Show HN: FAB – A benchmark for AI agents doing financial due diligence

Category: ai-ml

Tags: benchmark, llm-agents, finance, evaluation, dataset

Score: 6.3/10 (Innovation: 6, Technical: 6, Documentation: 7, Utility: 6)

FAB is a benchmark dataset and execution harness for evaluating LLM agents on financial due diligence tasks using a synthetic company data room. It provides 50 tasks, 160 documents, and 231 grading criteria, with published results and agent traces, making it a practical resource for evaluating agentic reasoning in finance.

Target audience: AI researchers, ML engineers, financial analysts evaluating agents

Repository: https://github.com/SecondState-ai/finance-agents-benchmark · Python · MIT · 5 stars

View on Hacker News