NewCompare models in chat agent tests.See the benchmark

Voxli tests your chat agent with simulated conversations to catch failures before release.

14 days. No credit card.

Built with our design partners

  • Kindly
  • Expertise.ai
“I can create tests, see what went wrong, tweak them - that would have taken hours before.”
Ryan Hoffman, Expertise.ai Read the story

Test conversations from start to finish.

Test with users who ask in different ways and languages. Review failed checks alongside the conversation to understand what went wrong.

See how conversation tests work

Let Voxli AI write the tests.

Voxli AI explores your connected agent and creates editable tests with instructions and pass criteria. Reuse them to check your next change.

Explore automatic testing

Turn real conversations into regression tests.

Explore conversation analysis
  1. Import

    Send real conversations through the API and find issues to investigate.

  2. Reproduce

    Ask Voxli AI to try reproducing an issue with your connected agent.

  3. Save a regression test

    Ask Voxli AI to create a test you can rerun after changes.

Compare versions before you release.

See what improved and what needs fixing before release. Compare runs the same tests on both versions. Add repeats to see how results vary.

Explore Compare

Monitor your agent with scheduled tests.

Run simulated conversations on a schedule and get Slack or email alerts for failures at or above your chosen severity.

Explore scheduled monitoring

Use cases

How teams use Voxli.

Connect your agent without changing it.

Test through your chat widget, your agent's HTTP endpoint, GitHub Actions, or the Voxli CLI. A founder can help with setup.

See connection options
Mattias LagergrenBjörn Rydahl

Run your first conversation test.

Try Voxli free for 14 days. Book a call if you want help getting started.

Start free trial See pricing