Skip to main content
Memory is the Slate agent’s durable mind. It’s where the agent keeps what it has learned about your brand, your priorities, and what works — written as documents you can read, edit, and approve.

​
Where to Find It

Go to Assets → Memory in the sidebar. The page header shows the agent, its autonomy mode, when it was last active, and a search box (“Search its mind…”).
Memory page showing the agent header with the Today and Files tabs

The agent's Memory

​
Leaving a Note for the Agent

Click Add a note / ask at the top to leave a correction, instruction, or request. The agent reads it on its next wake-up. Notes are logged into its Commitments — the agent’s working memory of open loops.

​
The Two Tabs

​
Today

A daily view of what the agent is doing and what needs you, in five sections: Needs your attention, Activity timeline, Tasks, Results, and Autonomy & controls. From here you can approve or reject the agent’s proposals, snooze or dismiss attention items, and adjust its autonomy mode. When nothing is pending, it says so — “Nothing needs you.”

​
Files

The memory itself: markdown documents the agent maintains, with revision history and your approval on changes. Three doctrine files anchor the mind: Around them, folders hold the agent’s research, initiatives, entity notes (pages, competitors, keywords, topics), journal, and inbox.

​
How Memory Gets Written

  • The agent proposes changes from its runs and scheduled maintenance. Proposed changes appear as Pending change with Approve / Reject buttons and a diff (See what changes).
  • You edit directly — open any file, click Edit, make changes, and add a one-line change note.
  • Your notes from the composer land in Commitments.
Every file keeps a History of revisions — view the document at any point, see what changed, and jump to the run that wrote it. If the mind is empty, Slate offers to Draft the mandate — a starting mission you review and approve.

​
Removing Memories

Files are archived, not hard-deleted: Discard moves a file to archived and it stops informing the agent. Edit and save an archived file to restore it.

​
What’s Next

  • Cowork — where the agent does the work its memory guides
  • Skills — reusable procedures the agent can load on demand
  • Brand Kit — the structured brand facts agents draw on