Testing that
proves itself.

Describe a GUI test. Get a recorded run and an independent verdict you can inspect.

It starts with a sentence

Write the steps and what should happen. Ryvetz finds each control on the live screen.

Explore Flows

Invite a teammate and confirm the seat

Executor actions
  1. Open Settings › Members

  2. Invite dana@example.com as an editor

  3. Send the invite

    Judge criterionThe members list shows dana@example.com as Pending within 10 seconds

  4. Revoke the invite

    Judge criteriondana@example.com is no longer listed within 10 seconds

View YAML
name: Invite a teammate and confirm the seat
app: acme-crm
account: staging

steps:
  - do: Open Settings › Members
  - do: Invite dana@example.com as an editor
  - do: Send the invite
    verify: The members list shows dana@example.com as Pending within 10 seconds
  - do: Revoke the invite
    verify: dana@example.com is no longer listed within 10 seconds
Illustrative example

A verdict that cites pixels

A separate judge sees only your criterion and the screenshot, then cites what supports its verdict.

Explore the Judge
The screenshot the judge saw Step 03 · Acme CRM
The example product, Acme CRM, on its Settings › Members page after an invite was sent: dana@example.com is listed with the status Pending, and the avatar beside the name failed to load.

The status the criterion asks about: Pending.

Also on this screen: an avatar that failed to load.

Judge · step 03 Passed

Judge criterionThe members list shows dana@example.com as Pending within 10 seconds

“dana@example.com is listed in Settings › Members with the status Pending.citation1”

Judge verdict · Acme CRM · illustrative example Judged from one screenshot, with one citation. The runner keeps checking for up to 10 seconds; this is the frame that passed.
Open full-size original (opens in a new tab)

What each side was given

A pass the judge is not confident in is recorded as a fail. Cheaper checks, such as an exact text match or a known-good replay, can settle a step before the judge is asked.

Executor received

  • The live screen
  • Step 03: Send the invite

Judge received

  • The criterion
  • One screenshot

Never shown to the judge

  • The executor's reasoning
  • Its logs
  • Its own claim of success

Evidence you can inspect

Replay the video, inspect the trace, and read the verdict. Team adds pull-request checks.

Explore Evidence
Ryvetz run view of the example product Acme CRM: recorded application video, passed steps, an event timeline, and findings. Illustrative example.

Run overview: an illustrative capture of the example product, Acme CRM.

Open full-size image

Guardrails for real accounts

Shared runners accept test, sandbox, and demo accounts only, with session guards enforced by the runner. Accounts with real money behind them, such as funded or live trading accounts, require dedicated Enterprise runners.

Explore runner safeguards
Nothing to install
You author flows in the browser or from the CLI. The runner, the browser, and the platform profile are ours to keep working.
Test accounts only
Credentials are encrypted at rest, injected only for the length of a run, and never shown back in the UI.
Dedicated when it matters
Accounts with real money behind them, such as eval, funded, and live trading accounts, belong on a single-tenant runner under Enterprise. Never on a shared slot.