-
-
The WASP console
-
WASP simulation results - main
-
WASP simulation results - detailed user
-
Example browser agent controlling site to simulate testing
-
WASP running 12 different browser agents
Inspiration
Recently, a friend who works on the UX Research team at Coinbase asked me to complete a quick 15-minute A/B testing exercise for $30. The effort it took didn’t match the compensation, and we were surprised by how costly and ineffective the current user research methods are. We saw an opportunity to build a better solution. They asked us to screen record our selves and explain our thoughts out loud finding different things on their documentation site.
What it does
WASP: Simulated UX testing using thousands of AI agents to help startups make faster, data-informed product decisions.
How we built it
The front end is built with React. The back end uses Flask, Postman, and Browser-Use, powered by the Akash Network.
Challenges we ran into
Browser-Use initially supported only GPT, Claude, LLaMA, and other paid models. Since we wanted a free alternative, we modified it to work with the Akash Network’s API and compute resources. Another issue was running multiple instances of Browser-Use, which was difficult at first, but we found a workaround.
Accomplishments that we're proud of
Initially, we struggled to come up with an original idea, but after some brainstorming, we landed on a concept that hadn’t been done before and built a working MVP within a few hours. We believe this project has real potential as a product.
What we learned
We learned how to work with LangChain and Browser-Use in a practical setting, and got more experience using Cursor with React.
What's next for WASP
We hope to fully build out WASP, ship it to market, and eventually raise a funding round to scale the venture further.
Log in or sign up for Devpost to join the conversation.