The agent defines test cases, explores your app, runs the tests and writes the automation. No developer needed: it can add and heal tests by itself, or your team can review each change first. Then the tests run on every release, on vikingQA or on your own infrastructure, and bugs are caught before your customers see them.
For developers: Every test it writes is plain Playwright in your git. Each change is a pull request. You merge it yourself, or you turn on automatic merge and vikingQA merges it for you.
Write a test for guest checkout. The acceptance criteria are in SHOP-412.
I read the ticket and explored the flow on Staging. Here is the test I propose.
Explored the app · 14 steps
Ran the test · passed
Committed 1 file
tests/checkout/guest-checkout.spec.ts
Added to your suiteAdded without a developer. Your team can review each change first instead.
The agent builds your tests, runs them, catches the bug and heals the tests when your application changes.
Tell the agent what matters, or point it at a ticket or a spec. It writes the test case: a title, a description of what to verify, and the steps. You correct it in plain language.
The agent opens your app in a cloud browser and walks the flow the way a tester does. It finds the real fields, buttons and states before it writes a line of code.
The agent runs the test in an isolated sandbox and reads the result. If the test fails, it looks at the evidence and tries again until the test passes or it can explain why not.
The agent turns the test case into an automated test that runs on every release.
Self-healing tests: when your application changes and a test is out of date, the agent explores again and heals the test. It adds the change by itself, or your team reviews it first. When the failure is a real defect, the agent does not change the test. It says so and files the issue instead.
Open a chat in a project and an environment. Write what you want tested, the way you would ask a colleague.
A run fails on a schedule or in your pipeline. The agent picks it up, reads the evidence and comes back with a healed test or a defect report.
Give the agent a standing task, such as a weekly review of the suite against the latest release notes.
A pull request opens in GitHub or GitLab. The agent reads it and drafts the tests for the change. A start from a ticket that moves to "Ready for QA" in Jira or Linear is on our roadmap.
A new tester needs weeks to learn how your product works. The agent starts from what your team already wrote.
The agent reads specs and acceptance criteria in Notion, Confluence and Google Drive, and tickets in Jira and Linear. You choose the sources for each project.
What the agent learns in one session is there in the next: how to log in, which data to use, which flows are fragile.
Each project has its own instructions for the agent. Write your conventions once: naming, test data, what never to touch.
Teach the agent a repeatable procedure for your project, such as how to create an order in your back office. It uses the skill each time the task comes up.
No developer needed. Anyone on your team can accept a change in vikingQA, or vikingQA can add and heal tests by itself. You choose.
The agent adds a test only after the test ran and passed. Each change is recorded in vikingQA and in the history of your repository: what changed and why.
Each session has a limit on AI credits. When the session reaches it, the agent stops and tells you what is done and what is left.
Stop the agent at any moment. It halts, and the work it already did stays in the chat.
Running your committed Playwright tests costs no AI credits. Credits pay for the agent's work: writing, exploring and healing tests.
No. It writes test files only. Each change is a pull request. You merge it yourself, or you turn on automatic merge and vikingQA merges it for you.
No. vikingQA can add and heal tests by itself, or your team can review each change first. You choose for each project.
The agent does not change the test to make it pass. It reports the defect, and it can create the issue in Jira or Linear with the steps, and the video and the trace attached.
Running your committed Playwright tests costs no AI credits. Credits pay for the agent's work: writing, exploring and healing tests. Every plan includes a monthly credit allowance. When you reach it, top up in the app. No plan change needed.
Yes. Each integration is connected once for the organization and then enabled per project. You choose the capabilities the agent has in each project.
No. It uses the environment you choose, with the variables and secrets you set for that environment. Secrets are encrypted and injected at run time.
The model providers we use do not train their models on data sent through their commercial APIs. This is a contractual commitment in their agreements.
Connect a repository and see your suite in vikingQA in minutes.