Memory¶
Store and retrieve conversation context and persistent knowledge.
What is Memory¶
Memory stores information for AI to reference across conversations. Two types serve different purposes:
- Short-term: Stores current conversation messages as context
- Long-term: Stores persistent knowledge independent of conversations
Memory Types¶
| Type | Storage | Scope | Use Case |
|---|---|---|---|
| Short-term | Conversation rounds | Single session | Maintains context within one conversation |
| Long-term | Database categories | Cross-session | Stores reusable knowledge for future use |
Short-term Memory¶
Conversation messages automatically stored in rounds. Each message includes role (user/assistant/tool) and content.
Long-term Memory¶
Structured knowledge storage with categories. Two owner types available:
| Owner | Purpose | Access |
|---|---|---|
| User Memory | User preferences and information | Available to all agents the user talks to |
| Agent Memory | Agent-specific knowledge | Only available to that specific agent |
Next Steps¶
- Short-term Memory - Conversation context management
- User Memory - Store user preferences
- Agent Memory - Build agent knowledge base