Every finance team is being asked the same question: Should we be using AI to build our models? The Ankura OCFO® team ran identical prompts through Claude for Excel, M365 Copilot (Agent Mode), ChatGPT in Excel, Shortcut (Free Tier – Open SourceModel), and Shortcut (Pro Tier) against a fictional company (Fieldstone Retail Holdings, approximately $180 million revenue, three entities).
The three test scenarios — a Long-Range Plan, 13-Week Cash Flow Forecast, and Customer Profitability Analysis — were selected because they mirror common financial modeling exercises in PE-backed finance and operations environments. AI was used to generate the fictional company data and over 500 word prompts for each scenario.
The goal was to establish a baseline comparison of AI financial modeling tools available on the market. What did we find? All five tools produced structurally complete models in 7-10 minutes, but no tool produced a production-ready model. Every tool had fundamental formula mechanics wrong or an incomplete, inconsistent structure. The value today is in scaffolding and structure; the formulas still require meaningful human rework. Critically, each prompt was delivered as a single pass with no iteration or follow-up, and identical prompts were copied across all five tools to isolate baseline capability. If a tool asked a follow-up question, the default was to accept its recommendation. With iterative prompting utilizing currently available features, results would likely improve.
Read more in our detailed report.
© Copyright 2026. The views expressed herein are those of the author(s) and not necessarily the views of Ankura Consulting Group, LLC, its management, its subsidiaries, its affiliates, or its other professionals. Ankura is not a law firm and cannot provide legal advice.
