Why Operating Models Matter More Than Ever in the Age of AI

AI is pushing a lot of companies to step back and look at how work really gets done inside the organisation. Not the planned version or the theoretical version, but the everyday reality. And when you look closely, you see that many operating models were never designed for the level of scale, complexity or customer expectation companies are dealing with now. AI makes that gap very visible, very quickly.

From my desk in Castelldefels near Barcelona (and a recent sunset below) I’ve been looking across different organisations and noticing the same thing: this isn’t something that applies only to fast-scaling startups. It’s present in banks, payments companies, fintechs and any organisation that has grown steadily over time. As volume increases, internal complexity increases with it. The operating model doesn’t automatically evolve to match that pace, which creates friction long before external factors do.

You start to see the same patterns repeat: workflows that drift from their original design, ownership that isn’t completely clear, and teams interpreting the same process in different ways. Exceptions rise, manual workarounds become the norm, and technology decisions made years ago turn into operational friction today. CS, Ops, Risk and Product often have very different views of what “the process” actually is. These gaps don’t usually show up in strategy decks, but they show up in day-to-day performance.

There’s plenty of data pointing in the same direction: internal execution is a far bigger determinant of success or failure than people often admit. Yes, there are high-profile cases where poor strategic decisions caused companies to fall behind, that absolutely happens. But when you zoom out and look across organisations in payments, fintech and BFSI, strategy usually isn’t the thing that breaks. The bigger issues tend to be internal i.e., how the organisation actually works day to day, and whether the operating model can keep up with what the business is trying to do.

The real problems show up inside the company: unclear roles, inconsistent processes, misaligned teams and operating models that haven’t kept pace. Many organisations looking at AI say they aren’t structurally ready for it. And companies growing quickly often find that internal complexity, not external forces, is what limits their ability to scale.

AI exposes the gaps because it needs clarity. It needs predictable workflows, clean handoffs and decision logic that is consistent. When those things aren’t in place, AI simply can’t operate. This is why early AI efforts often reveal issues that have existed for years. The technology highlights the inconsistency, it doesn’t create it.

A lot of operating models were built for a different era, a time when human judgment absorbed complexity, when demand was more predictable and when exceptions were manageable. That’s not the case today. Digital volume is constant, regulation is tighter and customer expectations move faster. AI adds another layer that depends on structure, not individual workarounds.

If companies want to make meaningful progress with AI, or simply want to scale without constant friction, they need to revisit the basics: how work actually flows from end to end, how decisions are made, how teams coordinate, and what AI should take on versus what humans should retain. They also need a clearer operating rhythm so complexity doesn’t build up in the background.

You see this clearly when organisations start mapping their exception paths. The “happy path” is usually the minority of work. The majority sits in the grey areas that no one formally owns. Once that becomes visible, it’s obvious why teams struggle with volume, consistency and speed.

Operating models that were “good enough” in 2015 aren’t fit for the pace of 2025.