Helping people find a gym that fits.
An internal AI-assisted fitness-discovery prototype designed to use location and a short preference conversation to recommend three relevant gym options.
Workflow overview
A guided way to find a fitness fit.
Location
Start from a supplied catalog of gyms.
Preferences
Ask five to ten questions, one at a time.
Shortlist
Select three gyms from the supplied list.
Structured return
Pass selections and context to the application.
The product question.
A long list of nearby gyms does not necessarily help someone decide where to train. The useful starting point is the person: what they want, their preferences and where they are looking.
FitLocal explored a discovery experience built around that information. The prototype flow collected a location, received a list of gyms and asked preference questions one at a time before returning three recommendations.
Xivic's contribution.
FitLocal was an internal Xivic product exploration. The 2024 prototype specification defined the conversation, the gym-matching step and the information passed between them. The design constrained the recommendation to a supplied gym catalog, rather than asking a model to invent options. The conversation gathered preferences; the application received a shortlist and the context behind it.
Where this project stands.
FitLocal remains an internal 2024 prototype. It has not launched as a consumer service.
A next evaluation would check whether the shortlist fits a person’s stated preferences, whether every recommendation comes from the supplied catalog and whether the person understands why each gym was suggested.
What decision should your product make easier?
Bring the user need and the information people struggle to turn into a useful choice.