Jo Kristian Bergum brings two decades of search infrastructure experience — including his tenure as Chief Scientist at Vespa — to his current role as CEO and co-founder of Hornet.dev, where he is building a retrieval engine purpose-built for AI agents. Attendees will benefit from his practitioner-level insights on why classical retrieval primitives like BM25 remain surprisingly effective for agentic workloads, and how the shift from human queries to agent-driven, multi-turn retrieval demands a fundamentally different architecture.

Recent talks (2)

Agentic Retrieval: Building Self-Optimizing Search SystemsBerlin Buzzwords 2026 · Jun 2026The Rise and Fall of the Vector DB CategoryLatent Space: The AI Engineer Podcast · Apr 2025

Public activity researched automatically · as of Jun 2026