An internal QA tool for a company in a highly regulated industry. It automatically scores every AI support chatbot conversation on three dimensions, routes uncertain verdicts to a human review queue, and surfaces trends in an analytics dashboard.
Separate upload from evaluation. Judge against real source material.
AI support chatbots in regulated industries can produce wrong answers, and the only way to know is to read every conversation yourself. There is no systematic way to measure quality degradation, catch a hallucination before it reaches a user, or give a QA team anything actionable. The chatbot ships; monitoring is hope.
Build a two-stage pipeline that separates upload speed from evaluation speed. Upload parses and queues immediately, returning 202 Accepted in under a second, no AI in the loop. Async cron handles evaluation: embed the question, retrieve the relevant knowledge base articles from the same RAG index the chatbot uses, and have Claude score the answer against actual source material, not general world knowledge. Verdicts feed a human review queue; the review queue feeds aggregate analytics.
- RAG-grounded judge: judgment rests on actual documentation rather than world knowledge
- Auditable thresholds: the verdict formula lives in application code, so they can be adjusted without a prompt engineering session
- Harmless as an independent gate: an answer that scores well overall but poorly on harmless fails regardless, and no weighted average overrides it
- Human review queue with accept or reject and reviewer notes
- Analytics dashboard in Recharts: daily, weekly, and monthly volume, resolution rates by question category, peak hour, and week-over-week trends
Upload to verdict in under 30 seconds per pair. The QA team gets a scored, reviewable record of every conversation instead of sampling from a queue and hoping the sample is representative.
The distinction between “did the chatbot know this” and “is this generally true” is the whole point of RAG-grounded evaluation. An LLM judge without retrieved context applies world knowledge, which is the wrong standard for a chatbot that only covers one company's documentation. Grounding the judge in the same source of truth as the chatbot is what separates a useful eval tool from a confidence score nobody can act on.