Skip to main content
Cortex has five nouns. Understand these and the rest of the product explains itself.

Brain

A brain is one self-contained memory: its own pages, its own inbox, its own members. You start with a brain of your own, My Work Brain, whose only member is you, and you can belong to shared brains your team builds together. Who can see a brain is exactly who is on its member list. More about brains.

Page

A page is one filed piece of knowledge: readable text about a person, a meeting, a decision, a topic. Pages link to each other, carry citations back to their sources, and keep their full version history, so nothing is ever lost to an edit. More about pages.

Inbox

Every brain has an inbox, and everything arrives there first: uploads, meeting transcripts, agent conversations. Nothing writes straight into the brain; items wait in the inbox until they are filed, by Cortex, by your agent, or by you. More about the inbox.

Skill

A skill is a filing instruction: how to read a certain kind of item (a meeting, a document, a conversation), where to put what it contains, and how to record where it came from. Your connected agents follow skills when they file things, so filing stays consistent no matter who, or what, does it.

Connection

A connection links an AI agent to your workspace, acting as you. Through it the agent reaches the brains you can reach, with the roles you actually hold, checked live on every call, and it says which brain everything came from. You approve each connection on a screen that names the workspace, and you can revoke any single connection at any time without touching the others. More about connections.

The promise, restated

Access in Cortex is the member list, and nothing else. Your own brain starts with one member, you, and stays that way unless you choose otherwise. Shared brains only ever receive what someone with Write access deliberately put there, and every write names its destination. Your agents can never do more than you can.