Uptime QA

Not just “is it up” — does the thing still work?

A page returning 200 tells you the server answered. It does not tell you the quote form still returns a quote. Uptime QA checks the task, not just the status code.

Two kinds of check

Simple up check

Fetches the page, confirms it is reachable and healthy, and optionally that it still contains text you expect. Returns in seconds. Good for “is it up?”

AI form check

An agent fills the form from your plain-English instructions, submits it, and judges the result against a success goal you write. A real browser drives the page, so it takes about a minute — and it catches the failure a status code cannot see.

You write the instructions in English, not a script

Say what to enter and which button to click. Any required field you do not mention is filled with a safe test value, so you describe the part that matters rather than every input on the page. Then write the success goal — how the AI decides pass or fail — in the same plain language.

On a schedule, with alerts that are not noise

Run manually, or every 5, 15 or 30 minutes, hourly, every 6 hours, or daily. Notify by email or Slack on failure, on status change, or on every run — so a flapping check does not train you to ignore it.

If the deployment has Slack or email switched off, the run still records its result and the alert is simply skipped. It does not fail the check to tell you about a missing integration.

Every run keeps its evidence

Open a test for its history: each run’s status, detail, HTTP code and duration — and for an AI form check, a screenshot of the page the AI actually judged. When a check fails you can see what it saw, rather than taking its word for it.

The Hub is ready. Your invitation is one sentence away.

Tell us what you are working on and we will open an account.

Request access