Release gate: an eval suite that blocks bad releases

The answer to "how do you know it works" is a number that is checked before every release.

The brief

The policy assistant from an earlier stage is about to be used by the whole company. The HR lead asks for proof that each update does not make answers worse.

What you will build

  • A golden set of at least 30 real questions with expected behaviour
  • Automated scoring for correctness, citations and refusals
  • A check that fails the build when the score drops
  • A one-page report a stakeholder can read

Requirements

RequirementDone when
Golden setBuilt from real questions, including ones that must be refused
ScoringRuns without a person, with the scoring method written down
GateA deliberately broken prompt fails the check
ReportShows the score over time and the worst failures

What to learn it from

Milestones

  1. Collect the questions and agree expected behaviour
  2. Write the scoring
  3. Run it once and record the baseline
  4. Add it to CI
  5. Break the prompt on purpose and watch it fail
  6. Write the report
How it is judged
Introduce a regression on purpose. If the gate lets it through, the project is not done.

Stretch goals

  • Check the judge against human labels
  • Track cost and latency alongside quality
  • Alert when production answers drift

Before you call it done

Worth remembering
  • The threshold was agreed, not invented
  • A regression was caught for real
  • A non-engineer understood the report
Back toAll frameworks

Every expert started right here.