bitHuman bitHuman docs
Status Get API key
API PlatformCLIApps SDKShowcase
ExamplesChangelogDownloadsCommunity Status
Get API key →
Docs / API Platform

API quickstart

Pick an existing agent and get a live, talking avatar on your page in two steps — no agent generation required.

The fastest way to start: pick an agent that already exists and embed it. No generation, no polling, no credits to create one — a live, talking avatar in two steps. (Want your own custom face later? See Generate your own agent.)

1. Pick an agent

Every agent has a short code like A78WKV4515. Browse the agent gallery and copy the code of any one you like — or use A78WKV4515 to follow along.

2. Embed it

Drop it onto any page as an iframe. It’s live and talking immediately — the user can speak to it and it responds:

<iframe
  src="https://agent.viewer.bithuman.ai/api/embed/A78WKV4515"
  allow="microphone *; camera *; autoplay *"
  style="width: 100%; height: 600px; border: 0;"
></iframe>

Replace A78WKV4515 with your chosen code. That’s the whole quickstart — open the page and start talking.

Note The iframe needs delegated microphone permission to hear the user. If your page sets a restrictive Permissions-Policy, the avatar loads but the mic stays silent — allowlist agent.viewer.bithuman.ai. For per-visitor session tracking and rate limiting, mint a short-lived embed token on your backend — see Embedding.

Going further

The steps above need no API key. The rest of the platform does — get an API secret at Developer → API Keys (free tier, no credit card) and export it:

export BITHUMAN_API_SECRET=your_api_secret

Verify it with the cheapest call there is — no credits, no agent needed:

curl -X POST https://api.bithuman.ai/v1/validate \
  -H "api-secret: $BITHUMAN_API_SECRET"

It always returns HTTP 200 — read the body: {"valid": true} means you’re set, {"valid": false} means the secret is missing or wrong.

Look up an agent

Fetch any agent’s details by code:

curl https://api.bithuman.ai/v1/agent/A78WKV4515 \
  -H "api-secret: $BITHUMAN_API_SECRET"

Make it speak from your backend

When an agent has an active session (the embed above, or a LiveKit room), push text into it and the avatar speaks it aloud:

curl -X POST https://api.bithuman.ai/v1/agent/A78WKV4515/speak \
  -H "api-secret: $BITHUMAN_API_SECRET" \
  -H "content-type: application/json" \
  -d '{"message": "Hello! Great to meet you."}'

Note /speak and /add-context require an active session. With no live room you’ll get a 404 (code NOT_FOUND, message "No active rooms found for agent <code>") — open the embed first, or start a LiveKit worker.

Voice without an avatar

Text-to-speech needs no agent at all — one call returns a WAV:

curl -X POST https://api.bithuman.ai/v1/tts \
  -H "api-secret: $BITHUMAN_API_SECRET" \
  -H "content-type: application/json" \
  -d '{"text": "Hello from bitHuman.", "voice": "F1"}' \
  --output hello.wav

See Text to Speech for languages, voices, and streaming.

Generate your own agent (optional)

Prefer a custom face and persona? Generation is asynchronous — it returns an agent_id immediately and takes ~2–5 minutes (250 credits):

curl -X POST https://api.bithuman.ai/v1/agent/generate \
  -H "api-secret: $BITHUMAN_API_SECRET" \
  -H "content-type: application/json" \
  -d '{
    "prompt": "You are a friendly fitness coach.",
    "image": "https://example.com/headshot.jpg"
  }'

Then poll GET /v1/agent/status/{agent_id} until ready and embed it exactly like step 2. See Agents for the full lifecycle.

Next steps