Step 3: Let GPT-5.5 run. It will work through the plan methodically. The key insight: GPT-5.5 is more confident when given explicit instructions from an external plan than when it generates its own. It stops second-guessing. It stops being lazy. It builds.

→ View original post on X — @godofprompt