While a session runs, GlazeBot takes frames from the screen or app you shared and sends them to an AI model that reads what’s happening. Another model writes a line in your character’s personality, and a voice model reads it out loud.
Characters don’t speak on every frame. GlazeBot picks who talks next and what kind of line fits: a quick reaction, a longer thought, two characters going back and forth, or nothing at all.
If you turn on voice input, they can respond to what you say too. To see what gets sent, read what GlazeBot sends.