Rather than iterating back and forth with an agent, have it build tools to iterate: many variations at once, several theories checked together.

Added to korrents 25 Oct 2026 · How quotes work · Something wrong? Tell us

Shared subject first, then similar wording. They may agree or disagree.

• An agent repeats its own history: whatever it did on the last change — ran the tests or skipped them — is what it will do on the next one, because it is predicting the next message in the conversation. DHDex Horthy15 Jul 2026
• The goal is not a better prompt but your own absence: arrange it once, hit go, and keep agents running for long stretches without you. AKAndrej Karpathy20 Mar 2026
• If you want loop engineering, build one loop at a time and keep them small: every part of the agentic-factory playbook is good advice except the part where you stop reading the code. DHDex Horthy15 Jul 2026
• Seeing each version an agent pipeline produces, instead of a black box, shows where the process is too weak. KZKyle Zantos11 Aug 2026
• When an AI agent runs out of forward progress, throw the project away and start over rather than trying to tweak it. KBKent Beck1 Jul 2026

See this claim on the map