Reproduce “FORCE-Bench: A Benchmark, Dataset, and Evaluation Harness for Agentic AI in Enterprise Finance” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Wolfgang M. Pauli, Sarah Panda, Kidus Admassu, Said Bleik, Ademola Okerinde, Jeremy Reynolds

FORCE-Bench evaluates finance agents across three workflows using 251 expert-annotated queries and eight finance-oriented response-quality dimensions. It scores assertions from 0 to 1, with partial credit, while weighting task types equally. Accuracy is assessed only for ERP obligation research; the Finance Agent scores higher than tested general-purpose systems on this measure. The comparison reflects this benchmark and setup, not every finance task or deployment.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.