Not another chatbot
You’ve talked to an AI assistant by now. You ask, it answers, you close the tab, and it forgets you existed. Useful — but it’s their assistant, borrowed for a minute, remembering nothing.
A personal AI OS is the opposite of that. It’s a small system that’s yours — one that remembers you, connects to your real stuff, does multi-step work on its own, and quietly gets more useful the longer you use it. Less like an app you open, more like an operating system for your mind and your work.
Spin the sphere below and tap any idea — each one drops what it means and a real example of it in action (hit another for a different one).
Spin the sphere. Tap any idea to see what it means — and a real example of it in action.
So what actually is it?
Underneath, it’s a handful of parts working together. You don’t need to be technical to get the shape of it — tap each piece below and it’ll tell you what it does, and where it’s heading.
Tap a piece of the system to see what it does — and where it's heading.
the loop it runs: discover → understand → build → publish → improve → ↺
Put simply: a memory that holds what matters to you, helpers that do jobs for you, tools that let it touch your real apps (with your permission), a knowledge base that answers with sources instead of guesses, a watchful layer so you always stay in control, and a publishing side that turns the work into things others can use. Around all of it runs a loop — discover, understand, build, publish, improve — that never really stops.
Why this matters — even if you never write a line of code
I think within a few years, having a personal AI system will feel as normal as having a phone. Here’s why it’s worth caring about now:
- It’s yours. Your notes, your context, your data — kept private and working for you, not rented back to you a question at a time.
- It compounds. A chatbot resets every conversation. A personal AI OS accumulates. Every note you add, every preference it learns, makes the next thing easier. It’s the rare tool that’s worth more in year three than on day one.
- It works while you sleep. The boring, multi-step chores — research, drafts, tidying, follow-ups — happen in the background and are waiting for you in the morning.
- You stay the boss. The good version of this asks before it does anything big, shows its work, and hands you the editor’s pen. Control isn’t a feature you bolt on; it’s the whole point.
The loop that makes it alive
Here’s the honest part, and the reason this is the capstone track: this website is one. Everything you’re reading runs on that loop — discover something, understand it, build with it, publish it, then improve from what happened. The essays, the courses, the labs, the little interactive toys — they’re the output of a system slowly learning to run itself, with a human still firmly in the loop.
So this track isn’t theory about some future product. It’s the blueprint of the thing you’re standing inside.
What it is today — and what it becomes
Today, a personal AI OS is a monorepo you can actually build: a public website, research agents, safe tool connections (MCP), a knowledge base, and an observability layer that shows what everything did and cost. Ambitious, but buildable — this site is the proof.
Tomorrow, it grows past what any one person maintains by hand: agents that improve their own tools, memory that spans years, a knowledge surface that spots your blind spots, and a system that publishes continuously while you steer. Not autonomous instead of you — autonomous for you.
What you’ll build in this track
The capstone pulls every other track together into one running system:
- Design a full personal AI platform — memory, agents, tools, knowledge, and observability — that fits together instead of sprawling
- Build a monorepo that scales from a simple blog to a full AI lab
- Wire agents to MCP tools and a live website, safely
- Observe everything: token costs, agent runs, errors, latency — so nothing happens in the dark
- Publish continuously from an agentic system without losing control
It’s built in public, and the courses land here as I build the real thing — so you’re not reading a finished manual, you’re watching (and rebuilding) a system come alive.
Who this is for
- Engineers who’ve finished the Applied AI track and want to build their own platform
- Anyone curious how this very site is built — and who wants one of their own
- People serious about building in public as a long-term practice
You don’t need permission or a company to start. A personal AI OS is exactly that — personal. This is where you build yours.