Most people asking "how do I connect an avatar to my agent" have already solved the hard part: an LLM-based assistant or agent platform that generates good replies. What's missing is a face and a voice for that agent to speak through. That's specifically the problem Trulience is built to solve - our avatars sit in front of the agent you already have, rather than asking you to rebuild it from scratch.
A Trulience avatar is the presentation layer - the face, the voice, the lip-sync, the gestures. Your agent, whatever LLM or conversational service is generating replies, keeps doing exactly what it already does. Trulience's job is to route the conversation between the two: text or audio in, speech and animation out.
If your conversational AI already runs on a platform like VAPI, ElevenLabs Conversational AI, or Agora's TEN framework, Trulience integrates directly with it: the platform keeps handling speech-to-text, the LLM, and text-to-speech, and Trulience takes care of the avatar's visuals and lip-syncs them to the audio it's given. Older conversational-service integrations such as Google's Dialogflow still work too, though most new integrations today are built around one of the platforms above or a custom LLM setup.
Trulience is compatible with all major LLMs, including hosted models such as OpenAI's ChatGPT, and connects to any custom REST endpoint - so a backend you've built yourself can be wired in directly, with your own RAG, knowledge base, and function calling handled on your side.
Once your agent's responses are flowing through, Trulience's SSML support lets the avatar do more than lip-sync the words. By injecting standard Speech Synthesis Markup Language tags into your agent's response text, you can have the avatar raise an eyebrow, smile, frown, or turn and look in a given direction while it speaks - all controlled from the same responses your agent already generates. See our developer docs on SSML authoring for the full tag reference.
Conversations run full duplex with acoustic echo cancellation, so there's no push-to-talk or waiting for a beep - people can interrupt and be interrupted the way they would with another person. Trulience avatars can hold natural spoken conversations in over 140 languages, and can be embedded wherever you need them: drop one into a page with a simple iframe, or use the JavaScript SDK for a tighter integration in a web or mobile app.
There are two ways in, and you don't need to pick one over the other: the Interactive Avatar Creator dashboard lets you sign up and have a fully responsive avatar live in a couple of clicks, with speech-to-text, your LLM, and text-to-speech all configured right there in the dashboard - no integration code required.
If you'd rather wire in the agent platform or LLM/REST backend you already have, the APIs, SDKs, and full integration reference are at docs.trulience.com. Either way, if you'd rather talk it through first, get in touch and we can help you scope the integration.