Vibe Vote

Not another restaurant chatbot—Vibe Vote is a decision room: collect group preferences, generate comparison cards, vote, then reserve instantly using Yelp AI API actions.

Comment 1

Inspiration

Choosing where to eat is surprisingly hard when it’s a group decision: everyone has different budgets, cravings, dietary needs, and “vibe” preferences. The conversation loops, links get spammed, and nobody feels confident enough to book.

We were inspired by Yelp’s new conversational AI API and its agentic reservation actions. We wanted to turn “endless group chat” into a single decision flow: collect preferences → compare options → vote → reserve.

What it does

Vibe Vote is a group decision room for going out:

  • Create a room and share a link with friends
  • Everyone submits quick preferences (budget, cuisine, dietary, vibe, time, party size)
  • The app uses Yelp AI API to:
    • discover relevant places
    • generate side‑by‑side comparison cards
  • The group votes on the top choices
  • The app confirms details and triggers a reservation action (where supported), returning the booking confirmation

How we built it

  • Product design:
    • A “room-first” UX to coordinate multi-person preferences
    • A “compare-first” UI to reduce decision fatigue before voting
    • A “confirm-first” reservation flow to make agentic actions safe
  • Backend (API):
    • Session/room state for participants, preferences, candidates, and votes
    • An orchestrator layer that routes conversation turns to Yelp AI API and formats results into comparison cards
  • Frontend (Web):
    • Room page (share link)
    • Chat + comparison cards + voting UI
  • Engineering for production direction:
    • Clear action confirmation step before booking
    • Structured error handling and user-facing fallbacks
    • Basic request tracing (e.g., request_id) for debugging and reliability

Challenges we ran into

  • Avoiding “generic chatbot”: It’s easy to build a chat UI that just recommends places. The hard part is designing a workflow that actually helps a group converge on a decision.
  • Action safety: Agentic reservations need explicit confirmation and robust fallback paths when a reservation isn’t supported or fails.
  • Demo constraints: We optimized for a 3‑minute demo that still feels like a real product, not a toy.

Accomplishments that we're proud of

  • A clear end-to-end experience that highlights Yelp AI API’s strengths:
    • discovery
    • comparisons
    • reservation actions
  • A product direction that can realistically evolve beyond hackathon scope:
    • stateful rooms
    • reusable core orchestration
    • room → compare → vote → book loop

What we learned

  • The biggest UX win isn’t “more recommendations” — it’s faster convergence.
  • For agentic actions, trust comes from:
    • transparent intermediate steps (comparison cards)
    • explicit confirmation
    • graceful failure handling
  • A strong hackathon submission is much easier when the product story, UX flow, and technical architecture reinforce each other.

What's next for Vibe Vote

We’re building Vibe Vote as a long-term, commercializable product:

  • Phase 1: polish the group decision room and reservation reliability
  • Phase 2: add a Date Night mode (preset vibes, 2-option showdown, “surprise me”)
  • Phase 3: add a Diet/Allergy mode (profiles + safer comparisons)
  • Monetization:
    • B2C premium tiers (saved rooms, recurring plans, concierge mode)
    • Later B2B expansion (venue-side funnel analytics and conversion tools, subject to API/policy constraints)

Built With

Share this project:

Updates