Agentic conversations
Tool calling and MCP for calendars, booking, and workflows. Agents decide in real time with your knowledge bases.
Agents that think, talk, and take action
Build agents with tool calling, MCP, knowledge, and live voice. Run inbound and outbound calls. See places and routes in 3D — inside the chat.
Start a chatWhy GLASS
Live voice
Talk with Gemini Live or Ultravox. Low latency, interruption handling, and streaming speech-to-text.
Your own agents
Name, voice, tools, an MCP server, and knowledge. Use the same agent in chat, live voice, and phone calls.
3D maps
Explore places and routes. Share where you are, drop markers, and ask for directions from the chat.
Call automation
Inbound and outbound calls via Twilio. Tasks for appointments, bookings, and follow-ups — with the same tools as chat.
Knowledge and memory
Custom and system knowledge. Mine facts from a chat and file them into memory so the next turn starts informed.
What you can do
Files, photos, and folders
Attach from your device or grant a local folder. Drop a photo and edit it just by asking an agent.
Language shortcuts
A short language hint is enough — even a half-word. The agent understands and translates.
Models you choose
Pick any model provided by OpenRouter, including image and video generation.
Share and fork
Share a chat with a link. Anyone can open it and fork a copy into their own workspace.
Media storage
Browse files from chats in a gallery — preview, download, rename, and keep folders organized.
Talk to the agent
Speak and listen. Voice control with streaming speech-to-text and spoken replies.
The right agent for the job
Create your own agents. GLASS assigns the right one. The planner helps research using several agents.
Push notifications
Get notified when a long-running task finishes — even if you left the tab.
Interface your way
Dark or light theme, accent color, dialogue text size, and UI language.
On every screen
Web
Open glass.cat in the browser. No install required.
iOS and Android
A native shell around the same workspace — sign in with Google, GitHub, or Apple.
Desktop
A Tauri desktop wrapper for the same app when you want it on the dock.
Frequently asked questions
What is GLASS?
A workspace for conversational agents: typed chat, live voice, phone calls, 3D maps, knowledge, and tools — in the browser and in the native apps.
Do I need my own API keys?
No to start. You can add Google, OpenAI, Anthropic, or OpenRouter keys later and assign a model per task (chat, planning, image generation, and others).
Can I use it on my phone?
Yes. The site works in mobile browsers, and the iOS/Android app is a native shell around the same workspace.
How do voice and calls work?
In-app live voice uses Gemini Live (default) or Ultravox. Phone tasks go through Twilio and use the same agent tools, knowledge, and MCP connections as chat.
Can I share a conversation?
Yes. Make a chat public and copy the /share/chat/… link. Anyone with the link can read it and start their own copy.
What happens to my data?
Chats are private by default. Shared links are opt-in. You can use your own provider keys so generation runs on your accounts.