Complex task prompting strategy: strong model, hypothesis, loop, and encouragement
sergeykarayev · x · 2026-09-02
Author shares a strategy for prompting complex tasks:
- Use a strong model (e.g., Opus)
- Explain the problem and hypotheses
- Define the loop: log, change, measure, repeat
- Add a touch of encouragement at the end.
More from Apps
- HubSpot Cuts Video Editing Time by 90% Using Descript API — descript · 2026-09-02
- Fable 5.1 officially released: AI video generation tool gets update — kevinnbass · 2026-09-02
- Clone any image style in 3 steps using ChatGPT and JSON extraction — EXM7777 · 2026-09-02
- OpenAI dev confirms active work on Prism scientific writing tool — OpenAI · 2026-09-02
- ChatGPT-Epic integration ships in two modes: patient context in chat, chat inside the chart — thekaransinghal · 2026-09-02
- OpenAI Launches Epic EHR Integration and Healthcare Data Plugins — thekaransinghal · 2026-09-02