An AI QA agent for the checks before every release
You write each end-to-end flow in plain language. BuniOD matches every step to your live page, adds the checks, and returns a scenario you confirm before anything runs.
What actually takes the days
The flow is rarely the hard part. Running it again is: the same clicks, the same screenshots, the same console lines copied into a note — before every release.
| By hand, before every release | With BuniOD |
|---|---|
| The steps live in a document that someone keeps up to date. | The steps are plain-language lines stored with the scenario and matched to your live page. |
| A person re-clicks the flow and watches the screen. | BuniOD runs the confirmed scenario in headless Chromium, on demand or once a day. |
| Screenshots and console output are collected by hand. | Run history keeps step and check results, screenshots, and console and network logs. |
| Tracing a failure back to a cause takes another pass. | Each failure comes with a plain-language explanation of which step or check failed and why. |
| A moved button breaks a script, and someone fixes the selector. | Self-healing element matching finds the element again and saves the new location after a passing run. |
No scripts, no selectors, no page objects.
Compare BuniOD with other toolsFive steps from a written flow to a run you can read
Connect
Paste your web app's URL. Public pages need only the URL; signed-in areas use HTTP basic auth or a test identity with its own inbox.
Write
Write the flow as plain-language steps with the expected result, the way you would explain it to a colleague who has never seen the app.
Generate and review
BuniOD's AI agent matches each step to the live page and returns a scenario with the checks attached. Adjust or regenerate it, then confirm it.
Run
Confirmed scenarios run in headless Chromium at a desktop viewport, on demand or once a day at a time you choose.
Inspect
Open the run for step and check results, screenshots, console and network logs, and a plain-language explanation of the first failure. The run stops there and does not retry itself.
Only confirmed scenarios are included in runs.
A flow in plain language
This is the whole authoring format. A person writes the steps; the agent turns them into a scenario you can review.
What you write
Sample data- 1. Click the 'Sign in' button
- 2. Fill in the email of the test user
- 3. Fill in the password of the test user
- 4. Click 'Continue'
- 5. Confirm the URL is /orders?success=true
What the agent adds
- The element each step refers to, matched on the live page.
- The checks that belong to the step — the URL, the visible text, the field value.
- A scenario you can adjust or regenerate before it is confirmed.
Example steps. Not a customer scenario, and not a result.
Six things you can open after a run
Step and check results
Every step and check in the run, in order, with its outcome.
Screenshots
A picture of the browser at the point where the run failed.
Console and network logs
What the page logged and what it requested, kept with the run.
Failure explanation
Which step or check failed and why, in plain language instead of a stack trace.
Self-healing record
When an element moved, the new location is saved after a passing run.
Business processes
Ordered groups of scenarios; if an earlier scenario fails, the dependents are skipped.
Two people own the release checklist. They need different things from it.
You keep the flow library honest. Confirm what the agent returns, read step results and logs when a run fails, and let self-healing absorb UI moves instead of a selector rewrite. Test identities give each flow its own inbox so the emails your app sends can be checked.
You write the flow in the language of the product, not of the DOM: click this, fill that, confirm the URL. The scenario comes back reviewable, and every failure is explained in words you can take to the next release meeting.
Invite teammates at no extra seat cost — there is no per-seat limit on any plan.
Where BuniOD stops today
These are the edges of the product as it exists now. Read them before you plan a release around it.
Desktop web only.
Runs use headless Chromium at a desktop viewport. Firefox, Safari and mobile browsers are not covered.
Native mobile app testing is coming soon.
It is in development and not available today.
One schedule: on demand or once a day.
Daily is the most frequent automated schedule.
Failed runs are not retried automatically.
The run stops at the first failure and waits for a person.
CAPTCHA can stop a run.
Use a test environment with CAPTCHA disabled, or a test key.
No code export.
Confirmed scenarios live in BuniOD, not in a repository.
Scenarios can create or change test data.
Use a dedicated test account and a non-production environment, and keep real customer data out of test identities.
Priced by the work the AI does — not by seats.
Metering is AI tokens. Projects, scenarios and team members are unlimited on every plan.
New runs and generation pause until the next period.
Work continues past the allowance.
Work continues past the allowance.
Every plan includes unlimited projects, scenarios and team members, daily scheduled runs, screenshots and logs, self-healing, business processes, and test identities with email checks.
A period is one month from activation. Input and output tokens are counted together, and unused tokens do not carry over. Paid plans are activated by email.
Questions we get
Is BuniOD an AI QA agent, or a script generator?
Does the agent find flows on its own?
Do I need to write code or install anything?
Which browsers does it test?
What happens when the UI changes?
Who confirms a scenario before it runs?
How is usage metered?
Write the flow once. Confirm it. Run it before the release.
Start with one sign-up, checkout or onboarding flow on your own web app. The Free plan covers the first scenarios without a card.
Desktop web apps only. Native mobile testing is coming soon.