Human-like AI avatars. In real time.
Agent Human is the video layer for voice AI. It turns your agent's speech into a photorealistic, lip-synced avatar stream — added to your existing pipeline in a few lines of code.
Free plan included · No credit card required
How it works
One extra step in the pipeline you already have
Keep your speech-to-text, LLM and voice. Agent Human sits after text-to-speech and publishes a lip-synced avatar into your WebRTC room.
Speech-to-text
Your STT
LLM
Your model
Text-to-speech
Your voice
Agent Human
Avatar video
Client
WebRTC stream
Ship a talking avatar in minutes
- 1Create a free account and generate an API key.
- 2Install the plugin for your framework.
- 3Add Agent Human after your TTS step — that's it.
agent_human = AgentHumanVideoService(
api_key=os.getenv("AGENTHUMAN_API_KEY"),
session_request=NewSessionRequest(avatar="avat_..."),
transport=transport,
)
pipeline = Pipeline([
transport.input(),
stt, user_aggregator, llm, tts,
agent_human, # ← add the video layer
transport.output(),
assistant_aggregator,
])Platform
Everything you need to put a face on your agent
Agent Human does one job — the video layer — and is built to do it exceptionally well.
Built for live conversation
Low-latency, GPU-accelerated video generation designed to keep pace with real-time voice agents — not pre-rendered clips.
Drop-in for Pipecat & LiveKit
Official plugins for both frameworks. Agent Human sits after your TTS step and publishes video to your room — no re-architecture.
Simple REST API
Create a session, stream audio, receive synchronized video. Works with any HTTP client and any WebRTC-based stack.
Multilingual lip sync
Lip movements stay accurate in whatever language your TTS speaks, so one avatar can serve a global audience.
Bring your own avatar
Start from any portrait photo — upload an image or pass a URL. No preset library lock-in.
Custom avatar generation
Generate a photorealistic, on-brand presenter from scratch when you don't have the right photo.
Use cases
Where a face makes the difference
People trust, engage and stay longer when they can see who they're talking to.
Customer support
Give your support agent a consistent, friendly face that's available around the clock.
Sales & lead qualification
Turn website visitors into face-to-face conversations that qualify and book meetings.
Education & training
Interactive tutors and role-play partners that learners can actually talk to.
Onboarding & assistants
Guide new users and customers through your product with a human touch.
Pricing
Simple, usage-based pricing
Start free, upgrade as you grow. Additional minutes are billed at $0.18 per minute on any paid plan.
Free
Evaluate the platform
- 60 minutes/month
- 5 minutes session limit
- 2 concurrent sessions
Explorer
For individuals and small projects
- 500 minutes/month
- 20 minutes session limit
- 5 concurrent sessions
- Unlimited custom image avatars
- 20 avatar generations/month
Growth
Most popularFor growing teams and products
- 3,000 minutes/month
- 60 minutes session limit
- 10 concurrent sessions
- Unlimited custom image avatars
- 60 avatar generations/month
Pro
For high-volume production use
- 8,000 minutes/month
- 180 minutes session limit
- 20 concurrent sessions
- Unlimited custom image avatars
- 180 avatar generations/month
Need higher volume or a custom setup?
Talk to us about volume pricing, higher concurrency and dedicated support.
What exactly does Agent Human do?
Agent Human is the video layer for voice AI. You keep your own speech-to-text, LLM and text-to-speech; we take the generated audio and stream back a lip-synced, photorealistic avatar video into your WebRTC room in real time.
Which frameworks and platforms are supported?
We provide official integrations for Pipecat (AgentHumanVideoService) and LiveKit Agents (AvatarSession), and support Daily and LiveKit rooms. You can also use the REST API directly, embed sessions via iframe, or share a hosted conversation link.
Can I use my own avatar?
Yes. Provide any portrait photo by URL or upload, and it becomes a talking avatar. On paid plans you can also generate a custom photorealistic avatar from scratch.
Does it work in languages other than English?
Yes. Lip sync is driven by the audio itself, so it works with whatever language your TTS produces.
How is usage billed?
Plans include a monthly allowance of streaming minutes and a concurrent-session limit. Usage beyond your allowance is billed at $0.18 per minute. The Free plan requires no credit card.
How do I get started?
Create a free account, generate an API key in the dashboard, then follow the Pipecat or LiveKit quick-start in our documentation. Most teams have a talking avatar running within minutes.
Give your AI agent a face.
Start free in minutes, or talk to our team about your use case and production requirements.