ABOUT VOIZUAL
A website that can talk back.
Voizual gives a website an AI agent for text chat and live avatar calls. A visitor presses Call and has a conversation with an agent that has its own instructions and knowledge — in the browser, with no signup and nothing to install.
How a conversation works
- Your agent, your knowledge. Add a profile portrait, choose an available voice, write instructions, and add the documents and web pages it should answer from. Choose a voice from the available catalogue. Voice cloning and voice design are not available in this version. Live calls currently use one shared presenter. Your uploaded portrait is your profile picture and call poster; it does not change the live face. Custom live avatars are not available in this version.
- A real conversation. Visitors speak or type. The avatar listens, answers, and can be interrupted mid-sentence, with live captions on screen.
- What comes next. A conversation can end in a summary by e-mail and a button to your own booking or sign-up page, and every conversation is in your admin.
What it runs on
Calls travel over WebRTC through LiveKit. Speech is handled by OpenAI's GPT-Live speech-to-speech model, and the avatar is rendered by Protoface's real-time avatar service.
Knowledge it answers from
PDF, DOCX, TXT and Markdown documents, and whole websites crawled and indexed into searchable passages, so answers stay grounded in your own material instead of guesswork.
Nothing to install
Voizual runs the service for you: there is nothing to host or install, and no keys of your own to manage. Transcripts are stored in your account, where you can read and delete them.


