Most businesses approach AI adoption believing they're making the right move. Leadership has watched the technology mature and heard about competitor progress, prompting the decision that the business needs to act. Once budget is approved and a suitable vendor identified, the project begins with objectives that seem clear and timelines that feel achievable.

‍

What rarely happens first is the question that determines whether any of it will work: what outcome are we trying to achieve and do we have the foundations to support it?

‍

That question sounds straightforward. In practice, most organisations skip it entirely. The consequences surface at exactly the moment the business is least prepared for them.

‍

Efficiency Isn't Strategy

‍

The most common starting point for an AI project is a cost or efficiency target. The board wants to reduce headcount in a particular function or automate a process that was previously manual.

‍

There's nothing wrong with efficiency as an outcome. The problem is that efficiency projects tend to automate existing processes rather than redesign them around the outcomes those processes serve. The organisation ends up automating something that probably should have been rebuilt, and the AI layer inherits all the structural limitations of what it was built on top of.

‍

As Kirsty Rutter, Head of Venture Capital at Lloyds Banking Group, puts it:

‍

"If you've got an organisational model that is still designed for a linear process, what you're going to get by putting AI on top of it is some efficiency. The design premise has to be what outcome do you want, not what process do you currently run."

‍

‍[Episode 49 – Leaders and Founders Podcast]

‍

This is the distinction that separates AI projects that compound in value over time from those that deliver a one-time efficiency gain and then plateau. Outcome-led design requires a different conversation at the start, one that most project briefs never get to.

‍

The Data Problem

‍

The second reason AI projects fail is data. Specifically, the gap between the data a business believes it has and the data it has in a state that's useful.

‍

This gap is more common than most leadership teams realise. The discovery usually happens after the project has been scoped and the vendor contracted, when the first development sprint reveals that the underlying data isn't in the state everyone assumed it was.

‍

Lee Provoost, CTO of Flagstone, made a deliberate decision to hold back from aggressive AI adoption in 2025. Instead, he focused on the organisational foundations that would make adoption genuinely useful when it happened. His reasoning was direct: businesses that haven't addressed their structural and data foundations won't get the return from AI they're expecting. The cost of discovering that mid-project is significantly higher than the cost of addressing it first.

‍

"There are so many basics that companies tend to forget," Provoost explains. "If you actually get your house in order, a lot of companies could easily get a 20 to 30 percent productivity boost without even touching AI or GitHub Copilot or any of those products."

‍

‍[Episode 50 – Leaders and Founders Podcast]

‍

Most AI projects assume the data infrastructure is ready when it was never designed for what AI requires.

‍

The Governance Gap

‍

An overlooked problem at the project scoping stage is governance, which becomes the primary focus of regulators and auditors after deployment.

‍

When AI systems are deployed without governance built into their design from the start, they create a category of risk that becomes very difficult and expensive to fix later.

‍

Kirsty Rutter, whose career has spanned innovation, risk, and regulation across some of the UK's largest financial institutions, is direct about where she sees the market failing:

‍

"The real winners are going to be either the tools that are written with risk and control built in, or services that serve that need. If you haven't got it, it is going to cause you so much trouble. That is still what I believe is where, in a couple of years’ time, it will be what stops all the really bad stuff happening."

‍

‍[Episode 49 – Leaders and Founders Podcast]

‍

For PE-backed businesses in regulated industries, this isn't theoretical. It's a live risk with direct implications for how regulators view the business, which in turn affects valuation and the path to scaling.

‍

You can't retrofit governance into an AI system that was built without it. The architecture must support it from the start.

‍

What Works

‍

The businesses that get compounding value from AI investment share common characteristics. They start with a clearly defined outcome rather than a technology. They audit their data before they scope the solution and they build governance into the design rather than appending it afterwards.

‍

Most importantly, they have technical leadership in place that can hold the business and the technology in mind simultaneously, asking the right questions when they're still easy to act on.

‍

The questions that determine whether an AI project will deliver value are almost always easier and less expensive to answer before the brief is fixed than at any point afterwards.

‍

How We Help

‍

At Gathered & Found, we help organisations understand whether their technical foundations can support AI adoption, looking at everything from how their data is structured to whether their architecture has the flexibility AI workloads demand.

‍

If you're planning AI adoption, start by understanding whether your platform is ready.

‍