AI agents that test software like real testers.
Oncomloops is an early-stage software company building autonomous QA tooling. Our agents crawl web and mobile apps, understand every screen with vision-language models, and turn what they see into executable, evidence-backed test suites.
Get in touchWhat we're building
Practical automation that replaces hours of manual QA work with verifiable, reproducible agent runs.
Test Discovery Engine
Autonomous crawling of web and Android apps. An LLM with vision reads each screen the way a human tester would, then emits structured test scenarios — happy path, negative, and boundary cases.
crawl → vision → scenariosAgentic QA Runs
Generated scenarios execute as real automation on Playwright and Appium. Every run ships with verifiable artifacts: screenshots, logs, and measured metrics instead of anecdotal pass/fail.
playwright · appium · evidenceCI Regression Pipelines
Suites plug into continuous integration so every build gets a full autonomous regression pass — with a human-readable report at the end, not a wall of noise.
ci/cd · reporting · gatesHow it works
One pipeline, three steps, fully autonomous after kickoff.
- Crawl. The agent explores the app — routes, forms, flows — and maps every reachable state.
- Understand. A vision-language model interprets each screen and reasons about what can break.
- Test & report. Scenarios run as executable automation, producing audit-ready evidence per run.