Third Person
nicolas@thirdperson.run

Offboard your functional testing.

An outside agent uses your API the way a new developer would. It finds the bugs that return wrong answers without an error, measures how long setup really takes, and turns what it learns into tests that keep running.

Get a free outside test

Send us your docs link. You get a short report with every finding reproducible.

What we look for

Silent wrong answers

Paging that skips rows, totals that disagree with lists, settings that quietly disappear. Your users won't see an error, only bad data.

Setup and usability

Time from landing on your docs to a working call, every place a new developer gets stuck, and docs that don't match what the API does.

Regressions

The same checks re-run on a schedule. When a change breaks behavior, you get the exact request that shows it.

How it works

  1. We test from the outside.Using only your public docs and our own account, the same view your customers have. No code access, no SDK to install.
  2. We check every finding.A separate step reproduces each one before you see it, and you get the request, the expected result and what actually happened.
  3. You keep the tests.They're plain, deterministic code with fixed assertions. AI writes them, but no AI runs at test time.
  4. We keep them running.On a schedule today, and on your preview deploys when you're ready.

Why now

AI has made code cheap to write and hard to validate. Teams can open twenty pull requests a day, and when the same AI writes the code and its tests, the tests can share the code's mistakes. An outside tester that sees only how your service behaves is an independent check.

We tried it on our own production API first. Working only from the public docs, it wrote 135 tests in about 16 minutes and found four bugs that returned wrong answers without an error. Our own test suite had missed all four.

How we test other people's APIs