Wiki / entities / wiki

Michael Ryan

Stanford CS PhD student, a contributor to the DSPy project, and one of three instructors of CS329Z: Engineering AI Agents with Diyi Yang and John Yang. Ryan taught the course's lecture on agent memory architectures, recorded on 14 October 2026.

The lecture builds on the agent-memory-types taxonomy, covers storage in files, databases and knowledge graphs, then walks through four patterns: skill libraries (Voyager), tiered memory (memgpt), reflective consolidation (Generative Agents) and memory maintenance (sleep-time-compute). Ryan's stated opinion is that a plain file system is a good default memory if you can get away with it, because the lesson from agents is to set up the simplest thing and let the model run the show.

Source: report

Linked from

Agent memory typesMemGPTParameters vs context memorySleep-time compute