Where a human and an agent edit the same document.
Not a chat window bolted onto a video editor. One document, one agent, one you — same state, same undo, same receipt.
The agent's attention is something you can see
One document
Every scene, track, clip, and keyframe has an address. The agent patches it. You see a timeline.
Deterministic edits
Same document, same machine, same pixels — every time. No "huh, that rendered differently."
Free forever
Every deterministic edit costs nothing. Iterate as much as you want.
Visible attention
When the agent is looking at 0:12–0:15, you see that on the canvas, live.
20 seconds, start to export
Three steps, every time
- Direct
Tell the agent what you want, in plain language or by pointing at the canvas.
- Review
The agent hands back a patch with a cost and a preview — nothing lands unseen.
- Ship
Apply it, undo it, or ask for another pass. The receipt says exactly what changed.
Podcast first
A pack is a set of verbs an agent knows for one kind of video.
Podcast Available
Podcast — trim filler, tighten silences, cut a highlight reel. Ships first.
More packs Coming
More packs come after this one proves the shape.
The format will be Apache-2.0. No company tier, no lawyer.
The document schema, the SDK, the engine, the CLI, the MCP server, and the player will all be open. Point your own agent’s harness at it.
Drop your email and we'll get back to you.
Deterministic edits are free forever
Iterate as much as you want at no cost. You pay when you reach for scale — cloud render, generative media, hosted evidence.
| Tier | What it covers |
|---|---|
| Deterministic edits | Free forever |
| Scale — cloud render, generative media, hosted evidence | Pays as you go |
Point your agent at it
Join the waitlist. We'll reach out when early access opens.