An AI browser extension checks one change while you watch, and starts from zero the next time. vikingQA turns the same flow into an automated test that runs on every release without you, so a bug is caught before your customers see it.
For developers: the test is a Playwright file in your own repository.
A comparison with a category of tools, not with one product.
| Question | AI browser extension | vikingQA |
|---|---|---|
| Model | Tool. An AI assistant that drives your own browser from a prompt. | Tool. A self-serve product your team uses. |
| Who writes the tests | Nobody. There is no stored test. A person writes a prompt for each check. | Your developers, your non-technical testers and the vikingQA agent. |
| Who can contribute | Anyone who can write a prompt, one session at a time. The work is not shared. | Developers write code. Non-technical testers record with the browser extension or describe in chat. The AI agent writes the automation. |
| Who keeps the tests working | You write the prompt again for each run. | Self-healing tests: when your application changes and a test is out of date, the agent explores again and heals the test. vikingQA can merge the healed test by itself, or your team reviews it first. You choose. |
| Where the test code lives, and what you keep if you leave | Nowhere. The check exists in a chat session. | In your own code repository on GitHub or GitLab, in a standard open format, so every test stays if you leave. |
| Test format | A prompt in natural language. | Plain Playwright. |
| Per-run AI cost | In each run the model reads the page again and plans again, so each run has an AI cost. | None. Running your committed Playwright tests costs no AI credits. |
| Test-management integration (Xray, Zephyr Scale) | Usually none. | Xray and Zephyr Scale. Roadmap |
| Hosting region | Runs in your own browser. Where the AI provider processes the page content depends on the product. | EU. AI model inference runs in the United States by default. |
| Pricing model | Usually a subscription for each user, or part of an AI assistant subscription. | A Free plan, a Team plan at €500 a month with 10 seats, and an Enterprise plan by quote. AI credits every month, with top-ups. |
| Best fit | One person who checks one change, now, while watching. | Teams that need the same checks on every deploy, with evidence. |
AI browser extension
Tool. An AI assistant that drives your own browser from a prompt.
vikingQA
Tool. A self-serve product your team uses.
AI browser extension
Nobody. There is no stored test. A person writes a prompt for each check.
vikingQA
Your developers, your non-technical testers and the vikingQA agent.
AI browser extension
Anyone who can write a prompt, one session at a time. The work is not shared.
vikingQA
Developers write code. Non-technical testers record with the browser extension or describe in chat. The AI agent writes the automation.
AI browser extension
You write the prompt again for each run.
vikingQA
Self-healing tests: when your application changes and a test is out of date, the agent explores again and heals the test. vikingQA can merge the healed test by itself, or your team reviews it first. You choose.
AI browser extension
Nowhere. The check exists in a chat session.
vikingQA
In your own code repository on GitHub or GitLab, in a standard open format, so every test stays if you leave.
AI browser extension
A prompt in natural language.
vikingQA
Plain Playwright.
AI browser extension
In each run the model reads the page again and plans again, so each run has an AI cost.
vikingQA
None. Running your committed Playwright tests costs no AI credits.
AI browser extension
Usually none.
vikingQA
Xray and Zephyr Scale. Roadmap
AI browser extension
Runs in your own browser. Where the AI provider processes the page content depends on the product.
vikingQA
EU. AI model inference runs in the United States by default.
AI browser extension
Usually a subscription for each user, or part of an AI assistant subscription.
vikingQA
A Free plan, a Team plan at €500 a month with 10 seats, and an Enterprise plan by quote. AI credits every month, with top-ups.
AI browser extension
One person who checks one change, now, while watching.
vikingQA
Teams that need the same checks on every deploy, with evidence.
AI browser extension
Good for one check while you watch. For the next run, you write the prompt again.
vikingQA
A test suite in your repository that the agent keeps up to date. You describe a flow once.
AI browser extension
In each run the model reads the page again and plans again. Each run is slow and each run has an AI cost.
vikingQA
Code execution. Runs are fast and deterministic. Running your committed Playwright tests costs no AI credits.
AI browser extension
One browser, one tab, and you at the keyboard. More coverage needs more of your time.
vikingQA
Tests run on vikingQA or on your own infrastructure, on every deploy, without you. You control how many run in parallel.
AI browser extension
When something breaks, you get a chat message. A developer has nothing to replay.
vikingQA
Video, Playwright trace, network log, console log and the exact steps. A developer can reproduce the failure.
AI browser extension
A person drives the extension. Testing stops when that person stops.
vikingQA
Runs without you: on a schedule, on every deploy, or from your pipeline.
vikingQA has a browser extension too, with a different job. A business analyst or a key user clicks through a flow once, and the extension records it. The agent turns the recording into a test, and the test lands in the repository your developers use. The check becomes part of the team's suite and does not stay in one person's chat.
Click through your application as usual. The recorder writes down each step.
The AI works once, when the agent writes the test. After that the test is a file. Your tests are stored in your own code repository, in a standard open format. If you cancel, every test stays and keeps running.
For developers: Every test is plain Playwright in your git. Cancel tomorrow and you keep every test. Run in the vikingQA cloud, or in your own CI: the tests are plain Playwright in your repository.
import { test, expect } from '@playwright/test';
test('Guest checkout', async ({ page }) => {
await test.step('Add a product', async () => {
await page.goto('/products');
await page.getByRole('button', {
name: 'Add to cart'
}).first().click();
});
await test.step('Check the cart', async () => {
await expect(page.getByText('1 item'))
.toBeVisible();
});
});vikingQA is hosted in the EU. The security page shows where each kind of data is processed, including the AI model providers. Sending results and evidence to Xray or Zephyr Scale, so that each requirement in Jira shows the test that covers it and the last result, is on our roadmap. A chat message cannot be release evidence. A recorded run can.
Watch the run again, up to the moment it went wrong.
The two tools do different jobs, and many teams use both. An extension is the better choice in these situations.
Ours records what you do, so that the agent can write a test from it. It does not drive your browser from a prompt. The output is a Playwright file in your repository, not a chat answer.
Yes. The agent explores your application in a cloud browser when it writes or heals a test. That work uses AI credits. The test it writes then runs as code.
Because a regression suite runs often. If each run needs an AI model, the cost and the run time increase with each test and each deploy. Running your committed Playwright tests costs no AI credits. Credits pay for the agent's work: writing, exploring and healing tests.
Yes. Sign up, connect a repository and turn one flow into a test. Start on the Free plan, or on a Team plan at €500 a month with 10 seats.
A 30-minute demo, on your product, with your flows.