how does the AI agent work?

The biggest challenge of AI-based testing is combining the deterministic nature of tests with non-deterministic AI to produce 100% valid tests.
explore the AI

Octomind AI Agent in a nutshell

Octomind uses AI for its common-sense knowledge and leaves the code generation to more predictable systems. Our AI agent navigates through the web app, understands it, and identifies relevant user flows that need to work. It mimics the user (e.g., clicks on the input field and enters email, clicks on sign-up to newsletter, etc.).

AI test case discovery

How do we discover relevant test cases for your web app without you telling us what to do? 

Our AI agent goes through the publicly accessible code in the DOM. It uses the vision capability of the underlying multimodal LLM to add visual context where the code insight falls short.  
a screenshot of octomind AI agent generating test steps

AI test case building - step generation

Our AI Agent interacts with the app as a user would, aiming to reach the goal of a user flow.

We are recording and storing the interaction chain of a test case in an intermediate representation. The corresponding  Playwright code which exercises the UI is generated deterministically on the fly just before test execution.
AI auto-maintenance
This feature is under active development and not publicly accessible yet.

First, we find out if a test failure is caused by a behavioral change of your user flows, the test code itself or a bug in your code.

In the case of a behavioral change, we pinpoint the failing interaction. We apply the AI Agent to detect the new desired interaction to achieve the original goal of the test case.
Using AI to fight flaky tests
This feature is under active research and not publicly accessible yet.  

We deploy active interaction timing to handle varying response times and follow best practices for test generation already today.

We are looking into AI based analysis of unexpected circumstances to handle temporary pop-ups, toasts and similar stuff that would break a test.

blogtopus: On AI, agents and app testing

50% developer rants, 50% tech news commentary, 100% cephalopod opinions on all things software. Worth the read.
We have launched the beta version. Browse our docs at your own pace or request a personalized demo, if you need to talk to our devs. Or just start for free, today.
Try out our beta