It resets when you close the tab. It knows what everyone else already knows. And when the company behind it pivots, your so-called assistant gets a lobotomy on someone else’s schedule.”

How they solve it

Build the library, own the librarian

Tan’s answer is a system he calls GBrain, built on a Karpathy-style knowledge wiki: 220,000 markdown pages, twenty-five years of his life diarized. Every email, every meeting, every decision and the reason behind it. The question that determines whether your agent is a genius or a goldfish, Tan says, is who decides which three books are open on the desk — and that librarian function is exactly what a well-built personal context layer provides.

The productivity numbers Tan cites hold up under adversarial scrutiny. He pre-empts the skeptic by running the numbers with every disadvantage applied: assume bloated code, half of it scaffolding, assume self-flattery. The floor is still 8×.

“It’s still 8x at the absolute floor and 10 times that in the middle of the range.”

“Apply the most pathological verbosity penalty you can stomach and assume the agent writes bloated code. Assume half of it is scaffolding. Assume I’m flattering myself, which is always a live possibility. It’s still 8x at the absolute floor and 10 times that in the middle of the range. The number is large no matter how you torture it. Now, this is just code — and if you’re at the beginning of your career, you’re in luck. This applies to design. This applies to product management. This applies to growth. This applies to every part of what you might want to do. The multiplier for coding is not just for coding. It’s for every piece of knowledge work. And it’s not just me. At YC, we get to watch this at portfolio scale. A year and a half ago in the Winter ’25 batch, a quarter of the companies had codebases that were 95% AI generated.”

But the multiplier is only part of the argument. The more consequential insight is about custody. If the cognitive patterns that make you productive — your workflows, your heuristics, your standards — are encoded in skill files that live in your own repository, the leverage is yours. If those patterns live in your employer’s tooling, you have sold another thousand guilders.

Skill files, Tan explains, are the mechanism that changes this. They are versioned documents encoding how you think about a domain — prompts, workflows, the reasoning behind decisions. For the first time in history, a craftsperson’s tools can follow them when they leave. The factory model of knowledge work assumes that cognition stays behind when the worker goes; skill files assume the opposite.

“For the first time in history, your cognition can be extracted, stored, versioned, and owned.”

“The craftsperson owned their tools. That’s what made them free. The factory broke that. The loom belonged to the mill. The knowledge workers assumed we were safe because our tools lived in our heads where nobody could confiscate them. Skill files end that. For the first time in history, your cognition can be extracted, stored, versioned, and owned. The only question is by whom? Remember the thousand guilders? That offer never went away. It got rebranded. Every comfortable arrangement where your judgment compounds in someone else’s repo is a thousand guilders a year — to show up, keep quiet, and stop building your own thing. And that’s why you should start a startup, because this is how you can actually make those skill files work for you.”

The startup, in this framing, is not just a business vehicle. It is the only arrangement where the skill files actually work for the person who built them. Inside someone else’s company, the leverage compresses back to zero when you leave. Outside, it compounds.

Takeaway

The quick version

Your personal context layer — emails, decisions, reasoning history, meetings — is the foundation. An agent without it is a stranger with a library card.
Skill files are the first technology that lets you extract and own your cognition rather than leaving it compounding inside someone else’s system.
The productivity multiplier from agents applies to every domain of knowledge work, not just engineering — design, growth, product, all of it.
Starting a startup is the structural move that makes skill files work for you rather than your employer.

“For the first time in history, your cognition can be extracted, stored, versioned, and owned. The only question is by whom?”— Garry Tan, Y Combinator

Spinoza spent his life grinding lenses to stay free. Tan’s argument is that the grinding is now optional — what remains scarce is not the capacity but the decision to build your own thing.