loops.guide
the guide

Loop patterns

These are common ways to set up an AI that checks its work. Start with the smallest one that fits your task.

pass fail: retry Start Model generate draft check output? checking rule Done
Retry Loop low cost low risk

Let the AI try, check if it worked, and try again if it did not.
plan valid feedback / revise Start Plan generate Checker check Done
Plan and Check Loop medium cost medium risk

Make a plan, do one step, check it, then continue.
Start Model make options Path A Path B
Try Options Loop high cost medium risk

Try a few possible answers, compare them, and keep the best one.
Start Model Ask person Done
Human Approval Loop medium cost low risk

Let the AI work, but ask a person before important or risky steps.
feedback: not good yet pass Worker produce output Checklist checker good enough? Ship
Checker Loop high cost low risk

Use a separate check to decide whether the AI's work is good enough.
next event wait new item / schedule Queue Agent do the work Update save result Idle
Triggered Loop medium cost medium risk

Start the AI when something happens, like a new issue or a daily schedule.
past runs improve the setup review History past runs Analyze better setup? Update saved change Measure
Improve Over Time Loop high cost medium risk

Use past runs to make the whole AI setup better.