How do you catch it when an AI assistant is looping instead of making progress?
retained text
I’ve been using ChatGPT and Claude for a long software project, and the failure I keep hitting isn’t bad code. It’s that the thing quietly stops progressing while still sounding productive. Today’s example: over about 14 exchanges it committed to “verify the competitors and pricing before committing development time,” then never did. Instead it built the app, deployed it, and tested it. When I finally forced the question, the answer was that two established competitors already do the job — and it had actually written that down itself in the README on day one, then kept going. Nothing in any single reply looked wrong. Each one was a reasonable next step. You only see it by holding 15 messages in your head at once, which I clearly didn’t. So, honestly: * Does this happen to you, or am I just prompting badly? * When it does, how do you notice? Gut feel, or do you actually do something? * If you do something — what? A rule in your prompt, a checklist, re-reading the thread, starting fresh, something else? * Has it ever actually cost you real time or money, or is it just annoying?
document fetched 2026-10-13 00:19 · canonical https://reddit.com/r/AI_Agents/comments/1x3u5d1/how_do_you_catch_it_when_an_ai_assistant_is