By Gert-Jan Wijman, VP & GM EMEA, Celigo

In the world of IT, there is rarely a period when some technology trend isn’t promising to deliver greater efficiency, productivity, and competitive advantage.

Few trends, however, have ever been met with the level of attention, expectation, and investment that AI is currently receiving. Usually, we would expect to see diversity in how businesses react to new technologies as they learn and experiment, but in a recent survey of more than 1,200 global enterprise Operations and IT leaders, Celigo found that 97% of respondents already view AI as ‘critical to driving operational improvements in the coming year’. That’s amazing when you consider that less than 10 years ago, there weren’t machines considered reliable enough to provide language or image recognition at a human level.

Of those 97%, the vast majority are already well into the swing of actively investing in AI: over three-quarters of businesses indicate that they have dedicated specific resources and budget to AI, while over four-fifths have a formal strategy or roadmap in place for AI implementation. However, usage does not automatically turn into benefits, and the sheer level of interest and effort in AI adoption only raises the stakes for businesses that need to show real ROI from their exploration of this new technology.

The data, and our experience based on working with IT customers, suggest that there are a few key questions which can point the way towards successful strategies that overcome roadblocks on the path to AI adoption.

Who leads the AI charge?

Whether the technology in question is a tailor-made solution or a plug-and-play tool,  the process is usually driven by IT teams. However, there are signs that for AI that isn’t the whole story. Just 26% of businesses, in fact, say that IT is at the forefront of their AI mandate, and over half allow users to implement AI solutions without formal IT oversight.

There are multiple reasons for this. For one, IT teams are often overburdened as it is, leaving them with little breathing room to take charge of something as all-encompassing as AI adoption. But at the same time, part of the promise of AI is the way that it can democratise access to technology, making complex processes more intuitive.

Indeed, 68% of businesses say they approve of a Citizen Developer mindset, in which knowledge workers are empowered to innovate processes in ways that were typically reserved for technology specialists. Such an approach has obvious benefits in terms of sharing the workload, and has the advantage that departments and teams are the experts in what capabilities would best augment their own workflows.

While there are clearly advantages to allowing citizen developers to play a role in implementing AI, it also exacerbates risks, particularly on grounds of security and data governance.To empower Citizen Developers safely, businesses first need a modern approach to integration.

Where does AI happen?

All AI applications start with good data. While any given department will have its key platforms for gathering and managing data – customer relationship management platforms, enterprise resource planning platforms, collaboration and productivity platforms, and so on – the best results will come when those data sources are brought together in a holistic way that can generate deeper insights.

The challenge of integration has been growing for a long time, as businesses lean on ever more cloud services to carry out day-to-day business. Having many specialised tools available can help teams to excel in their work, but it also makes connecting the business’s IT infrastructure together in a unified way exponentially more complex.

The arrival of AI is adding real urgency to this challenge: while employees may be able to find ways of navigating across many data sources, AI needs data to be available in a more frictionless way. Our survey found that businesses are expecting to exploit a huge diversity of data sources and types through their AI adoption, from cloud platforms and APIs to user interaction tracking and user feedback data.

In this context, investing solely in the end-goal of AI implementation risks either outcomes that underperform due to a lack of data or outcomes that create governance issues through inexpert data integrations. Attention should also be paid to technologies like Integration Platforms-as-a-Service (iPaaS), which can significantly simplify and normalise the underlying data integration challenge. Organisations should also place attention on the upskilling of staff through training so as to maximise the benefit of AI to the business.

How are AI benefits shared?

While security was the most common risk identified by respondents to our survey, 46% said that fears around jobs being replaced by AI are a concern in their organisations. As the Citizen Developer mindset suggests, however, AI is no different to any other technology in that it is ultimately by and for people.

Just as the adoption of specialised platforms by different teams can create data silos and integration challenges, permitting unchecked team-level innovation without IT oversight can ironically reinforce the very barriers that data integration aims to dismantle. This paradox highlights the delicate balance between fostering innovation and maintaining a cohesive, interconnected IT ecosystem. While team autonomy can drive rapid advancements and tailored solutions, it may inadvertently perpetuate isolation and fragmentation across the organisation’s data landscape. The challenge lies in cultivating an environment that encourages innovation while simultaneously ensuring new technologies and processes align with broader organisational goals for data accessibility and integration.

In order to maintain security while promoting the freedom to self-implement, it’s imperative that companies have a clear strategy on balancing the two. Establishing a clearly documented AI policy, for instance, can alleviate uncertainty over what is and isn’t allowed as people explore the technology. Creating an open culture of learning and experimentation can be helped with social feedback loops like lunch-and-learns, where non-technical employees share what has worked for them and IT leaders can offer their expert advice.

Over time, almost every business will experience AI as a critical driver of operational improvement. When so many businesses are investing so heavily, though, the real winners will be those who take the smartest path to the destination.