RAG evals and agent evals
By the end of today you can evaluate a retrieval pipeline stage by stage rather than end to end, and evaluate an agent on more than whether it eventually got the right answer.
YesterdayOn Day 76 you diagnosed a broken RAG system by hand. Today you make that diagnosis continuous and automatic.
TomorrowTomorrow, seeing these systems in production rather than in a test run.
Why this matters
End-to-end scores tell you something is wrong and never what. Evaluating each stage is what turns a vague quality problem into a specific fixable one.
- Evaluating RAG by stage
- Retrieval metrics
- Evaluating agents
- Trajectory assertions
Learn it
75 minCopy this into Claude or ChatGPT. It quizzes you before it explains anything, which is deliberate. The resources under it are how you check what it told you.
Today's Master Prompt
Free · sign inA prompt written for this day alone: your level, the exact scope, what to leave out, and an instruction to quiz you before it explains anything. Paste it into Claude or ChatGPT and it teaches you today's material.
Check it against something that is not a model
An assistant can be fluent and wrong, and on a topic you met today you will not catch it. These cover the same ground and were made by people who do this for a living, so they are what you hold the explanation up against. They are other people's work and we only link to them, so judge them for yourself.
3 hand-picked resources
Free · sign inVideos, official docs and articles covering the same ground, each opened and annotated by hand. They are what you check the assistant against on a day you cannot yet catch it being wrong.
Build it
55 minBuild a stage-level eval for your RAG system: at minimum retrieval hit rate and answer faithfulness, measured separately. Then evaluate your agent on ten tasks, five runs each, recording outcome, turn count and token cost. Deliberately break one tool and check the agent recovers.
Recall it
20 minAnswer out loud, reveal, then mark honestly whether you had it. That score is the only thing on this page you do not get to choose.
5 recall questions
Free · sign inQuestions you answer from memory, then grade yourself against the real answer. The score is carried into the mastery rating below it, so an honest miss cannot quietly become a tick.
Rate it
Completion and mastery are tracked separately. Be honest, because an inflated rating only means the concept resurfaces sooner.
Mastery tracking
Free · sign inRate yourself against five named criteria per concept. Completion and mastery are tracked separately, and anything you rate shakily comes back automatically on a spaced schedule.
Recap
- 01One end-to-end number says quality dropped and never where
- 02Retrieval evaluation needs ground truth, which is laborious and worth it
- 03An answer can be correct and unfaithful, which is luck rather than working
- 04Agents need several runs per case, or you are measuring noise
Your progress
Free · sign inMark days complete, pick up where you left off across devices, and watch completion and mastery diverge. Free, and the account exists only so ninety days of work cannot vanish with a cleared browser.