Field note
Starter AI eval sheet
25 representative test cases to check an AI workflow does its job before launch.
- Who it is for
- Teams writing their first evaluation set for an AI agent or LLM workflow.
- How to use it
- Fill in the cases, then reuse the sheet for each release decision.
- Format
- Field note, PDF, 6 pages.
What is inside
- 25 representative test cases
- Expected behaviour and pass criteria
- A reusable release-decision worksheet
Goes with the article How to evaluate AI agents and RAG systems before production breaks them.
Browse the research library
