arxivcs.AI2026-07-11
FORCE-Bench: A Benchmark, Dataset, and Evaluation Harness for Agentic AI in Enterprise Finance
Wolfgang M. Pauli, Sarah Panda, Kidus Admassu, Said Bleik, Ademola Okerinde, Jeremy Reynolds
Recent advances in large language models have accelerated deployment of agentic systems in operational finance. Existing benchmarks emphasize measuring general capabilities, instruction following, or safety, but few directly address the operational finance workflows that agentic…