Semantic Search Prototype

SteamWatchAI MVP-2
Steam games collage

Find Steam games by describing the experience you want.

This version searches across 10,000 popular Steam games using developer descriptions, stored in PostgreSQL and indexed in Qdrant via precomputed about_embedding. User queries are transformed into vector representations before being matched against the game-level index behind the scenes.

10,000 games indexed Qdrant + PostgreSQL Openrouter baai/bge-m3 query embeddings

Example: “I want a game where I can explore a dark world for hours and make meaningful story decisions.”

Run dataset ingest and index rebuild first, then search by mood, mechanics, setting or genre blend.