Reliable agents: memory, execution state and recovery
Tokyo AI evening on agent memory, execution state and recovery, with talks from Temporal, Zilliz and a Neo4j/MCP shared-memory demo.
- When
- Wed, October 14, 2026 · 18:00–21:00 JST
- Where
- Tokyo, Japan · In person
- Region
- Kanto (Tokyo)
- Organizer
- Tokyo AI
- Language
- EN
- Source
- Luma
Summary
A Tokyo AI (TAI) evening on building reliable AI agents, focused on the gap between what an agent remembers and the state of its work: what to retain for reasoning, what to record reliably, and how to resume after interruption. It is aimed at engineers building and operating agents, application architects, and technical leads.
Melanie Warrick (AI & DevRel Engineer, Temporal) presents "What an Agent Needs to Remember" with a live demo on durable execution state. Yulan Yan of Zilliz covers agent memory patterns from production AI applications, and Koji Annoura shows how to share persistent context across agents with Neo4j and MCP. Each talk includes Q&A, followed by an hour of networking. The venue in Tokyo is still to be announced.
About the community
Tokyo AI (TAI) is a large English-language AI community in Tokyo that brings together engineers, researchers, and founders. It runs frequent technical evenings built around a focused theme, typically a few expert talks with Q&A followed by networking, covering topics from agents and LLM infrastructure to applied AI in specific industries.
#ai-agents#agent-memory#durable-execution#llm#neo4j#mcp#reliability#temporal#llm-engineering#networking