Unity Evals

Monitor quality and agent performance.

Domain analytics, golden-SQL execution scores, and negative feedback — the same agent-monitoring views your team already uses to keep self-serve answers trustworthy.

Domain analytics

Questions, unique users, success rate, and satisfaction for each domain — with charts over the window you pick.

Execution score

Score generated SQL against golden tests: execution match, row count, column match, and value match.

Trends over time

Watch pass rates recover after you add metadata — then export every failed question from a run.

Negative-feedback review

See which topics miss (joins, aggregations, ordering) and open the interaction that got the thumbs-down.

Score the agent against golden SQL

Run evaluations and see pass rates for execution match, row count, column match, and value match — then improve the score from the database view.

Watch quality recover over time

Track execution, row, column, and value match across runs. Group history by domain or tag, and export every failed golden question.

Review the questions that missed

Negative feedback is grouped by topic — Aggregation, Join, Ordering — with the user’s comment, database, and a path back to the interaction.

Ready to make self serve analytics a reality for your company?

Deploy Unity Layer AI agents and empower your business and data teams with fast and accurate AI insights

Get Started