AI: It's Not About Doing More. It's About Doing What You Couldn't.

The Wrong Conversation

When we talk about AI in business, the conversation almost always gravitates towards productivity. Faster reports. Automated summaries. Meeting notes you don't have to write yourself. And yes, that's real. That has value.

But it's the wrong conversation.

The more interesting question isn't "how do we use AI to do more of what we're already doing?" It's "what becomes possible that wasn't possible before?"

Those are very different questions, and they lead to very different outcomes.

The Weight of the Past

The company I mentioned at the start has a piece of critical infrastructure at its core. Built over a decade ago by smart people with good intentions and the best thinking available at the time. They built it with the future in mind, which was admirable. But the future they imagined wasn't the future they got. The business grew, the product line expanded, and the foundational assumptions that underpinned the system quietly stopped being true.

Now, changing that system is a monster of an undertaking. We're talking multi million pound investments. Multi year programmes. New governance structures, migration strategies, risk frameworks. A cast of hundreds.

The result? The business has learned to work around it. They build products that fit within the constraint rather than products their customers actually need. Every year, a quiet toll is paid: opportunities not taken, customers not served, products not built.

That shrug from the CTO isn't resignation. It's a rational response to an impossible trade off.

Where AI Changes Something Fundamental

A problem that previously required a large team, enormous budget and years of runway can now be approached differently. A small group of senior domain experts (enterprise architects, solutions architects, people who actually understand the system and the business) working alongside a coordinated set of AI agents can do work that previously required an army.

The work doesn't disappear. It's still complex. But the shape of it changes.

Difficult problems get solved.

And when that happens, something else shifts too. The company stops being constrained by its past. It can build the products and services its customers actually need. It can engage in ways that were previously closed off to it.

The Compounding Returns

Multiply that single example across any company carrying similar weight. Legacy systems that decisions were made around rather than through. Technical debt accumulated not from carelessness but from honest trade offs made under pressure, years ago. Regulatory workarounds that became the default operating mode.

Every one of these represents capacity that is tied up in maintenance rather than growth. In cost to serve rather than revenue. In managing the consequences of the past rather than building towards the future.

And that reclassification has compounding returns. Fix one constraint and you unlock the ability to pursue opportunities that were invisible before. Fix several and you fundamentally change what kind of company you are.

Getting There From Here

That's what I'm most interested in exploring with companies right now. Not the productivity wins, though those are real and worth having. But the deeper question of what's currently impossible that doesn't have to stay that way.

The path to getting there isn't a big AI transformation programme. It's smaller than that, and it has to start somewhere sensible. You need to build the capability to work this way before you can take on the big problems. You need to understand how to direct AI agents, how to pair domain expertise with AI capability effectively, how to validate what comes out the other side.

The CTO and I still speak regularly. The conversation has shifted. He's starting to think about what he can get done.