Agent Memory: How to Build Agents that Learn and Remember
AI Summary
Letta's July 2025 deep dive on the three-layer memory model used in production agents: message buffer for recent turns, editable in-context memory for stable identity and project state, and external archival storage for everything else. The post works through when to read vs. write memory, how to handle memory pressure (what MemGPT called overflow), and why most teams under-invest in memory editing and over-invest in retrieval. It also lays out the agent-learns-over-time loop as a sequence of specific tool calls rather than hand-waving. Useful as a reality check after reading the academic papers — the production answer is messier than the benchmark setup suggests.
Original excerpt
[Letta Code Run agents locally inside your terminalLetta API Build agents into your apps with our APIResources
Blog Learn about product and research updatesCustomer Stories Read about Letta in productionDemos See Letta in actionModel Leaderboard Understand which LLMs work bestDeveloper Community Join the Letta community on DiscordCompany
About us Learn about our mission and teamCareers Join our team to work on open AIContact us Get in touch
Letta Developer Platform Use the Letta API to build agents that can actually remember and learn about your users over time. Open source, production ready, and fully model-agnostic.
Letta Code Letta Code is a memory-first coding harness, built on top of the…
18 more articles in this vault.
Import the full Agent Memory Patterns vault to Burn 451 and build your own knowledge base.
Content attributed to the original author (Letta). Burn 451 curates publicly available writing as a reading index. For removal requests, contact @hawking520.