Project 01
1st of 5 teams. An idea nobody else pitched won it.
Built the AI layer for Aria Hotels' travel co-pilot — a quiz-based hotel matcher, two culturally-tuned concierge agents, and a monitoring dashboard added after feedback surfaced a trust concern. Invited into Phase 2 of the final client product, leading front-end design.
Role
AI Product Builder (Phase 1) + Front-End Design Lead (Phase 2)
Context
Kimolian Academy × Aria Hotels
Timeline
Competition sprint + final product
Stack
Lovable · Figma AI · Miro · Useberry · OpenAI Agents · Supabase · Claude Code
1st / 5
Pitch competition winner
2
AI agents built & calibrated
Phase 2
Invited to co-build the client product
The problem
Sophisticated travelers want culturally immersive trips, but planning tools are fragmented across searching, planning, traveling, and reflecting. Aria wanted an AI co-pilot for every one of those stages — without guests needing a high-end travel agency to get that depth.
A second, quieter problem surfaced in the first feedback session. Aria is a small team who take customer care personally, and handing guest conversations to an AI meant handing over something they couldn't watch happen.
How I approached it
Discovery
Interviewed target travelers to understand both sides — what Aria wanted solved, and what guests actually needed.
Research
Ran the team through a broad–narrow–broad–narrow design-thinking process to converge on the problems worth solving.
Build
Shipped hotel and activity info, a best-fit quiz, a saved profile/itinerary, and two AI concierge agents with distinct personalities.
Feedback
Tested with travelers and Aria's team, who asked: how do we know the AI is giving guests the right answer? I scoped a hotel-side monitoring layer no other team addressed.
Pitch
Returned with a clearer story and the monitoring dashboard built in, turning a client concern into a concrete feature.
What I shipped
- Quiz-based hotel matching — a short quiz returns the best Aria property and curated activities.
- Eye-catching local-experience browsing across hotels.
- Profile-based trip planning — save experiences, build an itinerary, book events.
- Two distinct AI concierge agents rather than one general assistant, each with its own personality and focus.
- A hotel-side monitoring dashboard giving staff visibility into agent responses and a way to hold quality.
Result
Won the competition, 1st of 5 teams — then was invited onto the team actually building the product for Aria, with the competition ideas feeding into that build.
Led front-end design for Phase 2: authoring story card content for a story-driven discovery layer, and running QA on agent routing to help choose the technical provider. Phase 2 is due with the client soon.