Nexus StudioNexus Studio for leaders: what it does, and what you see in week one.
A short read for the person who decides, with no QA background needed. Three situations, each with the pain, what the product does today and what the first week looks like.
What do you need before you read on?
Nexus Studio is a Windows 11 desktop app and needs 16 GB of RAM as the minimum. It runs on your own machine, so your tests, results and evidence stay there. It is Windows only, real phones are Android only, and there are no cloud runners today. Last updated: .
A team that already has QA
The pain. Your testers already run Playwright, Selenium or pytest, and every red build costs someone a morning to sort out. A real defect, a flaky test, a slow network, a bot challenge and a crashed browser all look the same.
What it does today. Nexus Studio runs the tests you already have, and the Test Adapter imports the ones written elsewhere. Every run gets one of 14 verdicts, so a failure caused by the environment is not reported as a bug in the application. Each failure files a report, a screenshot and logs, a real failure keeps its video, and records can be sealed in a SHA-256 chain of custody. History ranks the risky tests, and the Executive Report gives a release call of go, caution or no-go. The tests stay plain Playwright, Selenium and pytest files, so nothing is locked in.
What you see in the first week. Install on one Windows 11 machine, import or record a few tests and run them on Chromium, Firefox and WebKit. You will see one verdict per browser, a folder of evidence for each failure in the Evidence Locker, and a History row for every run. Put your own numbers into the calculator to estimate the hours this work takes today.
A startup with no QA yet
The pain. Nobody owns testing. Releases rely on clicking through the app by hand, and no one has time to learn a test framework before the next deadline.
What it does today. The Nexus Recorder writes Playwright or Selenium code as you click through your app in a real browser, with no AI in the loop, and grades every selector with fixed rules. Tests run on your own machine on three browsers and on real Android phones. Watch re-runs a selection on a timer and alerts Slack or Telegram only when a state changes, and the CI/CD screen writes the GitHub Actions or GitLab CI file. The trial lasts 30 days with no card and every feature. The Individual plan is 1 machine with unlimited users and runs.
What you see in the first week. Record a first test, rehearse it in the sandbox, where nothing is written to evidence, and then run it for real. After three rehearsals it is ready for its first real run. By the end of the week a few recorded tests can be running on a schedule with alerts. How fast that goes depends on your app and your machine.
Building with AI (vibe coding)
The pain. When an AI assistant writes much of the code, nobody has read every change, and the tests it writes can pass without proving anything.
What it does today. The AI writer drafts tests with the AI tool you already use, such as Claude Code, Codex or opencode, or with a model: a local Ollama model by default, Anthropic or OpenAI when you choose. Every draft passes the Vet, seven checks without AI that include assertions that can fail and no sleeps or forced clicks. It then rehearses in the sandbox, and nothing is saved until you accept it. Self-healing proposes a selector fix but leaves the run red until you choose, and a retry that passes is FLAKY, never a quiet pass.
What you see in the first week. Point the AI writer at one flow, read the Vet result, rehearse the test and run it for real. You see which AI-written tests hold, which are FLAKY and which fail for a real reason, each with its evidence. Your own AI plan and its limits still apply: the Studio adds none of its own.
Try it on your own app.
Free for 30 days, with every feature, on your machines. No card needed.