Looking for testers: AI-native E2E test generator for microservices (free, early access)

Discussions
, , , ,
1

Hi MoT,

We’ve just finished the initial version of CratonAI and we’re looking for people with real microservice test suites to try it and tell us where it breaks. We built it because hand-writing and maintaining cross-service E2E tests (stitching data across services, keeping everything in sync as APIs change) was eating too much of our own time, and we’re guessing it’s eating yours too.

What it does — you provide:

  1. Your Swagger/OpenAPI specs — one per service involved in the flow
  2. Service info — e.g. base URL, auth requirements, dependencies on other services, exposed endpoints, and (for async flows) the Kafka topics/events it publishes or consumes
  3. The business flow, in plain language — e.g. “user signs up → payment service charges the card → notification service sends a confirmation”
  4. Data rules for each service’s input/output — e.g. field formats, required values, dependencies between services

You get back: a deployable E2E test suite packed in a docker image, ready to run inside your own infrastructure. We never touch your services or your sensitive data.

This is genuinely an early access build, not a finished product, and onboarding right now is hands-on rather than self-serve. No signup flow yet, so just a direct handoff while we build out proper self-serve access.

It’s completely free to try. If you work with microservices and are willing to point it at a real (or non-critical) flow, we’d love to get you set up. What we actually want and really appreciate is honest feedback like what’s broken, what’s missing, whether the output is usable in your CI, and it’ll directly shape what we build next. “It didn’t work for my case” is just as useful to us as “it worked.” We put more detailed information in our website: https://craton-ai.com/.

Separately, genuinely curious: for those of you testing microservices day to day, how are you currently handling cross-service E2E coverage — hand-written suites, contract testing, something homegrown, or mostly hope?

Happy to answer questions here too, including the hard ones about where this currently falls short.

Thanks,
Tony