AI Financial Modelling in Excel: What Actually Works Today

Financial modelling has been one of the last bastions of purely manual work in advisory. Analysts spend hours — sometimes days — building DCF models, comparable company analyses, and LBO frameworks from templates that haven't meaningfully changed in a decade. The templates work, but the process of populating them, linking the formulas correctly, and formatting the output for client presentation is time-consuming and error-prone.

AI-powered model generation is now changing that equation. Purpose-built tools can take a set of deal parameters — sector, revenue, margins, growth assumptions, deal structure — and produce a fully formatted, formula-linked Excel model in minutes. The output isn't a simplified summary or a PDF report. It's a native .xlsx file with proper cell references, assumption toggles, and sensitivity tables.

What AI-generated models actually look like

A well-built AI model generator produces output that's structurally identical to what an experienced analyst would build by hand. The assumptions page captures the key inputs. The income statement, balance sheet, and cash flow statement flow through with proper linking. The DCF valuation uses a standard WACC-based discount methodology with terminal value calculations. Sensitivity tables show the impact of changes to key assumptions on implied valuation.

The critical difference is time. A model that takes an analyst four to six hours to build from a template can be generated in under ten minutes. The analyst's role shifts from construction to review — checking assumptions, adjusting projections, and stress-testing scenarios rather than building formulas from scratch.

Where human review still matters

AI models are not a replacement for financial judgement. The generated output is only as good as the inputs, and deal-specific nuances — non-recurring items, unusual capital structures, sector-specific metrics — still require experienced review. The value of AI modelling is in eliminating the mechanical work, not the analytical work.

Three areas that always need human attention

First, assumption validation. AI will use the inputs you provide, but it won't challenge whether those inputs are reasonable in context. An experienced analyst will. Second, deal-specific adjustments. Every transaction has idiosyncrasies that a template — even an AI-generated one — can't anticipate. Third, presentation formatting for specific client preferences. Some clients want detailed footnotes; others want a clean one-page summary. That last mile of customisation remains human work.

The practical impact for boutique firms

For a boutique advisory team running multiple mandates simultaneously, AI model generation changes the capacity equation. Instead of one analyst spending a day building a model for a new pitch, the model is generated in minutes and refined in an hour. That analyst can now support two or three concurrent mandates where previously they could handle one.

The firms adopting this approach aren't replacing their analysts. They're making each analyst significantly more productive — which, for a team of five people, is the difference between handling four mandates and handling seven.