Atmosfy · AI Companion · Mobile + Web
Designing an AI companion for local discovery
I designed the standalone Tibbs AI experience across V1, V2, and V3, including in-app entry points and a browser-based demo that let people try the companion without downloading the app.
Project overview
Role
Product Designer
Platform
Mobile and Web
Scope
Product design across v1, v2, and v3
Atmosfy’s creator videos made it easy to discover inspiring places, but finding the right recommendation for a specific moment still depended on browsing feeds, searching, and filtering.
The opportunity was to introduce a conversational layer that could understand natural-language intent and connect people directly to relevant places and creator content, making discovery feel more personal, focused, and actionable.
Evolution across three versions
V1
Expressive exploration
A vibrant purple direction centered on personality, atmosphere, and card-based discovery.
V2
Refining the recommendation
A black-and-gold direction with stronger hierarchy, clearer actions, and a more focused recommendation card.
V3
Simplifying the conversation
A more direct conversational experience that brought prompts, recommendations, and Atmosfy content into one continuous interaction.
V3 experience
Design decisions
Starting a useful conversation
The opening experience uses lightweight prompts to show what Tibbs can do without forcing users through a lengthy setup. A clear first question helps people express what they want in their own words.
Making results actionable
Recommendations are presented as clear options rather than chat-only answers. Users can explore a suggested place, refine their request, or continue the conversation without losing context.
Connecting AI to Atmosfy content
Tibbs pairs conversational recommendations with relevant Atmosfy places and creator content. This grounds each response in something visual and explorable while keeping discovery connected to the broader product experience.
Browser demo
Created for potential investors and partners, the browser-based demo offered a glimpse of Tibbs as a standalone product outside the Atmosfy app, pairing guided onboarding with conversational discovery.
Designing Tibbs reinforced that a useful AI experience depends on more than the quality of its responses. The conversation needed a clear starting point, recommendations needed obvious next steps, and every suggestion needed to feel grounded in Atmosfy’s content ecosystem.
The project also explored an emerging interaction model: opening a social platform’s library of creator videos to conversational discovery. Instead of relying only on feeds, search, or filters, people could describe what they wanted and use AI to surface relevant places and videos.
Exploring Tibbs across multiple versions helped separate the core product idea from any single interface. The browser demo showed how it could grow beyond an in-app feature into a standalone discovery product while preserving the context and visual content that made it valuable.