TestMu AI Launches Agent Assurance for AI Agent Testing

TestMu AI has launched Agent Assurance, a product for testing AI agents before release, the company announced in a press release. The product covers conversational agents across chat, voice, phone, video, and image, as well as autonomous agents that call tools, write files, use APIs, and open pull requests.

Agent Assurance generates test scenarios from a codebase and invokes the agent to check actual behavior. It grades results using observed evidence such as changed files, produced artifacts, and verified tool calls.

The product reports pass and fail results, plus an unable to verify result for cases where evidence is missing. TestMu AI calls this number the assurance gap and excludes it from the pass rate.

The conversational agent category is generally available on the TestMu AI platform. The autonomous agent category is in early access and runs from the terminal as rook, with models executed in the TestMu AI controller.