Evals

Agents are fluent. Fluency is not the same as safe. If you are rolling AI agents into a real org, you already know the demo version of this problem. The model sounds careful. The slide says “guardrails.” Someone writes “do not touch production” in a prompt, and the room relaxes. That sentence is a sticky note. It is not a control. When the agent is in a hurry, or the chat got cut short, or it found a clever way to spell the same dangerous action, the sticky note is already gone. ...

This is part 1 of the Agent production system series. Start with the system map if you haven’t read it yet. On the interactive diagram: Eval gates. Demo agents look smart until they touch a real repo. In a demo, the agent writes the code, the code runs once, everyone claps. Nobody checks what happens the second time, or the tenth, or the time the agent decides the fastest way past a failing test is to delete the test. Then you learn the hard lesson: intelligence without a gate is just a faster way to ship a bad change. ...