During the SAP NOW AI event in Rotterdam, one message stood out clearly: AI only becomes valuable when it closely aligns with business processes.
Bart Van der Biest, director of SAP in the Benelux, shares this conviction during a conversation at SAP NOW. “Almost every customer conversation I have today also involves AI,” he says. “It’s about the beating heart of business operations: productivity, good governance, and the secure integration of AI into the ERP system and associated business applications.”
read also
SAP Now Rotterdam: Standardizing to Truly Move Forward
According to Van der Biest, it’s useful to distinguish three levels. First, there’s embedded AI: intelligence that has been “under the hood” in SAP solutions for many years, making daily work faster and more consistent. Above that come agents, which automate or prepare specific tasks, and finally assistants that enable agents to work together and bridge the gap with the end user. “Building an agent on top of one scenario isn’t that difficult,” he qualifies. “What’s difficult is orchestrating agents at scale, with authorizations, logging, monitoring, and a uniform way of control. That’s where our AI hub comes in.”
Control Room for AI
This AI hub is simply put the control room for AI in the SAP environment. It makes prompts and tasks model-agnostic and manages the connection with processes and transactions. “Technical teams need to realize how much is under the hood,” says Van der Biest. “At the front end, you communicate in natural language. But behind that, translation to the right SAP transactions is needed. We do this with a knowledge graph and a semantic layer that gives context to the data.” The latter is crucial, he emphasizes: without consistent metadata and semantics, AI outcomes quickly become unreliable.
Standardize where possible and automate where it makes sense.
Bart Van der Biest, Director SAP Benelux
Meanwhile, AI is also shifting towards implementation and change management. With Joule for Developers and Joule for Consultants, SAP aims to accelerate repetitive implementation tasks. Think of converting commands during migrations or generating documentation. Van der Biest speaks of a “factory approach”: standardize where possible and automate where it makes sense. “Much of consultants’ and developers’ work is repetitive. Recognizing and reusing patterns is exactly where AI excels.”
Productivity Gains
Why should large companies and governments invest in this today? Not just for speed. According to Van der Biest, the first tangible benefits lie in productivity gains and error reduction. “In certain back-office scenarios, we realistically target at least 30 percent productivity gains,” he says. “Whether you then need fewer people or deploy the same people for more valuable tasks is an organizational discussion. But the gain is there.”
At the same time, he tempers the hype: fully autonomous agents in a complex business context are the exception rather than the rule. “Most business tasks are not 100 percent defined. AI provides an enormous assist: visibility, suggested actions, and automatic handling of standard cases. Humans remain the decision-makers for deviations.”
Authorizations, Security and Compliance
Governance is another major pillar at SAP NOW. Many organizations fear shadow AI and data leaks through standalone co-pilots or public LLMs. Van der Biest: “In a business context, you need authorizations, security, and compliance. Exactly what SAP excels at. In our approach, access is ID-based: Joule ‘knows’ who you are and what you’re allowed to do. If you ask for the CEO’s salary without rights, you won’t get this data. It works just like in the underlying application.” This one-to-one translation of existing rights models to AI interactions reduces risk and adoption barriers.
On the technology front, SAP advocates for openness: not a one-size-fits-all model, but a platform that can drive multiple models depending on the use case. “The question ‘which model do you prefer?’ rarely has one answer,” states Van der Biest. “It depends on the task you want to solve. That’s why we keep the platform open and let the AI hub handle the orchestration.”
Data sovereignty
Related to this, he points to data sovereignty, which is particularly relevant for the public sector and critical infrastructure in the Benelux. “Many clients are subject to strict regulations,” he says. “Being a European player helps. We continue to offer all alternatives: hypersecure on a hyperscaler when possible, or fully European-hosted when necessary. The most important thing is freedom of choice and keeping AI close to the processes and data.”
How does a large organization begin without getting lost? Van der Biest sees a pragmatic path. Start with embedded AI where benefits are immediate, such as in planning, finance, or services. Meanwhile, get your semantic layer and authorization model in order. Set up the AI hub as an orchestration point, so pilots don’t become islands. Then add agents for clear, repetitive scenarios. Assistants should ensure agents work together intelligently. Measure systematically with ROI assessment and scale what works.
Less Talk
“AI has existed for a long time,” he concludes. “The difference today is the acceleration through generative AI and the maturity of integration. Those who combine this with good data management, governance, and process discipline see quick results. Those who implement it as a standalone toy mainly get headaches.”
The tone at SAP NOW AI was clearly Flemish pragmatic: less talk, more processes. AI that yields returns in SAP doesn’t look like magic, but like well-regulated normality. And that’s exactly what large companies and government organizations need. As Van der Biest summarizes: “Technology alone brings little value. Technology in function of your applications and processes: that’s where the value lies.”
This is a submitted commercial contribution from SAP. The editorial team is not responsible for the content.