The Library
Every guide, issue, and resource: AI systems, leadership, and infrastructure, from someone who's building it.
5 guides free to read in full. Members get every download and the complete Forge blueprint. Already a member? Sign in →
The Spark: Getting started
The Hammer: Intermediate
The Forge: AdvancedGetting Started with AI, Seriously
Not a ChatGPT tutorial. How I actually started using AI tools in my workflow: what clicked, what didn't, and the progression from curious to capable. A starting point for anyone who wants to move past the hype and actually build with it.
You Don't Need More AI Tools. You Need an Opinion About AI.
The engineers pulling ahead aren't subscribing to more tools. They have a clear point of view on what AI is for in their specific context, and that opinion does the filtering for them. Without a thesis, every new model release is just more noise to evaluate.
How to Move Fast with AI Coding
AI writes plausible code. "Looks right" and "is right" are different things, and the gap grows as the codebase does. The answer is a layered CI pipeline: six workflows, each catching a different class of failure, running on every PR. Python and frontend quality checks, backend and frontend unit tests, integration tests against real Postgres and Redis, and a changelog gate — all bound by a merge gate that blocks anything red. What each layer catches — and why "small change" is never a valid reason to skip it.
Why Documentation Matters for AI Context
The AI knows everything about code in general and nothing about your code specifically. You can burn tokens restoring context at the start of every session, or you can write it down once in a place where it always gets picked up. Three documents — config.md, architecture.md, and AGENTS.md — each serving a different purpose. The first time takes a couple of hours. Every AI-assisted task in the codebase benefits from them after that.
Do You Actually Know Where Your AI Spend Goes?
Your monthly AI bill is a number. A feature-level breakdown is a map. The number tells you how much you are spending. The map tells you whether you are getting what you are paying for. Two things: an instrumentation wrapper that tags every call at the source and writes token counts to a queryable table, and a regex-first pattern that eliminated 73% of LLM calls on one production feature. The mechanics and the business case.
The Forge Blueprint: Building and Operating a Multi-Agent AI System in Production
I spent months designing, breaking, and rebuilding a production multi-agent AI system. This is the complete account — the architecture that held, the decisions I got wrong, and what it actually takes to keep it stable when real users depend on it. Not a framework. Not a tutorial. A real system, fully documented.
Build From These
The templates and guides I actually use. Grab them, run them, adapt them.