A small team,
one wedge,
and a product that
tests itself.
Qanthos is building agentic test automation for QA teams — the people who own quality but shouldn't have to write Playwright. This page is the honest version of what we're doing and why.
Collapse the QA gap.
QA today is stuck between two bad options: hand-written Playwright that only engineers can maintain, or brittle record-and-replay tools that break every sprint. Both waste QA capacity on plumbing instead of coverage.
Qanthos collapses that gap. QA describes the behavior. Agents own the plumbing. Humans triage results. The upside — for the industry — is QA teams that ship more coverage per week without adding engineers, and product teams that stop treating end-to-end tests as a cost center.
Every roadmap item we ship — the GitHub App path, the runner isolation, the generator work — is subordinate to one outcome: real QA teams running real regression suites through us daily, with the results they trust.
Three decisions that define the product.
Agentic, not recorded
Most tools in our lane are a recorder plus a selector engine. We're an agent that reads acceptance criteria, generates real Cucumber/Gherkin, and repairs it when the UI drifts. That difference compounds every sprint.
Managed on purpose
One repo per project on our GitHub organization via a GitHub App. One runner path per tenant. Zero personal access tokens exposed to the customer. Structurally eliminating that class of pain is a real product surface, not a nice-to-have.
Self-tested, end to end
Every Qanthos release ships only after our own agent-generated suites run green against our platform — on the exact infrastructure we deliver to customers. The fastest signal we can have, and the same bar we hold our customers' coverage to.
Pre-market. Onboarding design partners.
We're not going to fake a customer logo wall. We're pre-market. The platform runs on dev trunk, our own UI regression suite is green, and our API regression suite is partial while the generator fills in the remaining scenarios. That's the real state, October 2026.
What we want between now and end of year is small and specific: a few real QA teams running real regression suites through us daily, with the results they trust. If that's a fit, we'd love to talk.
Design partners get direct access to the platform team, hands-on onboarding, and disproportionate influence on the roadmap.
- ✓ Internal UI regression greenEvery release run on managed infra
- ◐ Internal API regression partialPlaceholders being replaced by real generation
- ◆ GitHub App path in flightManaged-git per project · ERGA-474
- ○ BYOR deferred to enterpriseBring your own repo · post-launch
Design partners wanted.
If your QA team is stuck between hand-written Playwright and brittle record-and-replay, we should probably meet.