FAQ
Short, direct answers to the questions buyers ask us most: data and security, integrations, ownership, pricing and the pilot. If your question is not here, ask us in a 30-minute call.
Requirement text, the relevant code snippets and read-only data samples, only for the AI steps that need them. Secrets do not go to the model, and the CI runner works without internet access and without secrets, so no AI runs in CI.
Personal-data masking in the model gateway (national ID, IBAN, phone, e-mail) is available and must be switched on whenever customer data is processed; pilots can run on synthetic or non-production data.
Today, Anthropic Claude through the Anthropic API. Whether your data may be used for training is governed by the provider's commercial terms. Model access goes through a swappable model gateway; other providers such as Azure OpenAI are on the roadmap.
In your environment, as containers. Today it is installed with Docker Compose on a single Linux server next to your Jira and repository.
Not today. Users sign in with platform accounts and role-based access control (five roles). If SSO (OIDC/SAML) is a requirement, raise it early so we can scope it as a pilot condition.
Not yet. Today AI Testing Probe works with Jira, Xray, GitHub and GitHub Actions. Azure DevOps (Boards, Repos, Pipelines, Test Plans) is on the roadmap.
Not today. The product generates Playwright tests for web applications. Desktop (FlaUI) and mobile (Appium) code generation is on the roadmap. Antrek Labs delivers desktop and mobile test automation as a service today; the product does not generate this code yet.
The product generates new tests from your stories; it does not migrate existing suites today. Available today as an Antrek Labs service.
You do. Tests are plain Playwright TypeScript in your repository, and approved test cases are published to your Jira or Xray. If you stop using AI Testing Probe, everything keeps working without it.
You can generate test code with those tools, and many teams do. What AI Testing Probe adds is the governed flow around generation: every AI step waits for a named expert in Approval Gates, decisions go into a hash-chained audit trail, and each generated test runs against a deliberately broken build; one that misses the fault needs explicit, logged sign-off to merge (Break-to-Verify).
Code is filled into your templates and page objects instead of being written freely, failures are triaged into reviewed repair pull requests, and no AI runs in CI. Building and maintaining that flow yourself is possible; it is the part that takes the most time.
Usually a fixed-scope pilot, followed by a project or platform licence. We do not publish prices; we agree on them after the fit call.
[Pricing model wording, to be approved by our commercial team]
A pilot covers one application and a few real stories, with your analysts and engineers in the loop.
[Pilot duration and customer-side effort, to be confirmed by our commercial team]
In a 30-minute call we look at your stack and a few of your stories, and tell you honestly whether a pilot fits.