AI place search that shows its work.
When Foursquare acquired Superlocal, it didn't get folded away. Superlocal kept running as its own product, and I kept leading it from inside Foursquare.
Today I'm Foursquare's consumer tech lead. I run Superlocal, lead its next generation of conversational place search, and built that AI place search into an enterprise API, so the same grounded answers that power the app are available to any company building on Foursquare's data.
Design decisions
- 01
Keep Superlocal its own product
Instead of merging it into another app, Superlocal kept its identity and its users, now powered by Foursquare's place data. That gave Foursquare a live consumer product to design and test AI search with real people, not just an API.
- 02
The AI never speaks for the data
Every recommendation traces back to a specific, verifiable fact about a place. When a restaurant is called romantic, the UI shows the tips and signals that earned it the label. That's not a style choice. It's the central principle of grounded AI search, and it runs through every screen.
Justification UI: why a place is being recommended. - 03
Ask the way you'd ask a friend
"A romantic Italian place in Chicago." "Cheap cafes near me with outdoor seating." "A quiet place to work with good wifi." My job is designing how that conversation feels: the query patterns, the results surface, and the moments where the system has to earn trust.
- 04
One search, two audiences
The consumer app and the enterprise API run on the same AI place search. Designing for both kept the core honest: if an answer isn't grounded enough to show a developer, it isn't good enough to show a person either.
Highlights
- Consumer tech lead at Foursquare
- Kept Superlocal running as an independent product inside Foursquare
- Built AI place search into an enterprise API
- Designed the justification UI behind grounded recommendations
Next