When you're building with video, being 99% right is 100% wrong.
Letting the agent cook, using what it knows already, is correct. LLM's and models were trained on HTML/CSS, and leveraging everything the browser had. We help you package that into an MP4.
But there is more work to video than just the syntax, and to really get it right for agents and video.
Just like there is much more to a database, storage, etc.
What’s missing for real production? Agents are great, the models are great, but the gap is running video workflows in production (for agents and humans).
Large videos need proper streaming playback, cloud rendering when the browser isn’t enough, and easy human-in-the-loop workflows.
The enterprise needs secure ways for their agents to operate, that work.
Anyone who's built in video (with any scale and with any quality) knows how hard this problem actually is.
That’s why we built the full stack with
@editframe
• Custom <ef-video> HTML element with built in streaming playback
• Cloud rendering API (parallel rendering 1000s of videos)
• Declarative GUI toolkit (<ef-timeline>, <ef-scrubber>, etc.) so agents humans can actually collaborate with forms and editors
Editframe provides a format that LLM's have been fully trained on, works natively in the browser, works reliably at scale and in production.
This is what going from prototype to production in video really looks like. This is what does not exist.
A full agent video platform that actually lets you build.
Today,
@editframe emerges from stealth. Agents need video.
Editframe Agent Skills:
npm create @editframe@latest
Just prompt Claude Code, Cursor, or Codex and get a working video or a full interactive GUI.
This video was created just by prompting 👇