Start Here - Building an AI System
Why I'm building a full AI system in public, piece by piece - and how to follow along (and deploy it yourself).
AI feels like a lot. The pace, the jargon, the constant sense that everyone's already an expert and you're behind. Here's the thing nobody says: the knowledge is all out there now - what's missing is a place to actually get your hands on it and build, instead of just reading about it. That's what this is.
This is a build-in-public series where I take an AI system from an empty repo to a real, running architecture on Azure - one piece at a time - and you build it right alongside me. You don't need to be an expert to start. You become one by doing.
How it works
Every piece is small and self-contained. We start simple and layer up until the full architecture is standing. The pieces that deploy something open with a "Run this": the project and a one-command deploy to your own cloud. Grab it, run it, then read the walkthrough. It's a playground: poke at it, break it, make it yours.
If any of the words feel foreign - agents, tools, MCP, RAG, tokens - the Reading Room has a short, plain-English page on each. Read what you need, skip what you don't.
What we're building
An MCP server on an Azure Function App, then the connectors and data flow, custom instructions, wiring into the tools you already use (ChatGPT, Claude Code, Grok, Copilot), the service-side infra (RAG, vector, a local service API), and the agent that ties it all together. By the end it's not a diagram - it's a real environment you can deploy and build on.
Follow along
Everything's open. Take it, fork it, deploy it, make it yours. Start with the next piece and build alongside me.