AI & Agentic SystemsConference45min
The Anatomy of Memory in Humans and AI Agents
The talk explores how cognitive science insights can enhance AI agent memory design. It links human memory mechanisms to challenges in building and deploying reliable, context-aware agents, highlighting neuroscience lessons, practical storage and retrieval issues, and Redis’s open-source solution for production-ready agent memory systems.
Raphael De LioRedis
Artificial intelligence does not need to copy the human brain to advance, yet cognitive science still offers useful guidance. As agentic applications grow, understanding how humans store and retrieve information can help us design agents that act with greater context and reliability.
This talk connects human memory to the practical challenges of building AI agents with memory. We will review the main types of memory in the brain, revisit a landmark case in neuroscience, and relate these ideas to how large language models process information. We will also look at the real difficulties of taking agents to production, where choosing what to store and how to retrieve it becomes the core challenge.
Participants will learn:
- How insights from cognitive science can guide agent design
- The key forms of human memory and their relevance to AI
- What neuroscience reveals about the limits of large language models
- The main obstacles of deploying agents with memory in real systems
- How our work at Redis led to an open-source, production-ready agent memory server
- Practical ways to improve memory in AI agents
This talk connects human memory to the practical challenges of building AI agents with memory. We will review the main types of memory in the brain, revisit a landmark case in neuroscience, and relate these ideas to how large language models process information. We will also look at the real difficulties of taking agents to production, where choosing what to store and how to retrieve it becomes the core challenge.
Participants will learn:
- How insights from cognitive science can guide agent design
- The key forms of human memory and their relevance to AI
- What neuroscience reveals about the limits of large language models
- The main obstacles of deploying agents with memory in real systems
- How our work at Redis led to an open-source, production-ready agent memory server
- Practical ways to improve memory in AI agents
Raphael De Lio
Raphael De Lio is a Developer Advocate at Redis and a seasoned software engineer with over eight years of experience across industries and countries. Passionate about distributed systems, he specializes in Java, Kotlin, and building scalable, high-performance software. Originally from Brazil, Raphael spent six years in Portugal before settling in the Netherlands, where he also helps organize the Dutch Kotlin User Group. He loves blending code, community, and creativity to help developers build better systems faster—and have fun doing it.
comments.speakerNotEnabledComments