AI has become an important part of board-level discussions around growth, competitiveness, and resilience. However, there are serious challenges at the heart of many UK companies’ approach to technology.
Although they are plowing small fortunes into AI (nearly £16 million on average this year alone, according to new research from SAP and Oxford Economics), only 7% have a strategic, enterprise-wide investment plan for AI. In other words, much of today’s spending is piecemeal, ad hoc, and fundamentally short-termist.
In practice, this means that while piloting and implementing AI has shown some benefits, the full impact on productivity and growth is yet to be realized as organizations struggle to scale the results across their businesses. This is reflected in the fact that 70% of UK businesses are currently unsure whether AI is realizing its full potential. “Currently, most AI projects are technology-driven and focused on one business process or department, so they are not necessarily aligned with the strategic ambitions of the company,” said Sonia Nash, Head of Business AI at SAP UK & Ireland.
To meet this challenge, companies need strategies that integrate data, people, and governance. This allows you to bring experimentation and distributed tools out of the shadows, connect and scale them to deliver truly transformative results, from improving customer experiences and rapidly innovating to creating unique products and services.
The first step is to establish and embed cross-functional AI governance, which requires secure executive-level sponsorship. Once you have this in place, “the key is to define what your core business objectives are and whether and where AI can support them,” explains Nash.
In other words, before teams and departments jump ahead with siled proof-of-concept (PoC) experiments and support investments in new tools and technologies, organizations need to understand the “why” of AI—what business problems it can help solve as part of an overall strategy. “AI is a big umbrella term, and we often hear about the next bright new thing: agent AI,” Nash says. “But AI agents won’t necessarily have the right answer for every problem. Sometimes machine learning is the right answer.”
To scale your success stories and avoid wasted investments, you also need to establish effective KPIs that focus on employee recruitment, cost reduction, revenue growth, and other important metrics related to key business objectives. “We need to monitor not only the technical performance of AI, but also whether it is having real business impact,” says Nash. Additionally, “When a PoC lacks predefined KPIs, many teams have a hard time articulating a specific business case to management, making it difficult for organizations to commit to enterprise-wide investments.”
empower people
Employees are often enthusiastic about using AI, but lack the training, confidence, and support to use it safely and responsibly. This has led to issues with unauthorized “shadow AI” within many organizations. In fact, a study by SAP and Oxford Economics found that 68% of organizations say their staff uses unapproved AI tools at least sometimes, and 44% have already experienced a data breach due to shadow AI.
“If someone is experimenting with tools to create an AI agent and incorporate it into their data without thorough compliance checks, they can put their company at great risk,” says Nash. “But many employees don’t realize this because they haven’t received the proper training.”
More than half (60%) of organizations admit their staff haven’t completed comprehensive AI training, which Nash said is “essential” to experimenting with AI safely and productively. “Mandatory training addresses both sides of the coin,” she explains. “First of all, we need to make sure that people understand the risks involved in using unapproved tools, and at the same time introduce them to the approved tools at their disposal.”
In addition to strong guardrails and guidance on when, where, and how to use AI, the cultural change needed to bring AI out of the shadows also depends on transparent communication about its benefits, both in achieving individual and organizational long-term goals.
“You can’t just suddenly tell people, ‘Hey, you’re using this new tool now,'” Nash says. “You need to take them on a journey and explain why the company decided to bring AI into the organization, how it will re-engineer some business processes and what benefits it will bring.”
Employees also need to feel comfortable asking questions, reporting problems, and even making mistakes. Concerns about job security should also be addressed openly and fully. “There is still a lot of anxiety surrounding AI, and many employees worry that it will make some of their jobs obsolete,” Nash says. Therefore, leaders need to communicate that their company’s AI strategy is driven by a desire to “enhance and help humans, not replace them.”
The message that AI can help needs to come from both ends of the spectrum
Grassroots AI communities help foster enthusiasm for new tools by allowing employees to experiment together, share knowledge, and provide feedback throughout the organization. This is a model that recognizes that people learn best from their peers, not just from top-down orders.
“To ensure that AI can be used safely and compliantly, we need to send the message from both ends of the spectrum that AI can help, and we need to ensure that it is adopted and used in ways that deliver value,” Nash says. “Because even if you build a case for a great tool internally, if your employees don’t leverage it, the project is a failure.”
connected data
Without accurate, connected data, it is nearly impossible to scale AI and drive innovative results. Today, companies across all sectors are looking to build the strong data foundations they need for their AI efforts. However, breaking down silos to allow information to flow freely within an organization remains a challenge for many.
AI pilots often succeed in controlled environments because they use carefully selected datasets. However, it is difficult to scale successfully when there are problems at the organizational level. “Poor data quality and accessibility are one of the major barriers to bringing PoC to production,” says Nash.
However, resolving these issues may not be as difficult as some organizations think. “There are tools on the market today that can help you analyze the state of your data, identify what’s missing, fix issues and build a new foundation,” Nash explains. “So it’s not as difficult as it used to be.”
Moving unconnected legacy systems, processes, and data to the cloud can also help solve data challenges and enable end-to-end use of AI. “AI is now at the forefront of people’s minds as they move to the cloud, where they can benefit from faster innovation and break down data silos.”
One misconception Nash often encounters is that companies need to extract all their data from various sources, funnel it into a data lake, and build AI solutions on top of it. “You can do some interesting things there, but the problem with that approach is that the data isn’t always up-to-date, and it’s hardly real-time. This is necessary for AI to be truly efficient and useful.”
To get the most out of agent AI, it’s important to use a unified data fabric
Data lakes also lack the context needed for AI to deliver truly transformative results. Agents, and indeed other types of AI tools, need to fully understand data relationships, such as how purchase orders relate to packing slips, invoices, and customer inquiries. “These limitations can be overcome, but it’s often very difficult,” says Nash.
SAP’s Business Data Cloud offers a different approach. Data from all your organization’s applications, including SAP and non-SAP, is brought together in one curated layer. This creates the foundation for reliable, responsible, and relevant AI. “Having a unified data fabric is key to getting the most out of agent AI, as agents often work across end-to-end processes, which requires accessing data from multiple sources,” says Nash.
Partnering with organizations with deep AI expertise also plays a key role in helping companies navigate this rapidly changing space. “Trying to do everything yourself is very difficult because you may lack in-house AI talent and infrastructure,” Nash says. “There are vendors, consultancies and academic institutions that not only provide specialized skills and advanced platforms, but also help advance proven methodologies.”
For example, she believes it’s important to view AI not as a “one-time tool or investment,” but as a “continuous improvement loop, known as a flywheel effect,” where every component influences the next. In other words, SAP apps power mission-critical business processes that generate large amounts of rich, contextual business data. This data is fed into SAP’s Business AI to enable real-time intelligence, which is then embedded into SAP applications to create smarter workflows, predictive insights, and innovative user experiences.
UK businesses expect their return on investment in AI to nearly double over the next two years, increasing from 17% in 2025 to 32% in 2027. But achieving this will require many companies to move from piecemeal experimentation to strategic, enterprise-wide deployment of technology.
So the way forward for UK businesses is not simply to spend more on AI tools. It’s important to spend smarter to bring distributed experiments out of the shadows, ensure staff get the training they need, and connect fragmented data so that AI can deliver truly transformative results.
For more information, please visit sap.com/uk.
AI has become an important part of board-level discussions around growth, competitiveness, and resilience. However, there are serious challenges at the heart of many UK companies’ approach to technology.
Although they are plowing small fortunes into AI (nearly £16 million on average this year alone, according to new research from SAP and Oxford Economics), only 7% have a strategic, enterprise-wide investment plan for AI. In other words, much of today’s spending is piecemeal, ad hoc, and essentially short-termist.
In practice, this means that while piloting and implementing AI has shown some benefits, the full impact on productivity and growth is yet to be realized as organizations struggle to scale the results across their businesses. This is reflected in the fact that 70% of UK businesses are currently unsure whether AI is realizing its full potential. “Currently, most AI projects are technology-driven and focused on one business process or department, so they are not necessarily aligned with the strategic ambitions of the company,” said Sonia Nash, Head of Business AI at SAP UK & Ireland.
