How GUI Test Studio works
The whole idea in two minutes: record a test or describe it in a prompt, refine, run, read the report.
The idea
You test your application by hand already: you click through a scenario and look at the screen.GUI Test Studio turns that walk-through into an automated test.You show it once, it checks it on every run.
There is no script to write.A test is made of the steps you performed and the checks that say what correct looks like, laid out on a timeline like a video edit.
There are two ways to create a test.Record it yourself: GUI Test Studio generates the test from the recording, visually, ready to run and to edit at any time.Or describe the workflow in a prompt: your AI assistant records, verifies and runs it for you through GUI Test Studio's connector, Claude, ChatGPT or a local AI model we provide or you bring, and the result is the same test.
You open your app, type a user name and a password, click Sign in, and see "Welcome back, Anna".
GUI Test Studio records those steps and adds checks along the way: the Sign in button is there before it is clicked, and the welcome message appears afterwards.Every run replays the clicks and the typing, and tells you if anything looks or reads differently.
Two apps and a runner
One installer gives you three tools that work on the same test files.
| Tool | What you do with it |
|---|---|
| GUI Test Runner-Recorder | Record a new test, run a test, and read its Confidence Report. |
| GUI Test Studio | Shape a test on a timeline: add or adjust checks, name steps, group them into Scenarios, and keep the test up to date while your app is being developed. |
| gtr-cli | Run tests without a window, from a script or a CI/CD pipeline.Same runs, same results. |
| The prompt connector | Describe a test in a prompt, in Claude, ChatGPT with Codex, VS Code or to a local AI model we provide or you bring: the assistant records it, edits it, runs it and reads its report through the two apps. |
The four steps
Record
Do your scenario once by hand, or describe it in a prompt
Refine
Adjust it on the timeline, or ask for the change
Run
Replay it on every build
Read
The report says what changed and why
Refining is optional: a recording is a complete test the moment you save it, with its checks already in place.You come back to GUI Test Studio when you want to change something.
What a check is
A check (assert) is a moment where the test makes sure something is right: a button is there, a text says the right thing, a dialog has closed, a list has three rows.There are visual checks, text checks and layout checks.
1. Find
2. Compare
Expected
what you recorded
Found now
on the live screen
Same element, same look: passed
The GTS Engine + AI finds each element two ways.It first asks the application itself: a desktop app's accessibility information, the same information a screen reader uses, or a web page's own elements.When the application does not say, as with a game, a canvas, a remote desktop or a program drawn by its own toolkit, the engine finds the element by sight, on the screen, the way a person does.
Two ways to the same element are what make a test robust.It keeps finding a button after the code behind it changed, the window moved or the layout shifted, and it works whatever your app is: a website, a desktop program, a game or a phone shown on your screen, built with any technology.
Finding the element is the first half of a check.The second half always compares what is on the screen with what you expect, so a test catches what your users would see.
What you can use it for
| Kind of testing | How GUI Test Studio helps |
|---|---|
| End-to-end (E2E) tests | Replay a whole user journey on the real screen, across every window and app it touches. |
| Functional tests | Check that one feature does what it should: the right screen appears, the right text and values are shown. |
| Regression tests | Run the same tests on every build and learn at once when something that worked has changed. |
| Visual checks | Catch what other tests miss: a wrong colour, a missing icon, an overlapping button. |
| Tests written ahead | Prepare a test from design pictures before the feature exists. |
See Test a feature before it exists for the last one.
During development, not only at the end
Tests are not something you add once the product is finished.Create them while a feature is being built, and change them as it changes: record a new Scenario, move a step, replace what a check expects, on the timeline or by prompt.GUI Test Studio is made for that everyday editing, and every change can be undone.
| Traditional development (SDLC) | Development with AI agents (ADLC) | |
|---|---|---|
| Who writes the code | Developers, sprint by sprint | AI coding agents, many changes a day |
| The risk | A release breaks a screen nobody retested | Fast changes break screens faster than people can look |
| How the tests help | Run on every build, they replace the manual regression pass before a release | Run after every agent change, they are the eyes that confirm the screen is still right |
| When a test fails | The bug report goes to a developer | The bug report, with its steps and pictures, goes straight back to the agent |
Where to go next
Still stuck?Tell us what you were trying to do, and we answer within one working day.Ask our team