Tellers for developers

Ship the video feature.
Skip the video stack.

Upload, understand, search, edit, preview, and render video through one API—backed by an open timeline format, a CLI, an editing agent, and a unified layer across leading generative models.

Read the API docs Talk to an engineer
  • REST API and CLI
  • Open timeline format
  • No video stack to operate
API / VIDEO PIPELINE200 CONNECTED
REQUESTEVENTSRESULT
POST /create

{
  "asset_id": "ast_7f2c",
  "prompt": "Cut the best 60s",
  "output": "16:9"
}
01 · IMMEDIATEPreview URLplayer02 · PORTABLETimeline JSONOTIO03 · ASYNCRendered videoMP4

What changes

Build the product.
Not the plumbing.

  1. 01Integrate

    Add AI video without operating video infrastructure.

    Use one API for media upload, analysis, search, editing, preview, rendering, and delivery instead of assembling a custom stack.

  2. 02Orchestrate

    Give LLMs a predictable video layer.

    Generate structured, versionable timelines through an open OTIO-based format and render them through the same production system.

  3. 03Adapt

    Use new generation models without new integrations.

    Call leading video, image, voice, music, and sound models through one abstraction instead of maintaining vendor-specific pipelines.

The workflow

Request in.
Video out.

A small, composable workflow for product features, automated content pipelines, and LLM-driven video applications.

  1. 01

    Upload and index media

    Push media through presigned upload URLs or the CLI. Tellers analyzes and indexes each asset for search and editing.

  2. 02

    Request an edit

    Send a prompt or a structured tellers-timeline describing the video, constraints, and output you need.

  3. 03

    Stream progress and preview

    Consume server-sent events and open an instant player preview while the pipeline completes.

  4. 04

    Render or keep editing

    Trigger the final render, store the portable timeline, or continue refining it programmatically.

Inside one workspace

Everything the video stack needs.

  • 01

    REST API

    Upload, search, edit, preview, render, and deliver media with asynchronous jobs, progress events, and webhooks.

  • 02

    Open timeline format

    Use an OTIO-based JSON structure designed for version control, programmatic generation, and LLM output.

  • 03

    CLI

    Upload local media, trigger analysis, and ask the editing agent for a video directly from the command line.

  • 04

    Video intelligence

    Search scenes, transcripts, objects, entities, actions, and meaning without building a separate indexing layer.

  • 05

    Instant timeline player

    Preview multi-source timelines, mixed codecs, and graphic overlays before waiting for a full render.

  • 06

    Model aggregation

    Use leading generative media models through a unified configuration and billing layer.

Open where it matters

A managed video layer.
A portable timeline.

Tellers handles the infrastructure-heavy parts of video while keeping the composition explicit. Your application can generate, store, diff, and revise timeline JSON instead of treating every output as an opaque render.

  • ✓Open-source timeline schema and library
  • ✓Instant previews before final rendering
  • ✓Asynchronous events and webhook delivery
  • ✓Editable outputs instead of render-only results

Questions, answered

Before you
get started.

What does the Tellers API handle?

The API handles media upload, storage, analysis, indexing, semantic search, agent-driven editing, timeline previews, rendering, and delivery. It also provides a unified layer for generative media tools.

What is tellers-timeline?

tellers-timeline is an open-source, OTIO-based JSON format for video compositions. It represents scenes, assets, timing, audio, captions, graphics, and output settings in a predictable structure that software and LLMs can generate.

Can I build an LLM-driven video workflow with Tellers?

Yes. An LLM can create or revise a tellers-timeline or send natural-language editing requests. Tellers provides the media intelligence, preview, and rendering layer that turns that structure into video.

How do progress updates work?

Long-running operations are asynchronous. The create workflow streams server-sent events, and production pipelines can use callbacks or webhooks to receive completion and delivery information.

Do I need to build a transcoding and search stack?

No. Tellers is designed to centralize video preprocessing, indexing, search, mixed-source preview, and rendering so product teams do not have to operate those systems separately.

Can I use the timeline outside Tellers?

Yes. The timeline format is portable and OTIO-based. You can store it, version it, generate it programmatically, and use it as the source of truth for additional integrations or finishing workflows.

Start with one request

Bring the product idea.
Use the video layer.

Explore the API surface, timeline format, and integration patterns—or talk through the workflow with the team that built them.

Read the API docs Talk to an engineer