About Realtime Avatar

What we build

Realtime Avatar gives a character a face and a voice and puts it on a live call. You create an avatar from a single portrait, mint a session from your own server, and render it with one component in a browser or a native app. The conversation is full-duplex — the caller can interrupt, and end-of-turn is judged by what was said rather than by silence — and your own tools run inside it.

The platform never executes your tools and never holds your users' credentials. One mint endpoint on your server decides who may call and what the character knows; everything else is rendering and transport. That boundary is the product, and it is documented in full at the developer docs.

How it is priced

One meter: seconds on air, with the language model, the voice and the rendering all included in that rate. There is no separate charge for the model or the TTS, and no seat licence. The full table is on the pricing page, and as raw markdown at /pricing.md.

Built to be read by machines

Most of what reaches this site now is not a browser. Every page answers Accept: text/markdown with markdown at its own URL, every documentation page and blog post is mirrored by appending .md, the whole API is published as OpenAPI 3.1 with the scope each operation needs, and llms.txt indexes all of it. Coding agents can skip the guessing entirely with the realtime-avatar-mcp server described in the MCP docs.

Who operates it

Realtime Avatar is a product of The Influence Company (theinfluence.company), registered at 2261 Market Street, STE 36461, San Francisco, CA 94114, United States.

Questions, bug reports and anything that needs a human go to support@realtimeavatar.ai, or see the contact page for what to send where.