A site brain in your browser

Meet Byte.

A serious site-aware companion with a face, a mood and a little bit of attitude. Byte grounds replies in the site's catalogue and public docs, then runs the language model on your device instead of sending your conversation to an AI service.

LOCAL inference GROUNDED site knowledge PRIVATE memory ALIVE animated face

listening for you

Give Byte a brain

The first load downloads a small model into this browser's cache. Later chats can run without a cloud API.

Local model

The quickest option is a good first test. A WebGPU-capable browser is required for real model replies.

No model loaded yet

You can still try the clearly labelled demo mode below.

Site brainloading…

Loading the catalogue and public site guidance for grounded answers.

read replies aloud with your browser voice

Private memory

Only facts you explicitly save live in this browser. They are never added to the shared site brain automatically.

Privacy note. Byte's inference and site index stay in this tab; private memory and chat history stay in this browser's local storage. The model files and runtime are downloaded on the first load; your messages are not sent to the model host or an AI API. Optional browser speech is controlled by your browser or operating system, so leave that toggle off for strict local-only use.

Chat with Byte

demo mode

site context

Enter a message, then press Ctrl + Enter to send. Byte searches the local site brain before answering; optional browser speech is separate.