AI Deep Dive

Test.ai is the AI-First Framework
for software test automation.

 Here’s a glimpse how test.ai uses AI...

Explore the software

The first machine learning technique we use to train our bots is called reinforcement learning.

The bots receive points for reaching their goals and may lose points for actions that don't move them toward their goals.

 

Writing test cases

We use computer vision techniques to help with test case creation. First the bots chop up the screens using element detection. You then train the bots by labeling the elements. Now you can drag and drop the elements to create test cases.

Executing test cases

Test execution brings together reinforcement learning and computer vision. The bots use object detection and recognition to "see" elements in the software. Then they use reinforcement learning to interact with the elements to complete the test case.

Deep dive AI

The elements you label in test cases allows the bot to learn to navigate your app using supervised learning.

Label and icon once and change it’s postion anytime you want! Our AI bots use image recognition to interact with the icon regardless of where it is in the app.

Returning results

Because the bots use computer vision throughout the testing process, the reports include screenshots and display which element was interacted with at each step. Bug reproduction and reporting is a piece of cake.

 

Reducing maintenance

What if elements change in the application?

Maintenance measured by most teams is over 70% of the test case life cycle. AI is more robust and resilient to change. The computer vision (see above) doesn’t rely on traditional selectors so when they change the bots just keep on testing.

Wait then Watch

Our bots also perform intelligent waits to make sure a page is loaded and “settled” before interacting with the page. The bots avoid the traditional “flaky” tests without the need for explicit “waits” or “sleeps”.

Skip editing your code

If the bots don’t “see” an element or the redesign confuses the bots, thanks to computer vision, you don’t need to review any code to update the test case. Add a label to the new element. The bots are then “trained” automatically to see the new version of the element.
 

Get Further deep
in the AI

Read our whitepapers and case studies
  • Case Studies Using AI for Software Testing
  • Testers Don’t Test Anymore
  • Turning Testers into ML Engineers
  • View all resource material

Put the bots to work for you today