Special agents: the changing face of AI in business and how it’s driving ROI

Businesses have already seen what AI can achieve through successful pilots. The challenge now is scaling adoption across the organisation and embedding AI into the way businesses run.

As Christian Klein, SAP’s CEO, noted in May this year: “Today, most companies are experimenting with AI. Many of them can point to demos that impressed, pilots that worked, and tools that saved time in narrow tasks. Far fewer can say AI has changed their business across functions, processes, and teams.’’

Scaling AI across an organisation requires more than simply deploying additional tools. Critical business processes often span multiple departments and systems, drawing on data from different sources. For AI to have meaningful impact, businesses need access to reliable, connected data grounded in the right context.

According to SAP, the requirements for trust, context and governance become significantly greater as AI moves closer to complex business execution. Establishing these foundations is essential to ensuring AI can operate reliably and deliver long-term, sustained returns. James Bates, head of industry and customer advisory at SAP UK and Ireland, spoke on AI investment, ROI and challenges facing organisations during a recent special episode of the ITPro Podcast.

“If I think back to some of that foundational build-out when it comes to AI, the ROI for some of those things I think is always going to be challenging, but that investment is building for the future,” he said.

“If [businesses] aren’t investing in processes, data governance, (and) context, then actually I think that will slow them down, whereas investing in that now I think really accelerates that ROI going forward.”

Embracing agents built on strong foundations

One of the biggest evolutions in AI in the past 12-24 months has been the birth of agentic AI. Stanford University’s Human-Centered Artificial Intelligence Institute describes agentic AI as “oriented around ongoing task execution –– breaking down objectives, (and) coordinating steps”. This means making use of tools like web browsers and APIs to achieve aims with “minimal human oversight within defined constraints”.

It’s here where some of the real value of the future may be found.

According to SAP’s The Value of AI report, produced in conjunction with Oxford Economics, 87% of UK businesses feel agentic AI has a “moderate to very high” potential to transform their organization. Almost two-thirds (64%) of businesses have already piloted agentic use cases worldwide, with the UK ahead of the curve at 70%.

“AI largely to this point has been fantastic in terms of productivity use cases,” said Bates. “But if you think about the prospect of agentic AI and its ability to transform end-to-end processes, where actually you'll have agents really driving actions – albeit still orchestrated, I would stress, by people … there’s definitely a lot of excitement about it.”

That’s not to say every AI implementation has to be about agents, though.

“If I reflect on some of the conversations that I'm having with customers, a lot of it is around that potential of agentic AI. However, I wouldn't ignore some of the value of what [AI is] able to deliver today,” said Bates. He drew attention to some of the figures from the SAP and Oxford Economics research, which showed investment by the average UK company had risen from £16.2 million in 2025 to £20.6 million in 2026. Expected ROI has also grown from 17% in 2025 to 22% in 2026.

“If we think about where customers are investing and working and deploying capabilities today, it's also in building some of the foundations around AI,” he continued. “So things like data processes, skills, governance frameworks. I often think, actually, that the organizations and customers that will win in the long term are those that are focused on building those initial foundations rather than perhaps experimenting straight away with agents.”

AI skills don’t have to be specialized

One of the things that’s often less explored, when it comes to AI, is the questions of skills and AI. In fact, it’s something Bates will be speaking about during SAP NOW AI Tour UKI at the Birmingham NEC on 22 October.

During his appearance on the ITPro Podcast, Bates was asked how the UK is dealing with the relative lack of AI specialists and workers with AI experience in the workplace.

“I think that the tendency could be to think ‘well, do we need to go out and hire X number of AI experts?’,” said Bates. “People that understand the models and how they're architected will be important but, actually, I think the skills question is a question for the entire organization.”

“If I reflect on what we're working on, we're looking at this in terms of all roles. So what skills are needed to support AI enablement, regardless of what your role is,” Bates continued. “It's not about [needing] to create different roles, it's how AI can empower and drive productivity for all roles.”

At SAP, this has meant building a community of 9,000 AI ambassadors that spread understanding of how everybody can use AI to the greatest effect, regardless of what role they are in.

“There’s this industry term of … ‘democratizing AI’. I agree with that. I think this is something for all of us,” Bates said.

There’s still time to register for SAP NOW AI Tour UKI – click here to join SAP in Birmingham.

Download SAP’s The Value of AI report now