I scrolled through the entire readme and didn't see any mention of sqlite_vec. My feedback for the readme would be to optimize for signal- if it is a layer on top of sqlite_vec say what it does on top of that etc
sqlite-vec is a great vector index — Wax actually uses SQLite under the hood too.
The difference is the layer. sqlite-vec gives you vec_distance_cosine() in SQL. Wax gives you: hand it a .mov file, get
back token-budgeted, LLM-ready context from keyframes and transcripts, with EXIF-accurate timestamps and hybrid
BM25+vector search via RRF fusion — all on-device.
It's the difference between a B-tree and an ORM. You'd still need to write the entire ingestion pipeline, media parsing,
frame hierarchy, token counting, and context assembly on top of sqlite-vec. That's what Wax is.
Thanks for clarifying. If mv2s is a sqlite3 db file under the hood that is something I would like to see in the readme as it would make me more likely to use.
Stefan-H|13 days ago
kleton|13 days ago
ckarani|13 days ago
The difference is the layer. sqlite-vec gives you vec_distance_cosine() in SQL. Wax gives you: hand it a .mov file, get back token-budgeted, LLM-ready context from keyframes and transcripts, with EXIF-accurate timestamps and hybrid BM25+vector search via RRF fusion — all on-device.
It's the difference between a B-tree and an ORM. You'd still need to write the entire ingestion pipeline, media parsing, frame hierarchy, token counting, and context assembly on top of sqlite-vec. That's what Wax is.
kleton|13 days ago