The More the Better? A Comparison of Single- and Multi-Agent LLM Setups on a Financial Review Task
We test a narrow form of it in a controlled comparison of single-agent and multi-agent setups on a financial task, reviewing credit assessment memos against requirements.
ProofPaper ↗
Key points
- Practitioner guidance holds that a large language model reviews a document better when the work is divided among agents.
- One agent checks six requirements in one call.
- Two primary predictions are tested on four models at a Bonferroni-corrected level.
- In exploratory comparisons, sampling the single agent three times and flagging on any draw gives higher balanced accuracy than the team on those models, and the team's results depend on the instruction.
Sources (1)
- [1]The More the Better? A Comparison of Single- and Multi-Agent LLM Setups on a Financial Review TaskarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 20, 01:55 AM
We test a narrow form of it in a controlled comparison of single-agent and multi-agent setups on a financial task, reviewing credit assessment memos against requirements.
Practitioner guidance holds that a large language model reviews a document better when the work is divided among agents.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 20, 2026Cache Engineering for Agents
- Oct 20, 2026Automating ML for Science: Can Frontier Agents Climb Scientific Hills in the Wild?
- Oct 19, 2026anthropics/claude-code v2.1.304
- Oct 19, 2026You Are the Trigger: Speaker-Activated Backdoors
- Oct 19, 2026Same Simulation, Different Evidence: Where Retrieved Tool Output Costs LLM Agents Accuracy
- Oct 18, 2026StopLoss: Teaching Agents to Predict Failure and Stop Early
