The AI Pilot Scoping Starter Kit
Choose an AI use case, match the problem to automation, AI, or an agent, define a small pilot, and decide how you’ll measure it.
Pairs with How to Actually Get AI Working at WorkStarter Kits
Self-contained guides that pair with my posts, each at three levels: a manual version for a plain chat window, the full folder-based setup for Claude Code or Cowork users, and a section written for your AI agent to build from directly. Free, no signup, and the manual level needs no technical background at all.
Each kit includes a guide you can read on the site and the complete Markdown file you can hand to an agent. Start manually, build the folder-based version if you use Claude Code or Cowork, or give the file to your agent and ask it to set up the system with you.
Choose an AI use case, match the problem to automation, AI, or an agent, define a small pilot, and decide how you’ll measure it.
Pairs with How to Actually Get AI Working at WorkA one-page strategy doc and a decision-record template that give AI something real to execute against. AI can only move at the speed of your strategy if your strategy is written down.
Pairs with Executing at the Speed of StrategyA rulings log that turns your one-off corrections into standing rules, so you stop making the same edits twice. Starts as a single document; grows into a folder your AI tools read before you see a draft.
Pairs with Executing at the Speed of StrategyHave a model that didn’t write the draft check it against one rule you care about, and keep the verdicts where you can reread them. The smallest working version of the review system that built this site.
Pairs with the website-build postBuild the layers of your AI setup that persist between sessions: verified facts, standards, boundaries, a rulings log, routing, and automated checks. Four levels, from one pasted file to a configured agent platform.
Pairs with How Prompts, Context, Harnesses, and Models Shape AI OutputMore to come as posts are published. Subscribe on Substack →