/ About this event
Expect deep technical talks. No sales pitches.
Everyone is building agents on top of Slack, Notion, docs, and meeting recordings. And everyone is hitting the same wall: the agent works in the demo, then forgets what it learned, loses its place mid-task, or burns money re-reading the same screenshots and PDFs on every run.
This night is about what agents actually need to hold onto. Where you keep the state so a long job survives a crash. Where you keep the memory so the next run starts smarter than the last. And what happens when that memory has to cover more than text, from video to screens to logs.
Speakers from GMI Cloud, FalkorDB, and more, plus the teams shipping company-brain products in production.
Come if you're building agents that run longer than one conversation, and you're tired of watching them start over.
Agenda
5:00 PM | Doors open, food and drinks 5:30 PM | Welcome from The Multimodal Society 5:45 PM | Talk 1: Long-Running Agents Are a Compute Problem: Serving Memory-Heavy Workloads Without the Bill - Roan Weigert, DevRel AI Lead, GMI Cloud 6:15 PM | Talk 2: Under the Hood: Making Graph Memory Fast Enough for Agents in the Loop - Liz Zhang, DevRel Lead, FalkorDB 6:45 PM | Talk 3: Agent Memory: What to Persist, What to Forget, and How to Retrieve It - Vasilije Markovic, Founder, Cognee 7:15 PM | Talk 4: Memory for Coding Agents: A Persistent Graph of Code, Decisions and Reasoning 7:45 PM | Networking 8:30 PM | Close
Interested in speaking/partnering? Fill out a form here: tally.so/r/q4yjpO
This event is a part of #SFTechWeek—a week of events hosted by VCs and startups to bring together the tech ecosystem. Learn more at www.tech-week.com.
