AI in Project Finance Modelling: ChatGPT, Claude & Excel

A plain-English explanation of how large language models actually work — tokens, context windows, and why understanding this helps you get better results from AI in Excel.

How to write prompts that get you usable financial modelling output on the first try — role, context, task, and format, illustrated with real modelling examples.

What AI hallucinations look like in a modelling context, why they happen, and the habits that catch a confidently wrong formula before it reaches your model.

How training data bias shows up in AI-generated financial content — stale market assumptions, US/UK-centric defaults — and how to adjust for your specific deal or market.

Practical rules for keeping client and deal data confidential when using AI tools, including anonymisation techniques you can apply before pasting anything into a prompt.

The risk of leaning on AI too heavily — and other common pitfalls — with guidance on where AI adds real value versus where your own judgement has to lead.

A comparison of free versus paid AI tools for financial modelling work, what each tier actually buys you, and how to decide what's worth paying for.

Using AI to generate straightforward Excel formulas from a plain description of what you need — and how to check the output before it goes into your model.

Getting AI to build complex, multi-condition nested formulas, with a framework for validating them against known values before you trust the result.

How to use AI to break down formulas you've inherited or don't fully understand — turning a black-box SUMPRODUCT or nested IF into something you can explain.

A practical process for debugging broken or suspicious formulas with AI, including how to prompt for the fix and how to verify it actually solved the problem.

An introduction to Excel's LAMBDA function — what it is, how it differs from VBA, and why it's a powerful tool for reusable financial modelling logic.

Using AI to draft custom Lambda functions for financial modelling calculations, defining them in Name Manager, and testing them before rolling them into your model.

How to pair the LET function with Lambdas and dynamic arrays to write formulas that are both efficient and genuinely readable — with AI helping you draft them.

Building a reusable library of Lambda functions for common financial modelling calculations (like DSCR), with a naming convention that keeps your model organised.

An overview of where VBA still earns its place in an AI-assisted modelling workflow — formatting, automation, and the tasks formulas alone can't handle.

Using AI to generate VBA macros from a plain-English description of what you need, and the workflow for testing and refining the code it gives you.

A hands-on walkthrough of an AI-generated VBA macro that applies consistent formatting to a financial model — inputs, formulas, and links colour-coded to standard.

Building a working scenario-switching macro with AI — from InputBox to a Form Control button on the Cover sheet — so your model can toggle between cases with one click.

Using AI to draft a macro that detects and safely breaks a circular reference in a project finance debt schedule, with a clean iterative-calculation toggle you can trust.

Using AI to turn a project brief into a clear model specification — the inputs, outputs, and structure you need before you open a blank workbook.

Setting up the Cover sheet and overall tab structure for the project finance model, establishing the navigation and layout the rest of the model will follow.

Building the Assumptions and Input sheets that drive the entire model, with AI helping structure and label the inputs clearly for anyone reviewing the file later.

Building the Scenario (Scen) tab, including the master switch cell and the INDEX/CHOOSE logic that lets one control drive every assumption in the model.

Constructing the model's timeline — development, construction, and operations periods — the backbone that every other calculation tab will reference.

Building the revenue tab for the project finance model, linking it to the timeline and assumptions so it flexes correctly across scenarios.

Modelling construction costs and their debt/equity drawdown split, including the formulas that allocate each period's spend according to the gearing ratio.

Building the operating costs and working capital tabs, covering the recurring cash flow items that sit between revenue and debt service.

Constructing the debt schedule — interest, repayments, and the classic project finance circularity — and how AI helps structure the calculation logic.

Building the depreciation and tax tabs, connecting them correctly into the financial statements that depend on them.

Setting up the DSRA (debt service reserve account) and related reserve mechanics that protect lenders and affect the cash available to equity.

Pulling the model together into a full set of financial statements — income statement, balance sheet, and cash flow — that tie out and balance correctly.

Building the ratios tab — DSCR, LLCR, and other lender and equity metrics — that summarise the model's key credit and return characteristics.

Building a dashboard and error-check framework so every tab in the model displays a clear, visible confirmation that the model is balanced and error-free.

Wrapping the model with VBA automation — formatting, checks, and the scenario button — so the finished file is both robust and easy to operate.

Why model review matters — the cost of an undetected error in a project finance model, and how AI fits into (and doesn't replace) a proper review process.

Using AI to check formula consistency across a model, catching the one cell where a pattern quietly breaks among hundreds of otherwise-identical formulas.

Using AI to review a model's logic and assumptions for internal consistency, and understanding where its review is useful versus where it falls short.

Using AI to help design and interpret stress tests, checking that a model behaves sensibly under downside scenarios.

Using AI to draft model documentation — assumption books, review memos, and change logs — as a first draft you then verify and customise.

An honest look at what AI cannot do in a model review — commercial judgement, deal-specific context, and market knowledge — so you know where to rely on it and where not to.

A recap of everything covered in the course and how the pieces — AI fundamentals, formulas, Lambdas, VBA, and a full project finance model — fit together.

A look at where AI in financial modelling is heading, and how to keep building your skills as the tools continue to evolve.

Practical next steps for applying what you've learned to your own models, plus where to go for further practice and support.

Thank you for taking this course — a short note from me as you wrap up.