Get Place Go
An agentic AI travel agent for Pune that recommends places by "vibe" — atmosphere, noise, work-friendliness — instead of star ratings, using RAG over reviews and semantic vector search.
Most recommendation apps rank places by ratings. Get Place Go ranks them by vibe — ask it for "a quiet place to work with good coffee" and it understands what you actually mean.
--Vibe-based semantic search User queries in plain English hit a Supabase Edge Function that combines database filtering with Gemini re-ranking: places (with attributes like noise level, wifi, romantic factor) are scored against the query, and each result comes back with a match percentage and an AI explanation of why it fits. Reviews are analyzed with a RAG pipeline and embedded into pgvector for semantic matching, with a graceful fallback to unranked results if the AI response can't be parsed.
--Beyond search -Drag-and-drop itinerary builder** (dnd-kit) with AI-suggested day plans -Vibe categories — Work & Study, Social & Dating, Food & Experience -Favorites with personal notes, backed by Supabase Auth + Row Level Security
- aunched with 50+ curated places across two Pune neighborhoods (Baner, Koregaon Park)
--Stack React 18 + Vite + TypeScript frontend with TanStack Query; PostgreSQL + pgvector, Supabase Auth, RLS, and Deno edge functions on the backend; Zod validation on every edge-function input.