Researchers have introduced PeopleSearchBench, a new benchmark designed to evaluate the performance of AI-powered people search platforms. This open-source benchmark assesses four platforms using 119 real-world queries across recruiting, sales prospecting, and expert search use cases. A novel Criteria-Grounded Verification pipeline ensures factual relevance by extracting explicit criteria from queries and verifying results against live web searches. The benchmark measures Relevance Precision, Effective Coverage, and Information Utility, with the specialized agent Lessie outperforming other systems by 18.5% and achieving 100% task completion. AI
IMPACT Establishes a standardized evaluation method for AI people search tools, potentially driving improvements in accuracy and utility.
RANK_REASON The cluster describes a new academic paper introducing a benchmark for AI systems. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
