Enter to send · Shift+Enter for new line
Documents
Upload a document, then ask questions grounded in it.
Upload & index
Search your documents
Editor
Write a document, then ask AI to edit it. Private to you.
Saved documents
Memory
Things you ask the assistant to remember about you.
Stored memories
Notes & Tasks
Jot a note or track a task. Private to you.
Your items
Agent
Give the agent a task, or run deep research over your documents.
Compare
Run one prompt against two models, blind, then reveal.
Media
Transcribe audio, generate an image, or synthesize speech.
Transcribe
Generate image
Text to speech
Calendar & integrations
Call a connected external service (e.g. Google Calendar).
Tools (MCP)
List and invoke tools from a per-tenant MCP server.
Models
Serve and manage models. GPU capacity is scale-to-zero — the first request wakes it.
Served models
Add a model from Hugging Face
Enter a Hugging Face repo id to request installation. An operator reviews and applies it.
Your install requests
Hugging Face configuration
The token is stored write-only in the cluster secret store and consumed by the GPU inference plane; the control plane never returns it.
Settings
Preferences and workspace status.