AWS re:Invent 2025 - Beyond a Chatbot: How to Build Intelligent Agent Memory (AIM284) | AWS Events

The podcast is an educational session focusing on agent memory and its importance in improving the quality of AI agents and RAG pipelines. The speaker discusses the limitations of relying solely on large language models (LLMs) for AI systems, emphasizing that context is not memory and that proper memory engineering is crucial for reliable, long-horizon agent behavior. The session covers the anatomy of an agent, different application modes (assistant, workflow, and deep research), context engineering techniques, and the hierarchy of memory types (short-term, long-term, and shared memory). The speaker also touches on evaluating memory systems and common challenges, advocating for a shift towards designing systems that accumulate experience for more intelligent agents.

Outlines

Part 1: Problems, Context

Part 2: Agent Anatomy, Context vs. Memory

Part 3: Memory Engineering, Lifecycle

Part 4: Implementation, Evaluation

Sign in to continue reading, translating and more.

Open full episode in Podwise

AWS re:Invent 2025 - Beyond a Chatbot: How to Build Intelligent Agent Memory (AIM284)

AWS Events

Part 1: Problems, Context

Introduction to Agent Memory and Its Importance

Limitations of LLMs and the Need for External Memory Systems

Part 2: Agent Anatomy, Context vs. Memory

Context vs. Memory and the Anatomy of an Agent

Multi-Agent Systems and Key Agent Characteristics

Application Modes, Context Engineering, and Retrieval Pipelines

Part 3: Memory Engineering, Lifecycle

Evaluating Context Engineering and Introduction to Memory Engineering

Memory Lifecycle and Types: Short-Term, Long-Term, and Shared

Part 4: Implementation, Evaluation

Deep Dive into Memory Types: Short-Term, Long-Term, and Shared

Evaluating Memory Systems, Addressing Challenges, and MongoDB's Role

AWS re:Invent 2025 - Beyond a Chatbot: How to Build Intelligent Agent Memory (AIM284)

AWS Events

Part 1: Problems, Context

00:00Introduction to Agent Memory and Its Importance

Introduction to Agent Memory and Its Importance

05:02Limitations of LLMs and the Need for External Memory Systems

Limitations of LLMs and the Need for External Memory Systems

Part 2: Agent Anatomy, Context vs. Memory

11:31Context vs. Memory and the Anatomy of an Agent

Context vs. Memory and the Anatomy of an Agent

17:47Multi-Agent Systems and Key Agent Characteristics

Multi-Agent Systems and Key Agent Characteristics

24:53Application Modes, Context Engineering, and Retrieval Pipelines

Application Modes, Context Engineering, and Retrieval Pipelines

Part 3: Memory Engineering, Lifecycle

31:47Evaluating Context Engineering and Introduction to Memory Engineering

Evaluating Context Engineering and Introduction to Memory Engineering

37:54Memory Lifecycle and Types: Short-Term, Long-Term, and Shared

Memory Lifecycle and Types: Short-Term, Long-Term, and Shared

Part 4: Implementation, Evaluation

44:27Deep Dive into Memory Types: Short-Term, Long-Term, and Shared

Deep Dive into Memory Types: Short-Term, Long-Term, and Shared

50:02Evaluating Memory Systems, Addressing Challenges, and MongoDB's Role

Evaluating Memory Systems, Addressing Challenges, and MongoDB's Role