Ai Evals

1---2name: ai-evals3description: Help users create and run AI evaluations. Use when someone is building evals for LLM products, measuring model quality, creating test cases, designing rubrics, or trying to systematically measure AI output quality.4---5 6# AI Evals7 8Help the user create systematic evaluations for AI products using insights from AI practitioners.9 10## How to Help11 12When the user asks for help with AI evals:13 141. **Understand what they're evaluating** - Ask what AI feature or model they're testing and what "good" looks like152. **Help design the eval approach** - Suggest rubrics, test cases, and measurement methods163. **Guide implementation** - Help them think through edge cases, scoring criteria, and iteration cycles174. **Connect to product requirements** - Ensure evals align with actual user needs, not just technical metrics18 19## Core Principles20 21### Evals are the new PRD22Brendan Foody: "If the model is the product, then the eval is the product requirement document." Evals define what success looks like in AI products—they're not optional quality checks, they're core specifications.23 24### Evals are a core product skill25Hamel Husain & Shreya Shankar: "Both the chief product officers of Anthropic and OpenAI shared that evals are becoming the most important new skill for product builders." This isn't just for ML engineers—product people need to master this.26 27### The workflow matters28Building good evals involves error analysis, open coding (writing down what's wrong), clustering failure patterns, and creating rubrics. It's a systematic process, not a one-time test.29 30## Questions to Help Users31 32- "What does 'good' look like for this AI output?"33- "What are the most common failure modes you've seen?"34- "How will you know if the model got better or worse?"35- "Are you measuring what users actually care about?"36- "Have you manually reviewed enough outputs to understand failure patterns?"37 38## Common Mistakes to Flag39 40- **Skipping manual review** - You can't write good evals without first understanding failure patterns through manual trace analysis41- **Using vague criteria** - "The output should be good" isn't an eval; you need specific, measurable criteria42- **LLM-as-judge without validation** - If using an LLM to judge, you must validate that judge against human experts43- **Likert scales over binary** - Force Pass/Fail decisions; 1-5 scales produce meaningless averages44 45## Deep Dive46 47For all 2 insights from 2 guests, see `references/guest-insights.md`48 49## Related Skills50 51- Building with LLMs52- AI Product Strategy53- Evaluating New Technology