Hello, Robin here. I am one of the cofounders at Testzeus, and we had built an open source testing agent. Using AG2 framework, our agent uses tools like Playwright for browser interaction, Nuclei for security scanning or AXE for accessibility. So, with the release of Deepseek R1, and earlier launch of OpenAI Operator, I was curious if we could generate UI tests using Deepseek R1, and feed them directly to the open source testing agent for autonomous execution. And similar case with API testing, where we could feed an OpenAPI spec to Deepseek, ask it to generate tests and then directly run these tests with the testing agent.
To my surprise, this pipeline worked pretty well for both UI and API scenarios. Here is a video around my observations: <a href="https://youtu.be/-gbNDZ1eqnY" rel="nofollow">https://youtu.be/-gbNDZ1eqnY</a>.
We feel that engineering teams still struggle with velocity of delivery versus its quality, so agentic testing with open source solutions like AG2, Deepseek and TestZeus will change the world for better.