Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Can LLM Agents Be CFOs? A Benchmark for Resource Allocation in Dynamic Enterprise Environments

Yi Han, Lingfei Qian, Yan Wang, Yueru He, Xueqing Peng, Dongji Feng, Yankai Chen, Haohang Li, Yupeng Cao, Jimin Huang, Xue Liu, Jian-Yun Nie, Sophia Ananiadou

Abstract

Large language models (LLMs) have enabled agentic systems that can reason, plan, and act across complex tasks, but it remains unclear whether they can allocate resources effectively under uncertainty. Unlike short-horizon reactive decisions, allocation requires committing scarce resources over time while balancing competing objectives and preserving flexibility for future needs. We introduce EnterpriseArena, the first benchmark for evaluating agents on long-horizon enterprise resource allocation. It instantiates CFO-style decision-making in a 132-month enterprise simulator combining firm-level financial data, anonymized business documents, macroeconomic and industry signals, and expert-validated operating rules. The environment is partially observable and reveals the state only through budgeted organizational tools, forcing agents to trade off information acquisition against conserving scarce resources. Experiments on eleven advanced LLMs show that this setting remains highly challenging: only 16% of runs survive the full horizon, and larger models do not reliably outperform smaller ones. These results identify long-horizon resource allocation under uncertainty as a distinct capability gap for current LLM agents.

Can LLM Agents Be CFOs? A Benchmark for Resource Allocation in Dynamic Enterprise Environments

Abstract

Large language models (LLMs) have enabled agentic systems that can reason, plan, and act across complex tasks, but it remains unclear whether they can allocate resources effectively under uncertainty. Unlike short-horizon reactive decisions, allocation requires committing scarce resources over time while balancing competing objectives and preserving flexibility for future needs. We introduce EnterpriseArena, the first benchmark for evaluating agents on long-horizon enterprise resource allocation. It instantiates CFO-style decision-making in a 132-month enterprise simulator combining firm-level financial data, anonymized business documents, macroeconomic and industry signals, and expert-validated operating rules. The environment is partially observable and reveals the state only through budgeted organizational tools, forcing agents to trade off information acquisition against conserving scarce resources. Experiments on eleven advanced LLMs show that this setting remains highly challenging: only 16% of runs survive the full horizon, and larger models do not reliably outperform smaller ones. These results identify long-horizon resource allocation under uncertainty as a distinct capability gap for current LLM agents.
Paper Structure (52 sections, 3 equations, 7 figures, 7 tables)

This paper contains 52 sections, 3 equations, 7 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Financial Agent Benchmarks.
  4. Agent Environment Benchmarks.
  5. EnterpriseArena
  6. Dynamic Layered Environment
  7. Internal enterprise dynamics.
  8. External economic and industry dynamics.
  9. Information Acquisition via Organizational Operation Tools
  10. Tool budget.
  11. Trade-off Action with Environment Interaction
  12. book_closing.
  13. fund_raising_request.
  14. pass.
  15. Dataset Curation and Construction
  16. ...and 37 more sections

Figures (7)

  • Figure 1: Overview for EnterpriseArena Benchmark.
  • Figure 2: Aggregate fundraising actions taken by agents by timestep, averaged over 5 runs.
  • Figure 3: Aggregate book closing actions taken by agents by timestep, averaged over 5 runs.
  • Figure 4: Cash by timestep, averaged over 5 runs.
  • Figure 5: Failure Analysis 1: GPT-5.4
  • ...and 2 more figures