Drop your footage. Type the cut. Ship the video. FrameOS turns 47 clips and 6 hours of scrubbing into one chat command.
BY THE NUMBERS
Small chat.
Serious output.
EVERY EDITOR KNOWS THIS ONE.
47 clips. 6 hours in.
You opened the project at noon. It's dark now. You've watched the same ten seconds eight times. The render bar still says 23%.
You talk.
It edits.
Drop your footage. Type the cut you want. Watch FrameOS turn six hours of scrubbing into six seconds of typing.
CREW MODE
Meet the
crew.
Six capabilities. One timeline. Each role specialised, each accountable, each running on your choice of model.
Picks the shots.
Reads every clip, picks the keepers, orders the scene before a single cut happens.
Cuts. Paces. Grades.
Trims to length, drops the bad takes, color-grades to a vibe. No mouse-jockey required.
Word-perfect subs.
Frame-accurate timing, reads-the-room punctuation, burned in at export.
Three agents, one timeline.
Each role runs in parallel. Planner queues, editor cuts, captioner burns. Done before you scroll.
Gemini or Ollama.
One toggle in the desktop app. Same chat. Same timeline. Different room.
Local-first. Air-gapped.
Run the whole crew on your laptop. Your weights, your footage, zero round-trips.
BUILT FOR EDITORS
POWERED BY · HYBRID STACK
Ollama edits.
Gemini generates.
Editorial intelligence runs locally on Ollama. Whenever the cut calls for footage you didn't film, Gemini generates it on demand — Veo 3 for video, Imagen for stills. Frontier generation, no frontier bill on the boring work.
- Planning + intent parsing Ollama
- Cut decisions + pacing Ollama
- Captions + transcript search Ollama
- Generated b-roll video (Veo 3) Gemini
- Generated images + thumbnails (Imagen) Gemini
- Multi-modal scene understanding Gemini
Gemini
every pixel you didn't shoot
When the cut needs footage you didn't film, Gemini generates it. Veo 3 for video, Imagen for stills, native multi-modal for the scene reasoning behind every generative call. This is where the magic comes from.
- veo 3 · generated b-roll
- imagen · stills + thumbnails
- multi-modal scene grounding
Ollama
the workhorse
The everyday editorial brain. Runs on your machine. Plans the cut, parses your intent, decides pacing, burns captions, searches transcripts. Zero round-trips, zero spend, zero upload.
- plan + intent + pacing
- captions + transcript search
- 100% on-device · private
The result: cloud-quality generation, on-device speed for everything else. You pay Gemini for the pixels, not the chatter.
SHIP IT.
Ship a video this weekend.
Download the desktop app. Drop your footage. Type the cut.
/auto-edit · /caption · /export