arXiv cs.AIWednesday · May 27, 2026FREE

Helicase: Uncertainty-Guided Supply Chain Knowledge Graph Construction with Autonomous Multi-Agent LLMs

Helicase, introduced in a new arXiv paper (2605.26835), addresses the challenge of structural inference in supply chains, where answers require synthesizing information from multiple fragmented web sources. Unlike simple one-shot queries, questions like "Which Tesla components use lithium from Australian mines?" demand multi-hop reasoning. Helicase uses an autonomous multi-agent LLM framework to decompose high-level queries into executable investigation plans, dynamically constructing and analyzing knowledge graphs. A key feature is uncertainty-awareness: decisions are based on calibrated confidence in answer reliability, traceable to source quality and reasoning consistency. The system coordinates multiple LLM agents to retrieve, verify, and integrate information, providing traceable evidence for each inference. This approach aims to automate complex supply chain discovery tasks that currently require significant human effort.