Building an effective AI strategy

The AI rush is turning to disillusionment, as businesses risk digging up fool’s gold, warns Tom Lorimer at Passion Labs

 

In the recent rush to jump on the AI bandwagon, many businesses have lost sight of the problems they originally set out to solve. As a result, many are now weighed down by overlapping tools, half-formed strategies and unclear returns.

 

It’s an assessment that is echoed in Gartner’s 2025 Hype Cycle for Artificial Intelligence report, which observed that generative AI is passing through its gold rush phase and entering the “trough of disillusionment” as organisations seek to gain a clearer view of both its potential and its limitations.

 

But what does this actually mean, other than that the early excitement of the AI boom is now giving way to a period of reflection and refinement?

 

In a recent AI audit we conducted for a well-known publishing house, we uncovered a striking example of just how rapid AI adoption can outpace strategy. During one presentation, the publisher shared what they described as their “nightmare slide,” which contained the logos of a mass of AI tools they had formerly, and employees had informally, adopted in just 18 months.

 

When we dug a little further, we found that nearly half of their tools overlapped in terms of functionality; many were poorly integrated, and some were built on what can only be described as questionable open-source foundations with unclear data protocols. Despite the sheer number of tools at their disposal, few, if any, were delivering measurable impact. It wasn’t so much a strategy as a scramble.

 

 

The reality of the Gartner hype cycle

For me, this is the reality of Gartner’s ‘trough of disillusionment.’ But to understand where we are today, you have to go back a few steps and look at the activity of the last couple of years. They’ve been shaped by a kind of frenzied procurement cycle, with new tools adopted at a rate of knots, not necessarily because they met a specific need, but because they looked like they might.

 

As a result, overlapping systems, unclear ownership and fragmented workflows are now routine. For those that took this path, the result is no single source of truth, no end-to-end visibility and, in many cases, no real idea which tools are still delivering value. Which is why it’s time to change the way we think about AI.

 

 

De-implementation is now a critical innovation strategy

In short, businesses need to declutter and start to think strategically. This means auditing what’s already in use, cutting what doesn’t serve a clear purpose and rebuilding tech stacks around clear business goals.

 

For me, that process starts with a simple set of questions we refer to as NAFDA:

• What is the NEED, and what are you looking to achieve?
• Is there ALIGNMENT internally on the core workflows slowing you down?
• Have you reviewed the business case FINANCIALLY?
• How is the DATA currently stored and structured?
• How do you intend to ANALYSE the risks and impact?

Until those questions are answered, chucking more money or tools at the problem won’t resolve anything. In fact, I would go further and say that of all the issues faced by businesses, money isn’t one of them. Billions have already been poured into AI, often with mixed results.

 

According to Gartner, spending on Gen AI is expected to top $644 billion in 2025, an increase of 76.4% from 2024. In other words, there’s no shortage of cash to pump into these projects. Instead, we have an abundance of investment directed at tools without the necessary groundwork to support them. There’s no clear implementation plan. No change management. No training.

 

This isn’t just a waste of money. It also creates confusion and resistance. Teams don’t know what the tools are for. Leaders struggle to measure impact. And the people closest to the work often aren’t consulted at all.

 

 

Building an effective strategy around AI

So, what’s the best approach for building an effective strategy around AI? As I outlined earlier, the first thing is to be honest. It is only by conducting a thorough audit, and asking those all-important questions, that you will have true visibility of what you have and what you need.

 

Change management is also essential. From our conversations with businesses, it’s clear that trust, understanding and willingness to use AI vary hugely across organisations. So in the interest of best practice, it’s important to remember that any successful implementation depends on upfront support, guidance and training. In that regard, AI strategy isn’t separate from business strategy. They need to be one and the same.

 

Finally, you need to measure the impact on your investment. If the value is anecdotal, vague or hard to track, something’s gone wrong. If there are tools in your AI tech stack collecting dust or internal teams aren’t clear on their quantifiable impact, that’s a red flag. 

 

 

Measuring your AI investment

It’s a hard lesson to learn, but you’ll be amazed at how many organisations we see that have fallen into this trap. As an AI research and development lab, we do as we preach, and so we often run internal audits to ensure our tech stack is not bloated. After all, subscriptions add up fast, and if you’re not maximising that investment, you’re simply eroding the bottom line.

 

That means quarterly reviews between your management, finance and tech teams will provide those all-important checks and balances. Someone needs to be looking at usage and impact and holding leadership to account; otherwise, it just spirals. If you can’t track the ROI and the stack has grown without structure, then you’re dragging yourselves down.

 

This whole thing – the frenzy, the hype, the mad dash to spend – it’s not new. It happens every time the barriers to entry drop. It happened with the cloud. It happened with mobile. But we’re seeing it at an unprecedented scale now.

 

That said, the solution remains the same: get back to business fundamentals. Strategy. ROI. Customer value. Ask the right questions – the same questions you’d ask in any other part of the business – and apply them here.

 

Because the next phase of AI maturity won’t be defined by who has the most tools but by who has the most clarity. The real innovation today isn’t what you have; it’s knowing what to leave out.

 

---

 

Tom Lorimer is CEO at Passion Labs

 

Main image courtesy of iStockPhoto.com and Thomas Faull