An agent can remember the current conversation and still lose the customer the moment the session ends.
AI After 5 and MongoDB are hosting a New York build lab on November 17 around durable memory for AI agents. Samadnya Kalaskar, Senior Solutions Architect at MongoDB, will guide participants through a browser-based lab using LangGraph and Atlas Vector Search.
Memory changes what an agent can carry from one interaction into the next. It also changes the risk. A useful system has to decide what belongs in the record, what should expire, how the next retrieval is grounded and which user is allowed to recover the stored context.
The lab moves those questions into implementation. Participants will work with state, retrieval and longer-running agent behavior rather than treating memory as a larger prompt.
A laptop is required. Registration is free and approval-based. The exact venue details are supplied through the official registration process.
2026-11-17; 5:30-8:30 PM ET.
MongoDB New York, New York.