The KnowledgeBuild my Knowledge

A working memory for your business

Your company runs on what is in your head.

That is your most valuable asset. It is also your biggest risk.

Gibran RubioFounder, Green Tech Packaging

In London, you cannot drive a black cab until you have passed a test called The Knowledge. To pass, you memorize every one of the 25,000 streets inside six miles of Charing Cross, plus 20,000 landmarks, plus the fastest route between any two points. All of it in your head, no map. It takes most drivers three to four years. Scientists have studied their brains and found the memory part physically grows from the effort.

Here is the strange part. GPS did not kill The Knowledge. The apps have the same streets, and the cabbies still beat them, because raw data is not judgment. The Knowledge is not a list of streets. It is knowing which one floods when it rains, which one backs up at 4pm, which fare wants the scenic route. It is context. And for 150 years, the only way to move it from one head to another was to make someone spend four years building their own.

Every founder has The Knowledge of their own business. I run a company called Green Tech Packaging that did $3.6M last year, and for most of that time its most important database was my own head. Every price, every customer, which vendor makes what, who ships it, which SKU is about to stock out. None of it is worth much alone. The value is how it all connects, and that web lived in exactly one place.

This works until it does not. It does not scale, because I cannot be in every conversation. It does not transfer, because my team is each rebuilding a fraction of what I know, separately and imperfectly. And it is fragile in a way founders never say out loud. If I got hit by a bus, a huge amount of the company would go with me.

A business that lives in one person's head is not a company. It is a very stressful job you cannot quit.

So I built mine somewhere else. Not a folder of documents. A memory. Something my team can ask, that never forgets, and, the part that matters, that compounds. Most work in a business does not. You fulfill an order and it is gone. You answer a customer and the answer evaporates. A memory keeps every one of them and makes the next answer better. It is the difference between a salary and equity. One is linear. The other bends upward.

Early on it looks like nothing. A memory that knows my SKUs feels like a slightly better search box. That is the trap. In a year it is the thing that knows a customer is quietly opening more clinics and each one needs the same bags. It flags the stock-out before it happens. It drafts the quote in my voice. It remembers I always book Old Dominion because I have a rate deal. None of those facts is powerful alone. The connection between them is where the decision lives.

It is me, without the part where I have to be in the room for the company to be smart.

I am stepping deeper into sales now, chasing thirty accounts worth $6,000 a month each. The bottleneck is not effort. It is context. A great salesperson is just someone who never forgets what a customer cares about. I am turning that into software, for every account, forever. That is not a productivity tweak. It is cloning the one thing that made me good at this business, without cloning my calendar.

This is the part nobody sees coming. The founders who win will not be the ones who work hardest at reassembling context. They will build the machine that assembles it for them, and spend their attention on the work only they can do. That is what this is. Not a better search box. A moat, and one you can stand up in a few evenings.

Built on open source

Built on GBrain.

GBrain is the open-source memory system created by Garry Tan, President and CEO of Y Combinator, with more than 28,000 GitHub stars. I do not sell the software. I sell the setup that turns an empty database into your company's Knowledge.

View the official repositoryLive GitHub star count for garrytan/gbrain

The Knowledge is an independent implementation service. It is not affiliated with, sponsored by, or endorsed by Garry Tan, Y Combinator, or the GBrain project.

7 daysFrom kickoff to working Knowledge
3 sourcesConnected and organized
5 workflowsBuilt around your real work
You own itLocal-first with private backup

Why it is more than a search box

Not notes. An engine.

It knows how things connectAsk about one customer and it pulls their vendor, their pricing, their last order. It stores the relationships between facts, not just the facts, so one question walks the whole web.
Ask in plain EnglishIt finds by meaning, not exact words. “Who buys the black jars” turns up the account even if you never typed those words. You don't have to remember how you filed it.
It gets smarter overnightEvery night it re-reads the day, connects new facts to what it already knew, and writes down what it learned. You come back to a sharper brain than you left, with zero work from you.
Every answer shows its receiptsIt tells you where each fact came from, and says “I don't know” instead of guessing. Which means you can put it in front of a customer.
It is yours, in plain filesIt runs on your machine and backs up to a private repository only you control. Not another SaaS renting your own data back to you.
It compoundsEvery order, every answer, every decision makes the next one better. Most work in a business evaporates. This is the part that stacks.
Choose your path

Do it yourself, or have me build it.

The install takes about 15 minutes. Mapping your business, importing the right knowledge, and teaching it how you actually work is where the value is, and where I come in.

Path 01

Do it yourself

Free

Best if you are comfortable in Terminal and want to experiment with a personal brain first.

  • Follow the official GBrain bootstrap
  • Create a local memory on your Mac
  • Connect it to Claude Code
  • Create a private GitHub backup
  • Import and organize your own knowledge
  • Troubleshoot and build workflows yourself
Go to the free guide ↓
The free setup guide

Build your personal AI brain in about 15 minutes.

This is the current recommended Claude Code path from the official GBrain documentation. Use a new, empty folder, not an existing project.

  1. 01

    Install Claude Code

    Open Terminal on your Mac and run:

    curl -fsSL https://claude.ai/install.sh | bash

    Close and reopen Terminal, run claude, and sign in.

  2. 02

    Create a new folder for your brain

    mkdir ~/brain
cd ~/brain
claude
  3. 03

    Paste the official bootstrap prompt into Claude Code

    Read and follow every step of:
https://raw.githubusercontent.com/garrytan/gbrain/latest-stable/BOOTSTRAP_FOR_AGENTS.md
Goal: set yourself up as my persistent personal agent in this folder, with gbrain
as your memory. Interview me before writing any identity file. Never invent
answers. Ask before anything destructive. You are not done until
`gbrain bootstrap verify` exits 0.

    Claude Code will guide the installation, ask six identity questions, create the local memory, and connect a private GitHub repository. Approve each step only after reading it.

  4. 04

    Prove that it remembers

    Tell it one small thing to remember. Close the session, open a fresh session in the same folder, and ask for that fact back. If it returns the fact from memory, the setup is working.

  5. 05

    Fill the brain

    Start with notes and documents you already own. GBrain also supports consented imports for email, calendar, contacts, and offline archives. Connect one source at a time and review what is being imported.

Important:

Do not run bun install -g gbrain. The package named gbrainon npm is unrelated to Garry Tan's GBrain. Use the official bootstrap above or the documented GitHub install path.