about

Evaris is an evaluation platform for AI agents. You author an eval, run it against your agent, and publish the log. Evaris stores the raw artifact, derives structured records, and renders every run in a web UI: status, warnings, samples, scores, and transcripts.

Evals are written in Python with Inspect AI, the open-source evaluation framework from the UK AI Security Institute. Your system under test can be written in any language; evals call it over HTTP. Evaris is the storage, ingestion, and review layer around those runs.

We are a small team, currently in private beta. Questions, feedback, or early access: roshan@evarisai.com.