CURN
AI playbook
Write down one task so anyone on the team can run it with AI and check the result.
- Task
- Owner
- AI tool
- Last checked
-
Goal
The specific result the workflow is trying to produce.
- What does a good result look like?
- Who uses it, and when?
-
Context
The source material, constraints, audience, and task setup that shaped the result.
- Which documents or data go in, and where do they live?
- What must stay out of the AI tool?
-
Workflow
The exact prompts, sequence, model choices, and edits that got to a strong output.
Step Person or AI Prompt or instruction Edit or check 1 2 3 4 -
Checks
The review criteria that decide whether the output is good enough to use.
- What does a person check before it goes out?
- Which mistakes come up most?
-
Outcome
The final result, plus enough evidence to show that the workflow genuinely worked.
- Link a real example of the output.
- How long did the task take before, and how long with the playbook?
-
Health
A way to see whether the playbook still works as tools, models, and inputs change.
- When will someone run it again to check?
- What would make you change or retire it?
How to use it
- Start from a real task someone already does with AI, not a hypothetical one.
- Fill it in from the run that produced a strong result, not from memory.
- Ask a colleague to follow it and compare their result with yours.
CURN · curn.io/playbook-template