This section focuses on how agents get memories back into context — proactive vs reactive recall, query-time and tool-based retrieval, ranking, caching, and failure modes.
What you will learn
- Vector, keyword, hybrid, and fetch-based retrieval patterns
- When the agent should decide to remember versus when the system retrieves automatically
- Ranking, thresholds, multi-hop recall, and context-budget trade-offs
- APIs and tool-call designs for retrieval in agent frameworks
Chapters in this section
There are 18 chapters in this part. Open any chapter to read it on its own, or work through them in order.
- 1 Retrieval Types: Vector, Keyword, Hybrid, and Direct Fetch
- 2 Proactive Recall vs Reactive Recall
- 3 Query-Time Retrieval: Using the Current Message as a Search Query
- 4 Tool-Based Retrieval: Letting the Agent Decide When to Remember
- 5 Fetch-Based Retrieval: Retrieving Known, Bounded Memories
- 6 Multi-Hop Retrieval Across Related Memories
- 7 Retrieval for Reasoning Traces vs Retrieval for Final Answers
- 8 Similarity Thresholds and Precision Control
- 9 Ranking and Scoring Retrieved Memories
- 10 Combining Multiple Retrieval Strategies in One Query
- 11 Retrieval Latency and Its Effect on Agent Responsiveness
- 12 Caching Retrieved Memories
- 13 Retrieval for Multi-Agent Coordination
- 14 Personalized Retrieval: Scoping Results to the Right User
- 15 Retrieval Failure Modes: Over-Retrieval and Under-Retrieval
- 16 Designing Retrieval APIs for Agent Frameworks
- 17 Retrieval as a First-Class Tool Call
- 18 Balancing Retrieval Recall with Context Budget