-
Verifier Engineering: How to Build the Checks That Make AI Agents Trustworthy
I built a documentation agent to cover every endpoint in a backend API. It ran for 40 minutes, produced a clean 3,000-word spec, and confidently reported it had covered all 47 endpoints. I shipped it. Two weeks later, a colleague opened the spec to check a specific endpoint. It wasn’t there. We dug further. Twelve of the 47…
-
Loop Engineering Explained: How to Design AI Agents That Run Themselves
I had a multi-agent system that could handle complex store operations. It worked — when I was there to prompt it. Every morning I’d open a terminal, type the task, review the output, feed corrections back in, and repeat until the result was good enough. I was the loop. Then I went on vacation. The system sat idle for a…
-
Building Multi-Agent AI Systems: Patterns for Delegation and Collaboration
I had a single agent that handled store operations questions. It worked well — until it didn’t. A store manager asked: “Check which products are expiring this week, find alternatives from our supplier catalog, and draft a reorder suggestion for my regional manager.” The agent tried. It called the expiry tool, got 23 products, attempted to search the supplier…