Novarite
Powered by QVAC local AI

AI Game Discovery

Search by mood, genre, difficulty, length, or similarity. QVAC local matching recommends games from a real reference catalogue and highlights Novarite games when available.

Try an example

QVAC Local AI Engine

Novarite's discovery engine uses QVAC embeddings (GTE-Large FP16) to match natural language queries against a real game reference catalogue entirely on-device. Your search preferences never leave your machine — no cloud AI API, no tracking, no rate limits.

The hosted web page uses lightweight browser-side matching as a demo. Run the local QVAC CLI or start the API server for full semantic embeddings:

node qvac/game-discovery.mjs "underwater survival exploration"node qvac/server.mjs# then set NEXT_PUBLIC_QVAC_API_URL=http://localhost:4000

About this integration

Novarite integrates the QVAC SDK as its core game discovery engine. QVAC embeddings run locally and on-device to match player intent against game metadata using cosine similarity over GTE-Large FP16 vector representations. This directly addresses one of indie gaming's central problems: small games are hard to find, and players often don't know exact tags or genres. QVAC allows natural-language discovery while preserving user privacy.

On-device
Model runs locally via QVAC runtime
Private
Zero queries sent to cloud APIs
100+ games
Real reference catalogue across all genres