AI features

GlassChat ships with a built-in assistant that behaves like a contact, not a surveillance layer.

Chatting with the AI

Your GlassChat AI conversation is always available. It replies casually, remembers context you've shared with it, and matches your language — English, Hinglish, whatever you speak.

In groups, the AI only speaks when spoken to: @mention it or reply directly to one of its messages. It reads recent group context to answer well, knows who's online, and threads its reply to whoever asked.

Image generation

Ask in plain language or use the command directly:


/imagine a rain-soaked neon street, cinematic — count=2 | ar=16:9

Up to 4 variations per request, multiple aspect ratios.
Attach a photo (or reply to one) and ask for a remix — it's used as a reference.
Works in groups without needing to @mention the AI.

Voice replies

Ask for a voice note ("reply in voice", "bhej vn") and the assistant answers as a playable voice message.

Actions

The assistant can do things when you ask: send a message or GIF to a contact on your behalf, create a poll or an event in the chat, or place an AI voice call. Every action posts a visible confirmation — no silent side effects.

Hard privacy lines

The AI cannot join or read end-to-end encrypted chats. The server enforces this, not just the UI.
E2EE content is never included in any AI context, ever.
AI conversations are yours to reset: chat menu → Reset AI memory.