AI Features
Understanding Agent Memory
Agent Memory is the shared brain your AI agents use to remember what they have learned. Browse, filter, and curate memories across contacts, opportunities, and companies from one central page.
Agent Memory is where your AI agents store what they have learned between runs. When an agent sends a follow-up email, takes a note about a contact's objection, or makes a decision about an opportunity, it can write a memory so that the next time it (or another agent) touches that record, the context is already there.
Browse and search all memories at https://app.trustpager.com/ai-agents/memory.
**What a memory is**
Each memory is a short piece of knowledge an agent has written about a specific record in your workspace. It might be something like "Pitched automated follow-up sequences, contact responded positively, mentioned their team is stretched thin" or "Opportunity stalled, waiting on board sign-off before progressing." Memories are linked to the record they are about: a contact, an opportunity, a company, or the agent itself.
**The Memory page**
The Memory page at https://app.trustpager.com/ai-agents/memory shows every memory your agents have written. From here you can:
- Search memory content to find what agents have learned about a specific topic.
- Filter by subject type: Contacts, Opportunities, Companies, Self, or Campaigns.
- Filter by visibility: Shared (the workspace shared brain) or Private (visible only to the writing agent).
- Toggle a memory between Shared and Private using the visibility control on each card.
- Retract a memory so it stops surfacing to agents. Retracted memories are kept for audit but are no longer read by any agent.
**Inline memory panels**
You do not need to open the Memory page to see what agents remember about a specific record. Each contact, opportunity, and company detail page includes an Agent Memory panel showing memories linked to that record. The same visibility toggle and retract controls are available inline.
**The visibility model**
Every memory has one of two visibility settings:
- **Private** is the default. Only the agent that wrote the memory can read it. Use this for working notes specific to one agent's reasoning.
- **Shared** means any agent in your workspace can read it. This is how agents build a collective understanding of a contact or opportunity over time. Shared memories are the workspace shared brain.
As a workspace admin you can see both private and shared memories. Team members with standard access see shared memories only.
**Retracting a memory**
Retracting a memory removes it from the pool agents draw on, so they no longer see it in their context. The memory is not deleted; it stays in the audit history so you can review what agents knew at any point in time. Use retract when an agent has learned something outdated or incorrect that you do not want influencing future behaviour.
**Permissions**
Access to Agent Memory is controlled by the Agent Memory permission in your role settings, which you can manage at https://app.trustpager.com/settings/permissions.
- Editors get read, write, and delete access.
- Viewers get read access (shared memories only).
If you have connected an AI agent via API key and it cannot access memory, reconnect it from https://app.trustpager.com/settings/api so it picks up the new memory read permission.
**Finding Agent Memory in the nav**
Agent Memory lives under the AI Agents section. Open the More dropdown in the main nav and look for AI Agents. The full AI Agents area includes Agent Hub, AI Access, Oversight, Reports, Knowledgebase, and Memory.
**Related**
For developers and AI builders connecting agents via API or MCP, see the API and MCP reference at https://trustpager.com/help-center/agent-memory-api.
