AI Agents Seattle September
We will have 2 talks: 1. Graph Memory for Agents - from GraphRAG, to a graph of everything! | Jeremy Adams @ Neo4j Knowledge graphs have been earning their keep in recommendation, fraud detection, and analytical systems for almost two decades. With the rise of generative AI and agents, graphs are back in the spotlight as one of the best structures for delivering grounded context and memory to LLMs. One of the reasons for this is that unlike black-box methods like fine-tuned LLMs and vector search, graph retrieval is transparent and explainable. Beyond that, used correctly, graph's boost accuracy of generated responses compared to semantic search alone. And of course they feel familiar: quite similar to how we often organize information in a mind map or wiki. We'll dive into how graphs can represent knowledge, memory, and even become a map of your whole organization. 2. Codemode - The only tool your agent needs | Avery Townsend & Aaron Villalpando @ Boundary (creators of BAML) Tool calling is great, but still deterministic. Agent are amazing at writing code, so what happens, if an agent just creates its own tools on the fly? This is exactly how claude ultracode works. This talk will share how to do this, the dangers, and the safeguards you can put in place, so your agents works like magic. Speakers: Jeremy Adams (Sr Developer Advocate @ Neo4j Jeremy's technical and strategic streaks have taken him from startups to tech giants, in roles bridging technology, community, and customers. These days, he's focused on the intersection of developers, graphs, and AI. He's passionate about people, entrepreneurship, integration, and automation. Through roles at Neo4j, Dagger, GitHub, Twistlock, and Puppet, Jeremy has spent a lot of time zooming in and out—from hands-on technical details to architecture, strategy, and business. Along the way, he's developed an appreciation for how the pieces fit together, and for what it takes to turn technology into something people can actually