Bring your own LLM
Bring your own LLM means using your own choice of large language model to decide what an AI Avatar says, while the avatar platform handles only the voice and video. It keeps the conversation logic, prompts and model choice under your control.
How bringing your own LLM works
The avatar platform receives text or audio from your model and renders it, rather than deciding the answer itself. You can switch models, fine-tune one, or run one in your own environment, without changing the avatar.
Teams choose this to keep a model they've already tested, to meet data rules, or to control cost and quality per use case.
How Beyond Presence does it
Beyond Presence handles the real-time video layer while your LLM powers the conversation: OpenAI, Anthropic or a custom model. With the Speech-to-Video API you control the whole pipeline, including the LLM. See the Help Center.
Frequently asked questions
Can I use my own LLM with Beyond Presence?
Yes. Connect OpenAI, Anthropic or a custom model, and Beyond Presence renders the Avatar.
What stays under my control when I bring my own LLM?
The model, the prompts and the conversation logic. You can switch models without changing the Avatar.
Does my LLM affect Beyond Presence latency?
Yes. The LLM and voice take most of the reply time. The Beyond Presence Avatar render adds about 100 ms.