Testing & Automation
Web App Automation
Ask Gunamaya AI to open your app, click through it, and confirm a change actually worked — using the same Integrated Browser you'd use yourself.
The basic loop
The pattern that makes this useful is simple: you describe a behavior, Gunamaya AI makes a code change, then opens the Integrated Browser, navigates to your running app, and interacts with it to confirm the change had the intended effect — before ever telling you it's done.
Describe the goal, not just the code
Instead of "change this function to round after tax," try "the cart total is wrong when a discount is applied — find and fix it, and check that it's actually correct afterward." Gunamaya AI will investigate, fix, and verify.
It opens the browser
Gunamaya AI navigates to your local dev server (or whatever URL is relevant), using the same Integrated Browser tab mechanism you'd use manually.
It interacts with the page
Clicking buttons, filling in forms, and reading back the resulting page state — the same actions a person testing the flow by hand would take.
You see everything that happened
Every navigation, click, and read is shown in the conversation as it happens, not hidden behind a final "done" message — you can stop or correct course at any point.
What Gunamaya AI can check
- Visual state — does an element appear, disappear, or change as expected after an action.
- Navigation — does clicking a link or submitting a form land on the right page.
- Console output — did the change introduce a new console error or warning.
- Screenshots — capturing a before/after comparison to include in its explanation of what changed.
Why this matters more than a description
A model describing what code should do is guessing. A model that opens the actual running result and clicks the actual button is reporting what happened — the gap between those two is exactly where "looks right but doesn't work" bugs live. Grounding Gunamaya AI's answer in a real browser session closes that gap.
Point it at a specific URL
If your app runs on a non-default port or a specific route matters, just say so — "open localhost:5173/checkout and try the discount code path" — Gunamaya AI will navigate there directly rather than guessing.
Automated checks alongside manual browsing
Nothing about this replaces you opening the Browser tab yourself — the two work together. A common flow is Gunamaya AI doing the first pass of verification after a change, then you opening the same Browser tab afterward to look at the result yourself before considering the work done.
Where this is headed
Web automation today covers exactly the workflow above. See the Automation Roadmap for how this extends toward desktop applications, APIs, mobile apps, and dedicated security and performance testing.