Moneyball

Project Description

Before LLMs became mainstream, a team at Google developed an early large-language model that could cluster sentiment from conversations across the internet.

They asked us to explore how marketers might interact with this data—how they would search, explore, and make sense of real-time public sentiment.

I collaborated closely with a brilliant researcher to define the core value proposition, then designed and tested multiple rounds of prototypes. The goal: to make an unfamiliar search experience feel intuitive, flexible, and grounded in real-world use cases.

Adapting Google's design system to support this new interaction model required careful balance between consistency and invention.