Skip to main content

Akapulu Labs

​
Interactive AI Avatars

Photorealistic, low latency, expressive avatars for real time video calls.

​
Two ways to use it

Akapulu Labs hosts a public page for you. You pick an avatar, write what the conversation is for, and get a URL. Anyone with the link opens it and clicks Start Call. You do not run a bot, and that page never sees your API key. Best when you want a working call quickly, or something you can share.

​
2. Your own avatar chat

You run a voice pipeline. That can be speech-to-speech, STT → model → TTS, or anything else that produces assistant audio. Akapulu Labs adds the avatar video on top. Calls use Daily (WebRTC rooms). On this path you create the room. Three participants join it:
  • The browser is the person on the call.
  • The voice bot is your pipeline. It listens and does not publish camera or mic.
  • The Akapulu Labs worker publishes the face and the played audio.
In Pipecat, AkapuluAvatar sits inside the voice bot, after assistant audio, and sends that audio to the Akapulu Labs worker. It is not the worker itself. Best when the avatar should live in your product or next to your own bot.

​
Get started

  1. Sign up or sign in.
  2. Pick a path below.
In the dashboard, create a hosted link. Pick an avatar, write the system prompt, and choose whether to show This is AI. Save, open the URL, and click Start Call. You can also create the same row with an API key via POST /api/hosted-links/.

​
Path B: Your own voice bot

  1. Create an API key and a Daily API key.
  2. Either run the public example (Deepgram Flux, OpenAI, and Cartesia Sonic behind the scenes, local page + voice bot), or keep your own pipeline. create_akapulu_call creates the Daily room and the live session. Then add AkapuluAvatar.
See How it works and Your bot for how the room and live session fit together.

​
If you are setting this up with an agent

Have the agent follow Onboard your agent (skills, docs MCP, and the same paths with clearer ask steps).