Testing Claude in Excel: Building a Three-Statement Financial Model

Artificial intelligence is beginning to change how financial models are built, tested, and maintained in Excel. Join Tim Vipons, CEO of CFI, as he walks through and explores whether an AI tool can replicate a fully integrated three-statement financial model using the same structure, logic, and standards taught in CFI courses. The test is straightforward but demanding. A completed three-statement model has its entire forecast section removed, including all formulas. Only historical data and assumptions remain. That blank forecast is then uploaded into Excel using Claude by Anthropic to see whether it can rebuild a dynamic, fully linked income statement, balance sheet, and cash flow statement from scratch. The demonstration follows the full modeling process step by step, starting with importing historical financials, identifying forecast periods, and building formulas across all three statements. Supporting schedules for working capital, PPE, and capital structure are constructed along the way, just as a financial analyst would approach the task. As formulas are added, the model is reviewed in real time for structure, logic, and accuracy. The balance sheet is stress-tested, assumptions are traced through to outputs, and results are compared side by side with the original CFI course model. The final output is evaluated on correctness, linkage, and adherence to CFI modeling conventions. For analysts and finance teams, this raises important questions about how AI may support model building, quality control, and productivity in Excel. Understanding what these tools can and cannot do today is increasingly relevant for anyone working with financial models. Relevant for: Finance professionals, analysts, and students interested in financial modeling, Excel-based forecasting, and the practical impact of AI on finance workflows. Subscribe for more practical insights on financial modeling, Excel, and the future of finance skills.