Hornet.dev
Hornet.dev is a technology company specializing in search infrastructure, notably focusing on the integration of traditional lexical search algorithms with modern Large Language Models (LLMs).
Key Concepts & Research
- Agentic Search: Defined as “search inside an agent loop,” where retrieval is dynamically integrated into the reasoning process of an AI agent.
- BM25 Resurgence: Highlighted in recent analysis by CEO jo-kristian-bergum, demonstrating the “unreasonable effectiveness” of the 30-year-old BM25 lexical scoring function in contemporary LLM-driven contexts.
- Core Insight: Despite the rise of dense vector embeddings, BM25 remains critical for precision and recall in agentic workflows, often outperforming purely semantic approaches in specific retrieval tasks.
Related Notes
References
- Jo Kristian Bergum. “The unreasonable effectiveness of BM25 for agentic search.” BM25’s Unreasonable Effectiveness in LLM-Driven [concepts/agentic-search|Agentic Search]