Skip to content

Agent Memory

Concepts

  • Agent Memory Management - Memory management is the "engine" that enables an AI agent to move from a one-shot calculator to a persistent entity
  • The Four Memory Types of Agent Memory - The canonical taxonomy for agent memory comes from the Cognitive Architectures for Language Agents (CoALA) paper (Sumers et al
  • Long-Term Memory (LTM) Strategies - Long-term memory for AI agents spans three distinct types from the CoALA taxonomy: semantic memory (durable facts and knowledge), episodic memory (specific past experiences), and **procedur...
  • Research Papers & Technical White Papers on Agent Memory - This section curates key research papers and technical white papers that have shaped the field of agent memory management
  • Working Memory Management - Working memory (also called short-term memory) is the active state the agent is reasoning over right now — the running conversation and the scratchpad the model sees at inference time
  • Agent Memory Solutions - This page consolidates all memory solutions — open-source libraries, managed cloud services, and specialized platforms — and maps them to a Technology Radar adapted from the [Thoughtworks Radar...