01[ Research ]
Start with the people who do the testing.
Before anything was designed, I ran market research and structured interviews with QA engineers, developers and founders. The goal was to find and validate the real pain points in existing QA workflows — not the ones we assumed. Those conversations shaped Aurick from a loose idea into a defined product.
02[ Design ]
Make an autonomous system legible.
I explored interaction approaches through Figma wireframes and prototypes, then converged on Aurick's core workflow: AI-generated test cases, execution, session playback and human-in-the-loop review. Playback and review exist so people can see what the AI did and step in — trust was a design problem, not just a model problem. I then built that workflow on the frontend myself.
03[ Evaluation ]
Watch real people use it.
A 10-user closed beta gave us something to learn from. Analytics, session recordings and direct feedback pointed to friction points, and I A/B tested key flows to decide what to change.