Building an AI Agent with Memory: Microsoft Agent Framework + Neo4j

The article explains how to create a conversational AI agent with persistent memory using the Microsoft Agent Framework (MAF) and Neo4j, illustrated through the development of the TfL Explorer, a London transport assistant. This AI agent can remember user preferences and previous interactions, enabling it to provide more personalized and context-aware responses. The framework integrates the neo4j-agent-memory library, allowing the agent to access a shared Neo4j instance that stores both transport data and the agent's memory. This integration utilizes a pluggable architecture that combines various model providers and memory backends, such as Microsoft Foundry, Azure OpenAI, and Neo4j, to manage short-term, long-term, and reasoning memories. The article highlights the architectural benefits of using a single database for domain data and memory, which simplifies operations and enhances the agent's reasoning capabilities by linking memory entities to domain nodes. It also details how to set up the memory client, the types of memories managed, and the implementation of custom tools using the transport graph to execute Cypher queries directly within Neo4j. The agent's memory operations are non-blocking, and its tool calls return structured data, facilitating visualization in the frontend through Mapbox GL JS and Neo4j NVL.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.