AI designs your test strategy
From requirements, walkthroughs, defects and legacy assets.
Test suites that grow and heal themselves.
Meet Curator, the AI test team lead for enterprise QA.
Curator designs your test strategy, builds the automation and heals tests when your application changes. Everyday runs use saved replays, so there are no token costs to run your suite.
Lead agentCurator
CoversWeb, desktop and mobile
Cost to runZero token costs
Upkeep0.1 FTE for 598 scenarios
Watch it work
One request,
One request,
start to finish.
Under three minutes: Curator checks a requirement against your suite, designs the test with your approval, writes the scripts, runs them, and heals a step when the page changes.
Sound familiar?
Four things every
Four things every
QA lead tells us.
Each one is a job Curator takes off your team.
“Every release breaks half our scripts.”
Curator heals broken tests“We have 1,000+ scripts and nobody knows which test does what.”
Curator tracks every test“We can't hire automation engineers fast enough.”
Curator builds the automation“AI testing sounds great until you see the model bill.”
Zero token costs to run
People move on. The knowledge stays.
Suites grow with every engineer, and the know-how walks out when they’re promoted, move teams or leave. Curator tracks every test, what it covers and why, so handovers stop being a scramble.
Meet Curator
One lead agent.
One lead agent.
A specialist for every job.
Your whole QA team works with the same Curator, the same test models and the same rules. Behind it, specialist agents design, write, match and maintain your tests.
You bring
- A requirement
- A test case
- A defect
- A maintenance request
Curator
AI team lead
Checks every request against the suite you already have.
You get back
- Clear answers on what's covered
- Test models to review
- Gaps found early
- Tests kept up to date
- Automation, with less effort
What Curator does
Design, build, heal,
Design, build, heal,
evolve. Repeat.
Curator and its specialist agents cover the whole life of a test suite, from the first strategy to the tests you retire.
AI builds your automation
End-to-end, across web, desktop and mobile.
AI heals broken tests
Automatic repair when your application changes, so runs carry on uninterrupted.
AI evolves your test suite
Creates, optimises and retires tests on its own, so the suite stays lean.
Zero token costs to run
Saved replays run your suite, so there are no usage-based AI charges as it grows.
0.1 FTE
to maintain 598 test scenarios. One person, half a day a week.
Reuse, extend or create
Nothing gets built twice.
For every request, Curator first looks at what you already have, then makes one decision.
Reuse01
The requirement is already covered by an existing model. Nothing new is built.
Extend02
An existing model fits, so new scenarios are added to it.
Create03
No suitable model exists, so a new one is designed.
Maximum reuse · No duplicate test design · Less time on maintenance
How a request works
Curator proposes.
Curator proposes.
You decide.
Every request runs through the same five steps.
- 01
You ask
For a requirement, a test case, a defect or a fix.
- 02
It analyses
Checks the request against your existing models.
- 03
It proposes
A plan you can read before anything happens.
- 04
You confirm
Challenge it, correct it, then say go.
- 05
It builds
Models, scenarios and scripts, kept up to date from then on.
Nothing is created or changed until you confirm.How thorough: BasicStandardAggressive
Client stories
Already running
Already running
for our clients.
Names kept private. What each team was dealing with, what we built, and what changed.
Workforce management software · QA automation client
A day of upkeep, down to 2 hours.
Curator runs and maintains the suite. One engineer just reviews.
The challenge. After every regression run, an engineer spent a whole day reading test reports, tracing failures and making fixes to keep 598 scenarios healthy.
598test scenarios automated, across every module
2 hof one engineer’s time, down from a full day
Weeklyfull regression runs, analysed by AI
<5 minto diagnose a failure, down from 15–20
AI diagnoses. People decide.
- 01AlertEvery failure posts to Slack with the failed step and a log link.
- 02DiagnoseAI reads the script, data, logs and screenshot to find the cause.
- 03FixIt suggests a fix. A QA engineer approves it first.
Human-in-the-loop by design. Faster answers, and nothing changes without sign-off.
Safety app · Web, Android and iOS
95% less manual design effort.
From prompt to productivity, powered by Curator.
What we did. Curator built the app’s workflows and designs from prompts, then automated end-to-end regression across web and mobile, in development and staging, with runs the client can trigger themselves.
95%less manual design effort, through Curator
70%less manual testing effort
299regression scenarios automated
3platforms validated: Android, iOS and web
Prompt in. Tested app out.
- 01PromptWorkflows built from prompts, with no traditional scripting.
- 02DesignCurator generates the designs, so people review instead of drawing.
- 03TestE2E runs on real Android and iOS devices, offline and QR flows included.
Also on your list
See enterprise automation
Enterprise automation
The same agents and the same five-step loop, pointed at your business processes: bots that move data, read documents and handle routine work at scale.
Book a demo
Tell us what you need tested.
We'll show you Curator building and maintaining tests, then talk through how it would fit your suite. We answer every request inside one working day.