Jo Kristian Bergum
Jo Kristian Bergum is an AI Engineer and the CEO of Hornet.dev. He is known for his work on agentic search architectures and the practical application of lexical retrieval methods in modern LLM pipelines.
Key Contributions & Insights
- Agentic Search Definition: Defines agentic search as “search inside an agent loop,” emphasizing the iterative nature of retrieval within autonomous systems BM25’s Unreasonable Effectiveness in LLM-Driven Agentic Search.
- BM25 Resurgence: Highlights the “unreasonable effectiveness” of BM25, a 30-year-old lexical scoring function, in the context of LLM-driven agentic search, challenging the assumption that dense embeddings alone are sufficient for all retrieval tasks.
- Hornet.dev Leadership: Leads development efforts at Hornet.dev, focusing on optimizing search capabilities for AI agents.