Reopening your workspace Restoring your secure session and conversations…

YOUR PRIVATE AI SPACE

A little more room
to think.

Your agent can chat, search, remember, and build using the model on this machine.
Private, per-account conversations and workspace.

ACTIVE MODEL

Qwen 3.8 Flash

RUNNING ON

Orbit · local

CONTEXT WINDOW

262,144 tokens

A few places to startPICK A PROMPT ↗

Your product canvas

Ask the model to build a complete HTML experience. It will appear here as a live, interactive preview you can edit and export.

Loading apps…

HTML · EDITABLE

New coding taskLOLA · LOCAL MODEL

⌘

Build with Lola

Lola can work in this project, search and read the public web, and carry useful decisions into future tasks with persistent memory.

Loading project files…

No file selected

READ ONLY

ORBIT PREFERENCES

Settings

Choose a section to manage models, tools, and your agent fleet.

Choose a skill

Enabled

Installed skills make specialized diagrams and presentations available to the agent.

Select a file

Language model provider

Use the local Strata model or connect a hosted model. MiniMax M3 is preconfigured; the OpenAI-compatible option works with providers that expose a compatible chat-completions API.

Provider API base URL Model name API key Context window Let agents use my accessible DM and channel discussions as background context

Local Strata model selected

Model behavior

Sampling and output settings apply to hosted and local models. The loaded model selector appears when using local Strata inference.

Loaded local model Preferred local contextUsed by local Strata. Hosted models use the provider context size above. Temperature Top P Maximum output tokens

Active context windowThis is the live limit reported by Strata. A different size takes effect after Strata starts the model again.

System instructions

Deepgram voice input

Speech is sent securely from this server to Deepgram. Transcripts are added to the draft for review; nothing is sent automatically.

Deepgram API key Recognition model Language

Not configured

MiniMax speech playback

Generate natural voice audio for assistant replies and play it in the chat.

MiniMax API key Speech model Voice ID Speaking speed Read assistant replies automatically

Not configured

Settings are private to this account.

Users & access

Manage accounts, groups, sign-in access, and workspace membership. Changes are saved to the server database.

ADMIN

Select an account

Choose a user to manage access, sign-in, or generate a profile photo.

A little more room
to think.

Create a workspace

Start a direct message

Workspace people and channels

Invite by username

Workspace members

Channels

Share artifact

Share sent

Access shared item

Invite an agent to this conversation

Channel photo

Rename channel

Settings

Choose a skill

My profile

Profile details

Language model provider

Model behavior

Deepgram voice input

MiniMax speech playback

MCP servers

New MCP server

Lola

Capabilities

Model execution capacity

Users & access

Create user

Account

Select an account

Create group

Group

Select a group

Proactive intake

Add an Observe source

Sources in this workspace

Observed events

Space to think.Room to build.

Welcome back

A little more roomto think.

Choose a skill

Profile details

Language model provider

Model behavior

Deepgram voice input

MiniMax speech playback

New MCP server

Lola

Capabilities

Model execution capacity

Create user

Select an account

Create group

Select a group

Add an Observe source

Sources in this workspace

Observed events

A little more room
to think.