22nd AIAI 2026, 16 - 19 July 2026, Chania, Crete, Greece

Beyond Retrieval: An Explainable Legal Chatbot with Multi-Agent Large Language Models and Knowledge Graphs

Slany Emanuel, Buchmann Sara, Föhr Tassilo, Amling Jonas, Frummet Alexander, Lang Moritz, Scheele Stephan

Abstract:

  Large Language Models (LLMs) are transforming legal and audit practices, yet their adoption in high-stakes domains remains limited by concerns around reliability and transparency. To address these challenges, we present audit.ai, an explainable legal assistant that integrates a multi-agent LLM architecture with a knowledge graph backend. By grounding responses in structured legal knowledge, audit.ai enables transparent, verifiable reasoning. Our results hint at advantages compared to long-context LLM baselines in producing more helpful and better-supported answers. Through combined visual and logical transparency, audit.ai provides a foundation for trustworthy AI in domains where accountability is critical.  

*** Title, author list and abstract as submitted during Camera-Ready version delivery. Small changes that may have occurred during processing by Springer may not appear in this window.