Can we control which models the team uses?

Yes: workspace governance can pin allowed models and providers, and routing follows it for chat, agents, and the gateway API alike.

What can an agent do without approval?

Whatever the dial allows: read-only strips mutating tools entirely, ask mode confirms consequential actions, and full mode still runs under budget caps and traces.

Where do shared credentials live?

In the encrypted secrets vault: members and agents use them without reading them, and every reveal is logged.

All use cases

Team AI governance

Give the team AI, keep the controls.

Workspace roles, scoped agent permissions, an encrypted secrets vault, and full traces.

Try now See pricing

Set the boundary

One workspace per team or client; owner, admin, editor, and viewer roles scope what each member touches.

More about Workspaces

Scope the agents

Per-folder file access, per-computer permissions, and an autonomy dial from read-only to full.

More about Agents

Lock down credentials

Shared secrets members can use but never read, with every reveal logged.

More about Secrets

Keep the paper trail

Runs carry full traces, and share links are scoped and revocable, so review is reading, not archaeology.

More about Sharing

Built from these features

Workspaces

Shared folders, agents, computers, and chats: one container.

Agents

Skills, scoped permissions, persistent memory: any model you pick.

Secrets

Store API keys, tokens, and passwords: encrypted, never shown to agents.

Sharing

Public URLs for files, agents, and chats: granular, revocable.

Frequently asked

Ready to work this way?

Create a free account and start in seconds.

Try now