Building Evolving AI Agents Via Dynamic Memory Representations Using Temporal Knowledge Graphs
summary
Claims a practical memory architecture that layers Neo4j + LangGraph + Graphiti to store evolving agent memory, extracts entities with Pydantic entity schemas, and models time with bitemporal validity intervals. Demonstrates a toy build: two studies (2015 vs 2019) produce a relationship invalidated after 4 years while system ingestion time shows a 2‑second lifespan, and a chatbot user node (fictitious Sarah Chan) writes interactions into the graph for session personalization.
AI-written summary · about
chapters (7)
- 0:00Introduction and motivation
- 1:27Dynamic memory architectures
- 3:33Guidance design for memory
- 4:35Modeling temporal state changes
- 6:24Temporal knowledge graph example
- 11:42User behavior and interaction
- 13:27Open questions on time representation
related talks
51mMemory Masterclass: Make Your AI Agents Remember What They Do!
21mBeyond Recall: Creating Agentic Personas with Persistent, Evolving Memory
29mThe AI Agent Memory Landscape
31mTemporal Substrate Architecture: Building Persistent Identity for Autonomous Agents
21mMulti-Agent Shared Graph Memory: Building Collective Knowledge for Agents
18mArchitecting Agent Memory: Principles, Patterns, and Best Practices
46mDesigning Semantic Memory for Multi-Agent Systems with Python
26mStore Your AI Agent's Memory and Context in a Knowledge Graph