Before AI builds your models
Introduction: spreadsheets are here to stay
Introducing the blog series "Before AI builds your models"
Before AI builds your models
Introduction: spreadsheets are here to stay
Global gross capital formation is not an abstract statistic. It determines how our planet's finite resources - labour, materials, energy, land, and time - get deployed. In 2024, that number was $29tn (World Bank, 2024).
Since the 1980s, spreadsheet models have been our instrument of choice for justifying these decisions. Why has that endured? What happens when AI sends the production of high-stakes, weakly governed spreadsheets into overdrive?
Spreadsheets continue to be a terrible artifact to underpin these decisions: no version control, no audit trail, no standards enforcement, logic scattered everywhere, unusable collaboration tools, and increasingly complex requirements.
Today, huge monolith beasts with millions of formulas and hundreds of revisions are our day-to-day. 90%+ of the time, these models contain errors (Poon et al., 2024). In project finance, this is so challenging that $200bn+ of transactions were supported by specialist spreadsheet auditors in 2025 alone (Infralogic, 2026).
Over the decades, countless projects have tried and failed to move this critical business logic away from spreadsheets, into more controllable environments - a task akin to convincing the entire British population to learn Turkish because it follows a better-defined set of rules. Spreadsheets are the undisputed universal language of finance; they communicate business logic across teams, transactions, enterprise, countries and governments. They are here to stay.
In 2026, AI threatens to send this already fragile ecosystem into overdrive, promising massively increased productivity via autonomous, intelligent agents. The same risks and systemic flaws that we have learned to live with are about to be tested on a new scale.
In the following posts, I will explore techniques and strategies to keep control of what's to come.
This series is not about changing our spreadsheets. It's about changing how we build them.
Before AI builds your models - navigate series
Spreadsheets are not ready for AI [unpublished]
AI makes inconsistency easy [unpublished]
Spreadsheet modelling doesn't happen in spreadsheets [unpublished]
Model review was impossible. AI made it worse. [unpublished]
Can you prove an AI-built model is correct? [unpublished]
Excel is the perfect place to bury a mistake [unpublished]
Your model is doing too much [unpublished]
Escaping the spreadsheet [unpublished]
