See what AI assistants recommend, and why

People increasingly ask AI assistants what to buy. Optshun is a research workspace that shows what those assistants recommend to different kinds of buyers, which information drives the recommendation, and how it changes over time.

1. Generate synthetic observations. Thousands of isolated questions to several AI models, each about one buyer persona and one version of the product information.
2. Find what drives the answers. A relational feature network and a Relational Transformer model show which information the answers depend on.
3. Test and track. Change one fact at a time, repeat the study every month, and compare with real human research.
4. Act on it. Turn findings into content and paid tests, and record what actually happened.

The demo follows a fictional e-bike brand, Halden Cycles, through a full study.

Take the product tourBrowse the design system

This is a design prototype made in Penpot. All companies, products and numbers are fictional.