Related Contents to "AI"

The first question when working with a small software firm: what happens if the developer cannot be reached tomorrow? I explain why, in a project built with AI, the knowledge is already in the repository, why the real risk is access, and how handover day comes down to a checklist and a “hello”.

Part of this site's code is written by AI agents, and they do it from two separate command-line tools — Claude Code CLI and Antigravity CLI — with the same repository, the same rules and the same tools. I describe what it looks like to build the product that runs the agent, and the model behind it, as replaceable parts: skills, tools, two mirrored permission layers and where a rule actually lives.