By David, Chief Strategy Officer at &above, an AI product studio
Artificial intelligence (AI) holds masses of potential for business optimisation. It can support efficiency and productivity across the board. But it needs to be implemented properly. Too many companies find themselves struggling to gain worthwhile returns on their AI investment, but it’s not the tech at fault. More often than not, poor implementation and infrastructure are the reasons why AI fails.
Why AI isn’t a standalone solution
Many businesses feel shortchanged by their AI “transformations” because their expectations are too high. It’s not that AI isn’t up for the job, but rather that it isn’t a plug-and-play solution. AI doesn’t – can’t – work in isolation. Without clean data and a viable infrastructure, even the most advanced AI tech will deliver poor results. And for the businesses currently struggling to keep up, that means risking the loss of both operational efficiency and strategic relevance, as their poorly integrated technology fails to live up to its potential.
It’s a problem that is being amplified by the broader public narrative. While AI-driven job losses have dominated the AI conversation, the real disruption is actually coming from the operational and cultural changes required to make AI work. Businesses are caught, knowing that transformation is necessary, but struggling to implement it without criticism. All the while, technology continues to change and advance, making it even harder for companies to keep up.
A growing need for cohesion and willingness to change
Alongside predictive and generative AI, we’re now seeing the development of agentic AI. These are systems that are able to make and act on decisions autonomously. It’s one of the most fundamental changes since AI was developed, holding the key to genuine autonomy, allowing businesses to move on from the “grunt work” and focus on the things that can bring in revenue. And to support that even further, it can deliver the insights businesses need to effectively strategise, improve forecasting, and personalise customer experiences. But that value can only be unlocked when AI is implemented properly. If it’s used on an ad hoc basis, it’s easy for a fragmented ecosystem of disconnected tools and isolated initiatives to evolve, preventing the tech from living up to expectations.
It’s a problem also being felt by those businesses unwilling to give up on their legacy infrastructure. People become attached to the systems they know, so they attempt to modernise through piecemeal add-ons. But these are systems that can’t communicate with the new tech, leading to data silos and communication errors. What many businesses are failing to understand is that they don’t need another layer of technology, so much as a complete behavioural and cultural shift.
It’s not just the technology that matters
Successful AI adoption is as much about culture as it is about technology. For AI to deliver returns that justify the investment, organisations must also invest in the behavioural frameworks that help teams to work with it.
Effective and consistent measurement is another must for AI implementation. Because if you’re not tracking the relevant metrics – productivity, efficiency, revenue growth, customer satisfaction – you can’t see what’s working and what needs to be changed.
Scalability and replicability matter too. The more you can roll out successful AI use cases across teams and functions, the greater the impact and returns will be. But this doesn’t happen overnight. The AI value curve develops gradually, and many organisations overlook this, mistaking delayed returns for failure. That misconception, coupled with the issues surrounding data, is the cornerstone of why so many companies feel their AI investments have fallen short.
Data is the area where the majority of businesses struggle with AI. There’s a widespread failure to realise that AI can only perform at its best when data is clean, accessible, and fully integrated. When data is fragmented or siloed, it’s impossible for AI to do what it’s designed for. So, if you don’t restructure your data, or at least implement systems that clean and connect it as you go, AI won’t be able to deliver the results you expect.
Equally important is tool selection. When integrating AI, you need to choose tools that solve your organisation’s core challenges. Experimentation can be useful, but it must be disciplined and carefully piloted before gradual scaling. And you need to focus on structured, company-wide education to support that, ensuring that your teams not only understand how to use AI effectively, but that it’s a tool rather than a threat.
AI can’t solve all of your business’ problems. That’s not what it’s designed for. But with preparation, planning, performance measurement, and the effort to ensure strong cultural integration, it can bring enormous value. So, stop thinking of AI transformation as an easy fix to organisational glitches or a way to reduce your headcount. AI is at its very best when it’s informed, supported, and being used to support the capabilities of your people.
