Editing,
automated.
How AI agents actually change video editing — what they do well, what they get wrong, and how to wire them into real work.
The worst bug in a video editor is not a crash. It is the edit that looks finished and is not — the clip that comes out black, the voiceover that exports silent. Nothing warns you, so you assume the tool is broken and you leave. Otto is our answer to that.
Most advice for new creators is really advice for freelance videographers from 2014. The job has changed: the scarce thing is no longer the ability to operate an editor, it is having something to say and the stamina to keep saying it. Here is our definition of video creation, and the roadmap we would give anyone starting now.
MCP lets Claude call a video editor directly. Here is what that actually looks like in practice — the setup, the commands worth knowing, and the parts that still need a human eye.
Telegram video bots have existed for years, but they take commands like /trim and /rotate. This is different: you describe the edit in plain language and an agent performs it on a real timeline.
Most "AI video editing" is a set of buttons with a model behind each one. An agent is different: it reads the project, decides which operations to run, and executes them. That distinction changes the workflow.
Exporting the same 16:9 file to every platform wastes most of the screen and buries your subject. Repurposing properly means reframing, not just resizing — and it does not have to mean four separate edits.
Everything here is true of the shipped product. Try it on your own footage.
Start editing free →