There's no shortage of AI testing tools. Every month brings a new one that promises to write your tests, fix them when they break, and save your team hours. Many of them are genuinely good at what they do. So why build another one?
Because most of them ask for something in return that we think teams shouldn't have to give up: their test code, their framework, their data, or their budget. Testver was built around a different idea. Bring AI to the tests you already have, instead of moving your tests into a new tool. This post explains what that means in practice, and where it makes Testver different.
What most AI testing tools ask of you
Look closely at a typical AI testing platform and you'll usually find the same trade:
Rebuild your suite in their format. Tests are recorded or written into the platform's own steps, which run only on the platform's own runner.
Run on their cloud. Your application under test, and often your test data, has to be reachable from the vendor's servers.
Pay per seat, per run, per minute. For a team, that's typically $10,000 to $15,000 a year, before any AI usage.
Manage test cases somewhere else. Test management lives in a second tool, and the link between a written test case and the code that automates it is kept by hand, if at all.
If you ever leave, your suite stays behind. Testver takes the opposite approach on each of these points.
1. Your framework, your code, no rewrite
Testver works with the framework you already use: Playwright, Selenium, Cypress, WebdriverIO, Puppeteer, Cucumber, Robot Framework, pytest, REST Assured and Appium, in the languages your team writes. Point Testver at your test project and it finds your tests, your page objects and your configuration. There's nothing to import or convert.
Every test Testver writes, whether the AI writes it or you record it, is plain code in your repository, in your framework, following your conventions. It runs with your own command, in your own CI, with or without Testver. If you stop using Testver tomorrow, every test keeps working. There's no lock-in, because there's nothing to lock.
2. AI that knows your project
A generic AI writes generic tests. Testver's AI starts from your project. It reads your existing tests, your page objects and helpers, and how your project runs tests, and writes new tests the way your team would.
It reuses what you have. When you record a journey with AI Record, you can type @ to point a step at a helper your project already has, such as a login function, and the AI calls it instead of writing those steps again.
It knows what a change affects. The Code Map connects your tests to the page objects and methods they use, so you can see which tests a change touches before you run anything.
It answers questions about your suite. Ask the AI Assistant "which tests cover checkout?" or "which tests are flaky?" and it answers from your real tests and their run history.
3. AI that proves its work
An AI that writes a test and hands it to you is only doing half the job. In Testver, the AI runs the test it writes, reads the failures, and fixes them until the test passes. You see every file it changed and every run it made.
The same rule applies when things break later. When a developer renames an element and a test fails, Auto-Heal finds the broken locator in your code and works out the new one from the page the failing run actually saw. Then it re-runs every affected test to prove the fix. If the tests pass, the fix is kept, or offered for your review in Suggest mode. If they don't, your code is put back exactly as it was. Nothing is ever changed on a guess, and every fix can be undone.
And for steps too complex for a single click, AI Annotations let you describe them in plain words while you record: "the total must equal subtotal minus discount plus shipping plus tax", or "add every item under $50 to the cart". The AI turns each one into real code at exactly that point in the test.
4. AI only where it adds value
Many AI testing tools use a model for everything: every click, every page, every run. That's slow and expensive, and it adds up quickly when a suite runs many times a day.
Testver uses AI where it actually helps, and nowhere else:
Recording uses no AI. AI Record captures every element, its locators and the page around it on your machine. The AI is used once, to write the test from steps that are already exact.
Running your tests uses no AI. Scheduled and CI runs execute your framework directly, so a suite can run all day at no AI cost.
Most repairs use no AI. Auto-Heal tries the free evidence first: the recorder's saved alternatives, the failing page and the run's trace. A model is asked only when that evidence can't decide.
The result is a tool whose AI bill stays small even as your suite grows.
5. Test management that's connected to your automation
This is where many teams feel the biggest difference. In most setups, test management and test automation are two different worlds. Test cases live in a test management tool, and automated tests live in a code repository. Keeping them in step is a spreadsheet, a naming convention, or a plugin, and it drifts as soon as someone renames a test.
In Testver they're one system.
Write and organize test cases. Test Cases holds your manual test cases in folders, each with preconditions, numbered steps and expected results, priority, type and full version history. Test Plans group them for a release, with cycles for each round of testing and results recorded step by step, with evidence. Moving from another tool? You can import test cases from TestRail, Zephyr, Xray, PractiTest and qTest, or from JSON and CSV files.
Let AI draft them. AI Test Generation writes test cases from material you already have: a requirement, a PDF, a user story, a Jira issue, or even your existing automation scripts. It covers the happy path, the negative cases and the edge cases. You review them, keep the good ones and save them into your catalog, linked to the requirement they came from.
Link every test case to the script that automates it. Script Mapping connects each manual test case to the automated test that covers it, down to the exact test in your repository. Select a test case, pick the test from your discovered scripts, and click Map. One automated test can cover several test cases. From the other side, the Mapped TCs tab in the Test Explorer shows which test cases each automated test fulfils.
And Testver keeps those links honest. A mapping isn't a note that silently goes stale. Testver watches both sides:
Case Updated: someone edited the test case, so the script may no longer match it.
Script Updated: the automated test changed since it was mapped.
Script Missing: the test was renamed or deleted.
A single Sync Health score tells you how much of the map you can still trust, and the dashboard shows your automation coverage: how many test cases are backed by real automation, and which still need it.
The AI does the linking for you, too. When the AI writes a test, from AI Record, the AI Assistant or Explore & Automate, it files a test case for it and links the two. A test case created from a requirement stays linked to that requirement, so coverage counts it straight away.
From requirement to result, in one view. Put together, this gives you full traceability. The Traceability Matrix follows each requirement to its test cases, shows whether each one is automated, and rolls up the latest results: passed, failed or not run. When an automated test fails, you can raise a defect from the failure in one click, prefilled with the error and steps, and push it to Jira, where Testver shows its live status.
Why it matters: green test runs on their own don't tell you what's covered. Linked test cases do. With Testver, "is this requirement tested, automated and passing?" has an answer you can show in an audit or a release sign-off, without a spreadsheet.
6. Web, API and mobile in one place
Most teams test more than one thing, and most tools cover only one of them. Testver covers all three:
Web: record or describe a test, run it on Chromium, Firefox and WebKit, against any environment. See Web automation testing with AI.
API: send requests, chain flows, run data-driven and contract tests, and turn them into automated tests. See API automation testing with AI.
Mobile: record and run Appium tests on emulators, real phones or a device cloud (in beta). See Mobile automation testing with AI.
Around them sits everything a QA team needs every day: failure analysis that explains in plain words why a test failed, schedules, reports, and CI/CD pipelines that let GitHub Actions, GitLab CI, Jenkins, Azure Pipelines, Bitbucket or CircleCI run your tests on Testver and gate the build on the result.
7. Free and self-hosted
Testver is free. It runs on your own computer or server, next to your code, so your application, your tests and your test data stay with you.
For AI you have two choices. Use Testver cloud AI, where you pay only for the AI you use (the Starter plan is free and needs no card). Or run a local model through Ollama or LM Studio, where the AI is free too and nothing leaves your network.
When the whole team wants in, install Testver on one always-on server, or let us host it with Testver Cloud, an add-on that gives your team its own workspace. Testver Grid, also an add-on, runs your tests on cloud browsers in parallel.
What about an AI coding agent?
It's a fair question. Today's AI coding agents are capable: they can write a test, run it in a terminal and fix it. For one developer and one test, that can be enough.
Testing a product as a team needs more than that. Testver adds the parts an agent alone doesn't:
A recorder, so anyone can turn a walkthrough into a test, with no prompt to write.
Evidence for every run: screenshots, traces, video and history, kept for every test.
Repairs that are proven by a re-run before anything is kept.
Test management for the whole suite: test cases, plans, linked automation and coverage.
Unattended runs at no AI cost, on a schedule or in CI, gating a deployment on the result.
A web app the whole QA team can use, not just the person at the terminal.
At a glance
Your tests: plain code in your framework and your repository, not a vendor format.
Lock-in: none. Everything keeps working if you walk away.
AI that proves its work: it runs what it writes, and Auto-Heal verifies every fix with a re-run.
AI cost: recording, runs and most repairs use no AI at all.
Test management: built in, with test cases linked to the scripts that automate them, and drift flagged automatically.
Coverage: from requirement to test case to script to result, in one view.
Web, API and mobile: in one tool.
Hosting and price: free and self-hosted, with Testver cloud AI or free local models.
For the full side-by-side comparison, see Testver vs other tools.
See it for yourself
The quickest way to see the difference is to point Testver at a suite you already have. It takes a minute, your tests stay exactly where they are, and nothing changes unless you ask it to.
Run testver serve in your test project, then follow the installation guide to connect AI and record your first test. New to Testver? Start with Meet Testver.