Sume basics

Sume is fundamentally a video agent platform. People author and iterate in chat; partners invoke saved recipes over HTTP. Individual generation tools (Image, Video, Avatar, TTS, timeline, and more) also exist as HTTP APIs, but the main partner path is calling a sandbox Agent / Format that composes those tools — not stitching raw model calls yourself.

That composition is why you can ship deliverables a single Video 1.0 clip cannot: multi-minute host footage, B-roll, voiceover, and timeline assembly into a post-ready video.

For job admission, polling, and media.sume.com mechanics, see Core concepts.

In a nutshell

Agents

Agents is the chat UI where a person works with the sandbox Agent: write a brief, approve spend, inspect artifacts, and shape a house style over a few turns.

This is where Formats are authored — either by editing a Format in the library or by asking the Agent in chat to save a recipe (SKILL.md plus references). Interactive chat is the right surface when a human should stay in the loop.

Docs: Agents overview · Safe automation

Formats / Format API

A Format is a saved authoring recipe. Partners call it by handle and slug; Sume boots a fresh sandbox, loads the recipe, runs the Agent with generation tools, and returns artifacts plus optional structured JSON.

This is the surface most partners should integrate. One HTTP call carries the judgement and orchestration that would otherwise live in your own glue code.

Docs: Format API overview · Calling a Format · Bulk runs · Cookbook: embed a Format

Models

Models are atomic generation endpoints: create an avatar, render a talking clip, generate an image or a short video clip, add captions, and so on. They are real product surfaces — useful when you need one model invocation and nothing else.

They are supporting relative to Formats. A Format decides which tools to call, in what order, and how to assemble the result. If you only need a single clip or image, call the model; if you need a packaged workflow, call a Format.

Docs: Models overview · Video 1.0 · Avatar videos

Agent Completions / Scheduled

Not every agent task is worth saving as a Format.

SurfaceWhen to use
Agent CompletionsOne-off backend task; nothing worth saving as a recipe
ScheduledThe same saved task on a cadence (Actions)

Both run the Agent; Completions is ad-hoc, Scheduled is recurring. Formats stay the path when the recipe is fixed and only the inputs change.

SDK

The TypeScript SDK is a thin client over the same Developer API for Node / Bun / Deno / Workers, including subscribeFormatRun so you do not hand-roll polling. Everything it does is also reachable over plain HTTP.

Two more clients exist and still work, but they are not part of the primary path today: the CLI for local shells and scripts, and hosted MCP for clients that speak remote MCP. Reach for them when your environment calls for it, not as the default integration.

Dashboard

The dashboard is the human operator surface for the same workspace the API key resolves to:

Docs: API keys · Jobs · Usage · Billing

Developer API + media

DomainRole
api.sume.comPublic Developer API (/v1) and OpenAPI (/reference/json)
media.sume.comFirst-party generated media artifacts

Keys authenticate; the API is workspace-scoped. Generated outputs that belong to Sume are returned as media.sume.com URLs — that is the public artifact contract.

Docs: Public API · API reference · Authentication · Media inputs

Workspaces

API keys and spend resolve to a workspace. The key carries that context — do not send workspace_id in request bodies. Team-owned Formats are reachable on both vanity and opaque paths with a key created in that team workspace; a personal key fails with 403 workspace_key_required — see Team Formats need a team key.

What next?

  • Quick start — your first run, in the Agents tab or over the API
  • Format API — why Formats exist and how a run works end to end
  • Core concepts — jobs, admission, artifacts, usage