Let an agent create.Keep the project reviewable.
Create short outputs directly or build longer work scene by scene with uploads, patches, checkpoints and visual review.
Your agent builds the video.You can still open and change it.
JsonCut MCP exposes project-oriented authoring tools designed for iterative image and video creation.
Take the shortest useful path
Simple outputs can be created directly without forcing every task through scenes.
Build long work incrementally
Projects, scene patches and checkpoints avoid oversized single requests for longer videos.
Inspect visual progress
Images and configurable frame grids let agents review composition and motion before export.
Made in JsonCut.Open one to see the project behind it.
Every title, clip and caption in these is still a layer. Open the project and change whatever you like.

Remote Culture Manifesto
Open this example: Remote Culture ManifestoVela Institute · The AI Timeline
Open this example: Vela Institute · The AI Timeline
Arbor compliance workspace
Open this example: Arbor compliance workspaceIn the details.What you control by hand.
- Quick pathFinish small work without ceremony
- Short image and video requests can use a concise creation path without mandatory scene planning.
- Project pathBuild longer work incrementally
- Create a project, add scenes, patch only the relevant scope and keep each request small enough to inspect.
- UploadsWork with real files
- Temporary scoped upload URLs let an agent bring image, video and audio resources into the project.
- ReviewLook before final rendering
- Single images and configurable multi-frame grids expose composition, pacing and scene continuity to the agent.
What people make with it.
- Agent-created campaigns
- Long-form structured video
- Template discovery and fills
Before you start.The questions we hear most.
Do I need editing experience to use it?
No. You can ask the AI Director to make the change for you, and every result stays a normal layer you can adjust by hand later.
Does it remain editable after using AI?
Yes. The assistant changes ordinary project layers and versions, so you can inspect and refine the result.
Can the same result be automated?
Yes. Projects can become typed templates or be rendered through the API and MCP.