Use cases
Hotato finds what broke in your agent calls and pins it so it never ships again: local call forensics and regression guards, run on your own machine. Each use case below starts from one concrete failure, shows the loop solving it with commands you can run, and ends at a reproducible gate. Every page links to the docs and to the related use cases, so you can follow the whole path.
Voice agent testing
A voice agent can read a clean transcript and still fail the call: talk-over, dead air, wrong tool, a claimed action that never happened. Measure the call itself, then keep each catch.
Read the use case →CI for AI agents
Agents regress quietly when a prompt or model changes and nothing goes red. Pin a caught failure into a contract and gate every change with a deterministic, offline check.
Read the use case →Related writing: what your voice test suite still measures, the transcript passed, the call failed. Docs start at getting started.
Every use case starts with one call:
hotato autopsy ./call.wav