Ship more,
fear less
A QA pass on every push.
SuperGorilla uses agentic AI to run your app and test it the way a real user would, on every push or only when you ask. Every run comes back with video, screenshots, console logs and steps to reproduce. Nothing to write. Nothing to maintain.
Writing the code stopped being the bottleneck
Your agents can write a feature in an hour. Finding out whether a person can actually use it still means someone clicking through the app. It’s the slowest step left in your pipeline, and it’s the one nobody automated.
From push to report, without you in the loop
01
You push
A commit lands on a branch, from you or your agents.
02
We get your app running
Virtually any web app. Nothing for you to deploy or stand up first.
03
Agents use it like users
They work through what changed, plus the flows your business depends on.
04
Evidence lands on the PR
Video, screenshots, console logs and repro steps, posted straight back to the pull request.
The longer version, covering what we need from you and what happens when the app won’t start, is on How it works.
Not a red X. A recording.
Every run reports back like a colleague would: what it tried, what happened, and the evidence: video of the whole session, a screenshot at every step, the console errors, and exactly how to reproduce the failure. Judge any finding in twenty seconds without leaving the pull request.
supergorilla-aibotcommented 4 minutes ago
SuperGorilla ran a QA pass on this pull request. 3 of 5 steps passed · 3m 41s
Step 4: What broke
Order total renders as £NaN and the pay button stays disabled. Reproduce: add any item → checkout → apply the WELCOME10 code.
TypeError: Cannot read properties of undefined (reading 'total')
at CheckoutSummary src/checkout/summary.tsx:114Step 5: What broke
The sidebar layout breaks when collapsed: navigation items overlap the content area and the settings link becomes unreachable. Recording attached.
New changes, and the flows you can’t afford to break
User flows touched or added in code changes
Derived from the change itself. The agent reads what shipped, works out which user flows it touches, and tests them the way a customer would, on every push, before anyone has to look.
Protect critical user flows from regression
Record a flow with SuperGorilla once: walk through it, and it tests that flow automatically on every change. Signup, auth, checkout: covered without anyone remembering to check.
Coming soonYou choose when it runs
You control how often it runs, so you control the bill.
On push
Runs on every push to a pull request, or whenever the branches you select change. Feedback while the change is still in your head, not after you’ve moved on.
On demand
Ask SuperGorilla for a test right from GitHub, the same way you’d ask a teammate, and it answers by using the app.
One push is one test, however many flows we find.
SuperGorilla lives where you work
The report lands in GitHub, Linear, or Slack. When a run needs information or your approval, SuperGorilla messages you in Slack or Linear, the way a teammate would. No new dashboard to babysit.
Nothing to write. Nothing to maintain.
- ✓Virtually any web app
- ✓No test suite, no selectors
- ✓No staging URL to point at
- ✓Nothing breaks when the UI changes
- ✓Steer it like you’d brief a new tester
- ✓Guardrails the agents never cross
Day one it tests your app. Day ninety it knows it.
SuperGorilla remembers every run: the flows that matter, the gotcha in your onboarding, the step that always needs a retry, the thing that broke last quarter. Each test makes the next one faster and quieter.
L1
Organisation
Your conventions, across everything you ship
L2
Project
How this product hangs together
L3
Repository
The gotchas that live in this codebase
L4
Branch
What this line of work is trying to do
L5
Pull request
What already passed, what broke, what to retry
The sooner it starts, the more it knows.
It tests your app. Here’s how we treat it.
Granting a tool access to your repository and your app is a decision about trust. The security page covers exactly what SuperGorilla can see, how test credentials are stored, how long recordings are kept, and what memory retains.
- ✓Bring your own API key on Enterprise: model traffic runs through your own provider account
- ✓Test credentials stored encrypted; dedicated test accounts recommended
- ✓Recordings and screenshots kept for you, purgeable on request
- ✓Repo access scoped to reading the change and posting the report
- ✓Guardrails bound every run: pages, actions and data the agents never touch
Flat plans, plus usage.
Starter
£50/mo + usage
- ✓3 concurrent runs
- ✓Up to 3 members
- ✓30 tests included / mo
- ✓Email support
Team
£100/mo + usage
- ✓10 concurrent runs
- ✓Up to 10 members
- ✓100 tests included / mo
- ✓Email support
Enterprise
Talk to us
- ✓Custom concurrency & members
- ✓Custom test volume + usage
- ✓Bring your own API key
- ✓Slack support with SLA
- ✓Onboarding help & custom integrations
Beyond included tests, usage-based pricing applies. You control when tests run, so you control the bill. Full pricing →
The six things people ask first
What does it cost?+
Plans start at £50/month with tests included, plus usage-based pricing beyond those. Enterprise plans can bring their own API key.
Will it work with my stack?+
Virtually any web application. SuperGorilla runs your app and works out how to use it, and if it needs a hint, you steer it in plain English.
What can you see in my repo?+
Access is scoped to what a run needs: reading the change and posting the report back. The full breakdown lives on the security page.
How does it log in to my app?+
With test credentials you provide, stored encrypted. We recommend dedicated test accounts, and cover the details during onboarding.
When do mobile and desktop land?+
Web is live today. iOS, Android and desktop support are in development; book a demo and tell us which one you need.
How do I get started?+
Click Get started and connect a repository. No waitlist. SuperGorilla learns your app with every run, so the sooner it starts, the more it knows.
Longer answers, Playwright comparisons included, on the FAQ page.