This Week in Neo4j: Ontologies, Agent Memory, GraphRAG, Visualisation and more

Neo4j released AgentMemory for .NET, bringing graph-native short-term, long-term, and reasoning memory to Microsoft Agent Framework and Semantic Kernel developers, and published recordings from its Graph Track at AI Engineer World's Fair, including talks on ontology-based semantic layers and agentic system failures. The company also introduced Bloom's whole-graph projections for running algorithms across entire databases and highlighted a GraphRAG build with Spring AI and Goodreads data.

This Week in Neo4j: Ontologies, Agent Memory, GraphRAG, Visualisation and more Staff Community Manager 4 min read Welcome to This Week in Neo4j, your fix for news from the world of graph databases The Neo4j Graph Track from AI Engineer World’s Fair is now up on YouTube – hear from Emil Eifrem, Neo4j about “Thinner Agents on a Smarter Substrate: The Ontology-based Semantic Layer”, Subbiah Sethuraman and Abhilash Asokan, ZS Associates on “Why We Killed Our Multi-Agent Pipeline” or hear Frank Coyle’s talk on why agentic systems fail without a formal ontology sitting outside the model as a logical guardrail. Also this week: AgentMemory for .NET brings graph-native short-term, long-term and reasoning memory to Microsoft Agent Framework and Semantic Kernel developers; a hands-on GraphRAG build with Spring AI, Neo4j and Goodreads data that flags the real wiring gotchas; and Bloom’s new whole-graph projections, letting you run algorithms across your entire database instead of just what’s on screen. Share your experiences and influence the future of Neo4j products: Join the Neo4j User Research panel https://message.neo4j.com/NzEwLVJSQy0zMzUAAAGZN8qvbggSloC9gXkMyg5KMvlV10ryIrNUXCa QC8dLTLXTjPbtfIiUvrsfzG2kLxc3ReSbQM= It’s a chance to connect directly with product development teams, get paid compensation, hear about what we are working on and more Happy Graphing, Alexander Erdl COMING UP Livestream : G.A.M.E.R.S: Graph Agents with Multimodal Entities and Reasoning Schemas https://youtube.com/live/8otCxnotd2E on August 11 Conferences : Find us at testµ, virtual https://www.testmuai.com/testmuconf-2026/ on August 20-21, Agentsnexus, Bengaluru https://agentsnexus.io/ on September 4-5, AGNTCon + MCPCon, Tokyo https://events.linuxfoundation.org/agntcon-mcpcon-japan/ on September 10-11 & Big Data, Paris https://www.bigdataparis.com/ on September 15-16 Meetup : Meet us in Berlin, DE https://globalai.community/e/0pqyb95l on August 25, Tokyo, JP https://jp-neo4j-usersgroup.connpass.com/event/400616/ on August 26, Tokyo, JP https://luma.com/tokyo-hack-night-08-27-26 on August 27 & Pune, IN https://www.meetup.com/graphdb-pune/events/315585209/ on August 29 All Neo4j Events : Webinars and More https://neo4j.com/events/ FEATURED COMMUNITY MEMBER: Hamza Farooq https://www.linkedin.com/in/hamzafarooq/ Hamza Farooq https://www.linkedin.com/in/hamzafarooq/ Hamza is a Founder by day and a Professor by night. His work revolves around the realms of NLP and Multi-Modal Systems. Connect with him on LinkedIn https://www.linkedin.com/in/hamzafarooq/ . Just this week, he was teaching in a free online series: Forward Deployed Engineering Series https://maven.com/lls/69dbb0 where they covered how to design AI systems, communicate technical ideas through visual storytelling, use Knowledge Graphs and GraphRAG, engineer reliable context pipelines and apply proven architectural patterns used in production. ONTOLOGY: Neo4j Graph Track @ AI Engineer https://www.youtube.com/watch?v=Sir59K8ZDPU&list=PLKLSkgrxaJAU&index=7 Recordings from the Graph Track during AI Engineer World’s Fair have now been published and you can watch them all. One particularly popular talk was “Why Agentic Systems Need Ontologies” by Frank Coyle at UC Berkeley. He argues that most agent failures, from brittle tools to fragile handoffs, are symptoms of one missing layer: a formal ontology sitting outside the model as logical guardrails. LLMs reason probabilistically over domains they only half understand, and no amount of prompt engineering closes that gap. AGENT MEMORY: AgentMemory for .NET: A native sibling to Neo4j Agent Memory https://neo4j.com/blog/developer/agentmemory-for-net-a-native-sibling-to-neo4j-agent-memory/ José Latorre and Zaid Zaim built a .NET implementation of Neo4j Agent Memory https://neo4j.com/labs/agent-memory/ – bringing graph-native short-term, long-term and reasoning memory to Microsoft Agent Framework and Semantic Kernel developers who’ve had thin pickings compared to the Python ecosystem. It’s independently verified against Neo4j’s own Test Compatibility Kit, and pairs with AgentEval, a companion .NET toolkit for benchmarking memory quality over time. GRAPHRAG: GraphRAG in Practice Using Spring AI, Neo4j, and Goodreads Data https://dzone.com/articles/graphrag-spring-ai-neo4j In this hands-on walkthrough, Akmal Chaudhri builds a full GraphRAG pipeline with Spring AI and Neo4j, using a Goodreads books-and-reviews dataset to combine vector search with graph traversal. Beyond the working code, it flags the real gotchas that trip people up when wiring this up for the first time – like Spring AI’s default index naming and required node properties. GRAPH VISUALISATION: Seeds of Science, Visuals in Bloom: Introducing Whole-Graph Projections https://medium.com/neo4j/seeds-of-science-visuals-in-bloom-introudcing-whole-graph-projections-7bcbfe62ab5c Bloom can now run graph algorithms across an entire database, not just the current Scene – a big deal if you’ve ever been limited to local, on-screen views for centrality or community detection. Jeff Gagnon walks through the difference with a concrete example pepper vs butter as recipe ingredients , showing how whole-graph results can tell a very different story than a local one. STARTUPS: Icite https://www.linkedin.com/posts/wesmullins great-conversation-on-friday-with-brian-okeefe-share-7484960411734130688-8sN9/ Icite is a multi-data-source identity threat detection and response ITDR platform built on a simple insight: identity is multi-surface, spanning event logs, configuration state, and the relationships between people, permissions, and systems that a graph is uniquely suited to model. CONTINUOUS LEARNING GraphAcademy : Build a Product Recommendation Dashboard on Aura https://graphacademy.neo4j.com/courses/aura-no-code-project – no code required Learn on Your Schedule : Go deeper into graph intelligence on Neo4j’s On-Demand webinar library https://neo4j.com/events/on-demand/? event level=intermediate Workshops : Join our virtual classrooms workshops from Fundamentals to GenAI https://neo4j.com/events/? event type=workshops New Webinar : Build trustworthy AI: A fireside chat with IndyKite – Americas https://go.neo4j.com/WBR-CON-260813-IndyKite-fireside-chat Registration.html , Europe, Middle East & Africa https://go.neo4j.com/WBR-CON-260813-Fireside-Chat-IndyKite-EMEA Registration.html , Asia Pacific https://go.neo4j.com/WBR-CON-260813-Fireside-Chat-IndyKite-APAC Registration.html POST OF THE WEEK: Sipirtu https://bsky.app/profile/sipirtu.com Ontologies are resurfacing as a foundation for agentic AI, with UC Berkeley and Neo4j highlighting their role in providing structured, logical frameworks for reasoning. Source: Latent Space — Sipirtu @sipirtu.com 30. Juli 2026 um 15:10 Please share it if you like it 051230