One POST queues the render. Poll the record or listen to the live status stream, then receive the MP4 and its full editable scene graph. Included from Plus up, metered at 2x plan minutes. No minimum.
POST /v1/explainer
{
"input": "Explain how recurring billing works",
"length": "1:00", // optional, in 30-second steps up to "6:00"
"aspect": "16:9", // 16:9, 9:16, 1:1
"voice": "sketchie:sulafat", // Nora, the default narrator
"language": "en"
} Creation is asynchronous. The API responds immediately with 202 Accepted and a queued record.
HTTP/1.1 202 Accepted
{
"id": "10f00eee-d2d6-4a1b-b708-f0391faaa85b",
"status": "queued",
"prompt": "Explain how recurring billing works",
"lengthSeconds": 60,
"voice": "sketchie:sulafat",
"language": "en",
"aspect": "16:9",
"videoUrl": null,
"sceneGraph": null
} Fetch GET /v1/explainer/:id until the record is ready, or subscribe to GET /v1/explainer/events for live status updates. The ready record carries the MP4 URL and scene graph.
GET /v1/explainer/10f00eee-d2d6-4a1b-b708-f0391faaa85b
{
"id": "10f00eee-d2d6-4a1b-b708-f0391faaa85b",
"status": "ready",
"videoUrl": "https://media.sketchie.ai/....mp4",
"sceneGraph": { "scenes": [ ... ] }
} The API does not just hand you a file. It hands you a document. Send a plain-language edit through the shipped edit endpoint:
POST /v1/explainer/{id}/edit
{ "instruction": "Make the second scene shorter and warmer" } Seconds later you have a new video with one scene changed and everything else untouched. The new version stays in the same explainer’s history, so you can inspect or revert it through the API.
Sketchie ships an MCP server, so any MCP-capable agent can generate and edit explainer videos as native tools: generate from a doc, inspect the scene graph, request an edit, re-render. If you are building agents that should answer with video instead of a wall of text, this is the shortest path that exists.
Choose Nora, Miles, Dean, Clara, Piper, or Wren from GET /v1/voices. Nora is the warm default narrator, and every catalog voice can speak every supported language. Voice previews are included in the catalog response.