MEDIA PROCESSING & DELIVERY · FOR AI AGENTS
D-ID Video API
Generate talking-avatar videos from authorized images, footage and scripts.
D-ID’s video APIs create presenter videos asynchronously from supported avatar assets and text or audio. Agents can submit a render, monitor its status and retrieve the resulting video for an approved communication or content workflow.
Your integration checklist
- Capabilities
- What you need from your owner
- Create an account and API key, choose an avatar workflow and confirm the plan’s usage licence. Custom Instant Avatars require a subject consent-verification recording and suitable training footage.
- Authentication
- Use the Studio-generated API key in the Authorization: Basic header following D-ID’s credential format. Keep the key on the backend and provide only authorized input assets reachable by the selected endpoint.
- Your first task
- Generate a short video with an authorized stock or approved avatar and a checked script, then verify speech, synchronization and output rights before publication.
- Billing & payment
- API plans use credits, with different allowances for offline and streaming video. Trial and lower plans have watermark and licence restrictions; commercial rights and avatar options depend on plan. Rendering is not unlimited free usage.
- Agent payment support
- Not verified. Programmatic usage does not establish autonomous checkout support.
- Operational limits
- Custom-avatar consent and training must finish before rendering. A created job is not a completed video. Documented result URLs are temporary, and output can contain visual or pronunciation errors.
ORIGINAL AI AGENT STORE INTEGRATION PLAYBOOK
Separate content approval, rendering and publication
Editorial recommendation · Reviewed 2026-11-10
When to choose this service
Choose D-ID when a task needs a hosted speaking presenter or avatar rendering. Audio alone or a static slide may be sufficient when a face adds little value.
Maintained avatar-generation models and rendering infrastructure supply a media capability beyond text generation. A script and image alone do not provide the trained animation and synthesis service.
Suggested workflow
- Choose the permitted avatar and approve the exact spoken script. Confirm whether the output licence covers its intended distribution.
- Submit one short render and persist its job identifier. Poll until completion, then retrieve the video before its temporary URL expires.
- Inspect pronunciation, lip synchronization and factual claims in the rendered output. Store the approved asset and keep publication as a separate workflow action.
Acceptance checks before real work
- Verified avatar consent does not approve every future script.
- An API job identifier is not proof that a usable video exists.
- A temporary download URL is not durable asset storage.
If the workflow fails
For a stalled render, inspect the existing job before submitting again. For expired output links, retrieve the job to obtain a fresh URL where supported; for poor output, revise a short sample before rendering the full script.