SimbaStack · 2026-05-21 · notable
Framedex — A 5-Year-Old MacBook Runs Gemma 4 31B Locally to Index a Year of Video Into a Plain-English Knowledge Base
Open-source tool that indexes a video archive into per-clip markdown sidecars using local Gemma 4 31B, WhisperX transcripts, and face embeddings. The write-up hit the Hacker News front page.
Local-first tool that turns an unlabeled video archive into a plain-English searchable knowledge base.
What is it?
Framedex is an open-source Python tool that catalogs a personal video archive instead of editing it. For every clip it writes a markdown sidecar describing what is in the footage, so the whole library becomes searchable in plain English. Everything runs locally on consumer hardware.
How does it work?
It runs Gemma 4 31B locally via LM Studio to write prose scene descriptions, WhisperX for transcripts, and insightface for face embeddings, then extracts metadata like lighting, time of day, and GPS coordinates. Claude orchestrates the pipeline, and the output is plain markdown files saved next to each video.
Why does it matter?
Once an archive is searchable in plain English, finding the right clip stops being the bottleneck. The author argues the valuable layer is the index, not an AI editor, and shows a five-year-old MacBook indexing a year of footage overnight with no cloud upload of personal video.
Who is it for?
local-AI tinkerers and videographers
Try it
github.com/Simbastack-hq/framedex