Prompt I use in the video is here.
Free Github spec, free TypeScript template, free MIT-licensed reference implementation — the three resources you need to build your own AI system with a coding agent.
In Lesson 2 I walk you through the three resources that ship with the Personal AI Architecture and get you set up with everything you need.
What you’ll learn
- The Personal AI Architecture Repo — the Spec, the Primers (for your AI coding agent), and the Blueprints (drift-guards your agent will follow)
- The TypeScript Template — a working starting point you fork on Github
- BrainDrive — MIT-licensed reference implementation you can use out of the box
- One setup prompt with Claude Code (works with any coding agent) that forks the template, clones both repos, reads the primers, and runs the conformance tests on a clean baseline — so you’re ready to start building in Lesson 3
The lock-in claim this lesson lands: if you need a team of developers to build and maintain your AI system, you’ve signed up for a different kind of lock-in. The architecture is simple enough for one developer + a coding agent. That’s not a constraint — it’s the whole point.
Resources
Full course playlist: Topics tagged paa-course
The Personal AI Architecture: GitHub - Personal-AI-Architecture/the-architecture: MIT licensed user-owned AI runtime: 4 components, 2 APIs, zero lock-in. Your Memory is the platform. · GitHub
TypeScript Template: GitHub - Personal-AI-Architecture/ts-architecture-template: Contract-first TypeScript template for the Personal AI Architecture: Memory, Agent Loop, Gateway, and Auth with API contracts, conformance tests, and lock-in guardrails to help developers build swappable, local-first AI apps fast. · GitHub
BrainDrive (MIT-licensed reference implementation): https://braindrive.ai
Discussion
Trying the setup prompt? Drop questions or what tripped you up here — I’ll answer.
Previous: Lesson 1 — Build Your Own AI System (Simple Architecture). Next: Lesson 3 — 5 Step Workflow for Building Your AI System.