Saved in:
| Main Authors: | , , , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2604.11304 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910124831932416 |
|---|---|
| author | Lau, Elaine Dücker, Markus Chaudhary, Ronak Goh, Hui Wen Wei, Rosemary Kumar, Vaibhav Qunbar, Saed Gogia, Guram Liu, Yi Millslagle, Scott Borazjanizadeh, Nasim Tkachenko, Ulyana Danquah, Samuel Eshun Schweiker, Collin Karumathil, Vijay Devalaraju, Asrith Sandadi, Varsha Nam, Haemi Arani, Punit Epps, Ray Arif, Abdullah Bhaiwala, Sahil Northcutt, Curtis Wang, Skyler Athalye, Anish Mueller, Jonas Guzmán, Francisco |
| author_facet | Lau, Elaine Dücker, Markus Chaudhary, Ronak Goh, Hui Wen Wei, Rosemary Kumar, Vaibhav Qunbar, Saed Gogia, Guram Liu, Yi Millslagle, Scott Borazjanizadeh, Nasim Tkachenko, Ulyana Danquah, Samuel Eshun Schweiker, Collin Karumathil, Vijay Devalaraju, Asrith Sandadi, Varsha Nam, Haemi Arani, Punit Epps, Ray Arif, Abdullah Bhaiwala, Sahil Northcutt, Curtis Wang, Skyler Athalye, Anish Mueller, Jonas Guzmán, Francisco |
| contents | Existing AI benchmarks lack the fidelity to assess economically meaningful progress on professional workflows. To evaluate frontier AI agents in a high-value, labor-intensive profession, we introduce BankerToolBench (BTB): an open-source benchmark of end-to-end analytical workflows routinely performed by junior investment bankers. To develop an ecologically valid benchmark grounded in representative work environments, we collaborated with 502 investment bankers from leading firms. BTB requires agents to execute senior banker requests by navigating data rooms, using industry tools (market data platform, SEC filings database), and generating multi-file deliverables--including Excel financial models, PowerPoint pitch decks, and PDF/Word reports. Completing a BTB task takes bankers up to 21 hours, underscoring the economic stakes of successfully delegating this work to AI. BTB enables automated evaluation of any LLM or agent, scoring deliverables against 100+ rubric criteria defined by veteran investment bankers to capture stakeholder utility. Testing 9 frontier models, we find that even the best-performing model (GPT-5.4) fails nearly half of the rubric criteria and bankers rate 0% of its outputs as client-ready. Our failure analysis reveals key obstacles (such as breakdowns in cross-artifact consistency) and improvement directions for agentic AI in high-stakes professional workflows. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_11304 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | BankerToolBench: Evaluating AI Agents in End-to-End Investment Banking Workflows Lau, Elaine Dücker, Markus Chaudhary, Ronak Goh, Hui Wen Wei, Rosemary Kumar, Vaibhav Qunbar, Saed Gogia, Guram Liu, Yi Millslagle, Scott Borazjanizadeh, Nasim Tkachenko, Ulyana Danquah, Samuel Eshun Schweiker, Collin Karumathil, Vijay Devalaraju, Asrith Sandadi, Varsha Nam, Haemi Arani, Punit Epps, Ray Arif, Abdullah Bhaiwala, Sahil Northcutt, Curtis Wang, Skyler Athalye, Anish Mueller, Jonas Guzmán, Francisco Artificial Intelligence Existing AI benchmarks lack the fidelity to assess economically meaningful progress on professional workflows. To evaluate frontier AI agents in a high-value, labor-intensive profession, we introduce BankerToolBench (BTB): an open-source benchmark of end-to-end analytical workflows routinely performed by junior investment bankers. To develop an ecologically valid benchmark grounded in representative work environments, we collaborated with 502 investment bankers from leading firms. BTB requires agents to execute senior banker requests by navigating data rooms, using industry tools (market data platform, SEC filings database), and generating multi-file deliverables--including Excel financial models, PowerPoint pitch decks, and PDF/Word reports. Completing a BTB task takes bankers up to 21 hours, underscoring the economic stakes of successfully delegating this work to AI. BTB enables automated evaluation of any LLM or agent, scoring deliverables against 100+ rubric criteria defined by veteran investment bankers to capture stakeholder utility. Testing 9 frontier models, we find that even the best-performing model (GPT-5.4) fails nearly half of the rubric criteria and bankers rate 0% of its outputs as client-ready. Our failure analysis reveals key obstacles (such as breakdowns in cross-artifact consistency) and improvement directions for agentic AI in high-stakes professional workflows. |
| title | BankerToolBench: Evaluating AI Agents in End-to-End Investment Banking Workflows |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2604.11304 |