{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,11,16]],"date-time":"2026-11-16T16:37:12Z","timestamp":1794847032537,"version":"4.9.1"},"reference-count":0,"publisher":"IOS Press","isbn-type":[{"value":"9781643686707","type":"electronic"}],"license":[{"start":{"date-parts":[[2026,7,2]],"date-time":"2026-07-02T00:00:00Z","timestamp":1782950400000},"content-version":"unspecified","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by-nc\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,7,2]]},"abstract":"<jats:p>AI is increasingly used in legal workflows, but trusted deployment requires reliable performance across heterogeneous tasks. Single-model deployment is brittle: smaller models can be efficient on retrieval-style extraction, while larger models are stronger on interpretive reasoning but costly. We present Legal Adaptive Multi-Model Router (LAMMR), a routing architecture that combines rule-based signals (intent cues, document-length heuristics and confidence fallbacks) with data-driven routing, plus a task-aware evaluation layer that scores extraction with overlap-oriented metrics (F2\/Jaccard), and reasoning with LegalBench-aligned correctness scoring. Across 1,093 tasks (999 reasoning, 94 extraction) and 2,186 model-level evaluations, a single model architecture based on Llama Maverick attains extraction\/reasoning scores of 0.2988\/0.6425, while learning-only LAMMR reaches 0.3978\/0.6578. This corresponds to +33.15% extraction, +2.40% reasoning, and +3.70% overall improvement versus Llama Maverick. These results indicate that routing policy is associated with improved legal-AI accuracy and reliability, with cost treated as a secondary operational consideration.<\/jats:p>","DOI":"10.3233\/faia260493","type":"book-chapter","created":{"date-parts":[[2026,7,2]],"date-time":"2026-07-02T09:56:49Z","timestamp":1782986209000},"source":"Crossref","is-referenced-by-count":0,"title":["Learning and Reasoning for Legal AI via Intelligent Adaptive Multi-Model Routing with a Smart Evaluation Layer"],"prefix":"10.3233","author":[{"given":"Aditya","family":"Mukhopadhyay","sequence":"first","affiliation":[{"name":"Manipal Institute of Technology, Udupi-Karkala Rd, Eshwar Nagar, Manipal, Karnataka 576104, India"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sara","family":"Ghane","sequence":"additional","affiliation":[{"name":"DT Services and Consulting, Avenue des Cerfs 11, 4031 Li\u00e8ge, Belgium"},{"name":"The Belgian Research Lab for AI in INdustry (BRAIN), Cantersteen 12, 1000 Bruxelles, Belgium, E-mail: brain@dtsc.be"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Marco","family":"di Gennaro","sequence":"additional","affiliation":[{"name":"DT Services and Consulting, Avenue des Cerfs 11, 4031 Li\u00e8ge, Belgium"},{"name":"The Belgian Research Lab for AI in INdustry (BRAIN), Cantersteen 12, 1000 Bruxelles, Belgium, E-mail: brain@dtsc.be"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"7437","container-title":["Frontiers in Artificial Intelligence and Applications","HHAI 2026"],"original-title":[],"link":[{"URL":"https:\/\/ebooks.iospress.nl\/pdf\/doi\/10.3233\/FAIA260493","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,2]],"date-time":"2026-07-02T09:56:50Z","timestamp":1782986210000},"score":1,"resource":{"primary":{"URL":"https:\/\/ebooks.iospress.nl\/doi\/10.3233\/FAIA260493"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,2]]},"ISBN":["9781643686707"],"references-count":0,"URL":"https:\/\/doi.org\/10.3233\/faia260493","relation":{},"ISSN":["0922-6389","1879-8314"],"issn-type":[{"value":"0922-6389","type":"print"},{"value":"1879-8314","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7,2]]}}}