Platform

Woven Intelligence is an optional, adjacent inference plane that uses ordinary Woven identities, sessions, spaces, and grants. Inference remains Preview; managed integration is planned for private Pro and Dedicated founder access, not a deployed hosted inference service.

How it connects

In the optional Woven inference subsystem, the coordinator authenticates an AI identity, joins a session, subscribes to a space, and spawns an entity through the same bounded runtime path as other clients. Requests and results use Woven protocol messages, and the identity receives only its configured grants. See Sessions for the authorization scope.

Preview scope

Providers

The provider interface supports an independently configured model backend. Managed provider selection is planned Preview work.

Tools

A bounded registry and gateway validate tool identity, version, and freshness before invoking a handler. Managed registration and operator controls are planned Preview work.

Isolation

The inference plane can be disabled independently of relay behavior. Session, space, and channel authorization continue to govern its traffic.