Memory
Memory is where you see what Nexus remembers about your work, and where you can add a note for your assistants to remember. Your assistants use these memories to personalise their answers and actions.
The page is titled My Memory, with the subtitle "What APEX has learned in the context of your work." (APEX is an older product name.)
Where memory appears in Nexus
Memory shows up in several places. They are related, but different.
| Place | What it is |
|---|---|
| Memory (this page) | A personal list of your own recent memories, with an Add memory form |
| Chat context panel → Memory | The memories for the chat you have open. See Memory in chat |
| Project memory | A panel on a project's page. See Projects |
| Cortex Memory | The full memory surface with the Explorer, spaces, graph and more, with its own menu and permissions |
| Settings → Learned Rules | Memories Nexus learned from your knowledge base documents (see below) |
Open Memory
Click Memory in the left sidebar ("Agent memory and knowledge"). You see it only if you have Nexus admin permission. There is no feature switch.
If you open the link without permission you see Access Denied ("You do not have the required permissions to access this page (nexus.admin)…") and a Return home button.
The Memories tab
The first tab, Memories, lists the memories saved for you. Only your own memories appear, newest first.
Each card shows:
- The memory text. Long text (over about 220 characters) is cut short with Show more / Show less.
- A type label, such as semantic or decision.
- The topic, if it has one.
- The date it was created.
- An orange Expires (date) if it has an end date.
- An Open link to see it in the Cortex Memory Explorer.
At the bottom: "(N) memories saved · your agents use these to personalise their responses and actions."
| Situation | You see |
|---|---|
| Loading | Three grey blocks |
| Nothing saved | A database icon and "No memories yet." |
| The list could not load | The same "No memories yet." There is no error message |
Memory types
Each memory has one of eight types. These are plain-language explanations.
| Type | Typically means |
|---|---|
| Episodic | Something that happened |
| Semantic | A fact or a preference (the default for new memories) |
| Procedural | A way of doing something |
| Policy | A rule to follow |
| Temporal | Something tied to a time or a period |
| Causal | A cause and its effect |
| Decision | A choice that was made |
| Predictive | An expectation about the future |
Add a memory
Use this to tell your assistants something you want them to remember.
Click Add memory
Click + Add memory at the top right. A box titled Add memory opens.
Screenshot placeholderScreenshot to add: The Add memory box with the text area, type menu and Save and Cancel buttons. Write the memory
Type in the box ("e.g. Prefers SAP transaction codes over menu navigation"). It can be up to 4,000 characters.
Choose a type
Pick one from the menu. The default is semantic.
Save
Click Save. The list refreshes and the box closes. Cancel clears it. Save stays disabled while the text is empty.
A memory you add this way is private to you, stored in your personal project, and marked as written by you.
Open a memory
Click Open on a card to see it in the Cortex Memory Explorer (/memory/browser?detail=…).
- Memories cannot be edited or deleted from the My Memory page. They are saved as they were written.
- In Cortex, only the person who created a memory can correct or delete it. Anyone who can read it sees its history.
- Memories shared with you show "Shared by (name)", without Correct or Delete buttons. Administrators have no override.
- The creator can share a memory by reference with projects, teams or the whole organization. It stays theirs, and un-sharing never deletes it. Sharing with the whole organization asks for a second confirmation.
Who can see a memory
This page shows only your own memories and has no visibility setting. Who can see a memory made during a chat depends on that chat's memory visibility: Private, Team or Org. See Memory in chat.
Memories are only captured automatically from chats inside a project.
The Learned Rules tab
Not the same as Settings → Learned Rules
There is a separate page at Settings → Learned Rules (/nexus/settings/learned-rules), which needs Nexus admin permission. It works. It lists "Memories Nexus learned from your Knowledge Base documents — as opposed to memories saved explicitly during chat." Scroll to load more, and click the × on a card to forget it. Its empty state reads "No knowledge-base-derived memories yet."