Skip to main content
Memories are the shorter-lived counterpart to knowledge: your facts, preferences, observations, decisions-in-progress, and project context. They give your agents continuity — what you told them yesterday still counts today. Where: Sidebar › Knowledge › Memories.

​
Memories vs Knowledge

​
The Memories view

A standard table view (see Data Views for all table controls). Memories have their own add button and a dedicated edit dialog with the same fields as the chat confirm card: The edit dialog also offers Delete (“This cannot be undone.”) — the one place a memory is removed for good.

​
Creating memories

  • In Chat: /save (facts and preferences), /stage (active project context), /note (quick observation), or Save as Memory under any reply. Every save shows a confirm card first.
  • By agents: agents with the Save Memory permission record what they learn while working.

​
How agents use memories

Like knowledge: high-importance memories auto-inject, agents can be pinned to specific memories, and everything else is searchable on demand (/recall). An agent’s memory access scope (Just me / My department / Everything) is set per agent — see Agents. Memories also feed the learning loop: the nightly self-learning sweep reviews memories and may propose promoting one into permanent knowledge or a skill — those proposals land in Skills Review and the admin Inbox.

​
Permissions

Your memories are yours: personal-scope memories are visible only to you (admins manage the system but the visibility picker controls sharing). Department and company scopes share a memory deliberately.