From Cornerstone to Control Plane for Enterprise AI

Readers will appreciate from our earlier interview with Sylvie Sollod, Global Vice President of Platform Product Marketing for SAP BTP, that SAP BTP’s role in 2026 has evolved and now settled as a part of the SAP Business AI Platform as the foundation of SAP’s AI strategy.

The open question is execution. Whether customers use integration modernization, clean core methodology, and governed AI to turn that foundation into one working unit, rather than treating each as a separate project.

In an interview recorded before SAP Sapphire 2026, Steffen Pietsch, Head of SAP BTP Product Management at SAP, framed the shift plainly in an SAPinsider Expert Exchange conversation with Robert Holland. The strategic case for business AI, he said, is largely won. The question now is more about, “How do I apply it in the intended way,” he noted. That maturity showed up again after the interview, when SAP confirmed at Sapphire 2026 that SAP BTP became the governed foundation of the new SAP Business AI Platform rather than being displaced by it.

Pietsch called SAP BTP “the cornerstone of modern SAP landscapes for almost every enterprise regardless of the industry or size of the organization or geography.” His point? The role has not changed even if the branding has. What changed, though, is that SAP’s AI now relies on platform capabilities.

Clean Core Is the Operating Discipline

The through-line connecting everything Pietsch discussed is clean core, and he was quick to correct its reduction to custom-code hygiene. “Clean core is more than just extensibility,” he said, listing five dimensions: business processes, extensibility, integration, data, and operations. “Manage all five, and you can truly speak about managing your core in a clean and lean fashion,” Pietsch observed.

The conversation highlighted two shifts that matter for practitioners. SAP moved from a strict, near-binary clean-or-not judgment mode to a four-layer model where organizations can better assess how clean they are, reflecting real environments that are rarely fully compliant.

Pietsch also pushed back on a persistent misconception. “Clean core does not mean you must not develop,” he said. “The discipline is decoupling, accessing SAP code only through released APIs so that you still build differentiating innovations without breaking upgrades. BTP is where that decoupled innovation lives.”

Where Clean Core Meets AI Readiness

The most urgent clean core dimension in 2026 is integration, which has a deadline. Pietsch urged customers still on SAP PI/PO to “not postpone it too far,” citing the end of maintenance for that stack. But SAP Integration Suite is not a like-for-like swap. According to Pietsch, it is an enterprise-wide iPaaS solution that supports all the integration patterns, spanning process, API-led, event-based, and hybrid scenarios.

That breadth reframes the migration. Replacing PI/PO with SAP Integration Suite still means moving some existing data flows from A to B; the technology differs, but the interfaces persist. The mistake is stopping there. A migration that does nothing more than recreate legacy flows satisfies a near-term transition while leaving the modernization value on the table. In contrast, rationalizing integration patterns during the move reduces complexity and builds the governed base on which AI depends.

The economics back the case. Pietsch cited a joint customer study showing SAP Integration Suite customers realize, on average, more than $500,000 in cost savings from retiring legacy integration systems, alongside an ROI of more than 345%. The AI link is explicit: Agentic AI benefits from an event-driven architecture that can trigger an agent or a workflow, and governing APIs and MCP connections run through Integration Suite.

Governed, Open, and Built for Enterprise Data

Pietsch concluded by calling SAP Business AI platform “truly the foundation for everything AI at SAP,” with governance as the differentiator. This means access to leading LLMs in a governed, secure fashion, plus tooling such as the Prompt Optimizer. He was candid about model limits, noting, “LLMs tend to hallucinate or cannot handle tabular data. That is why SAP built RAPID for structured enterprise data. And SAP’s platform stays open in all dimensions.”

What This Means for SAPinsiders

Do not treat clean core as just code cleanup. CIOs and enterprise architects should assess cleanliness across business processes, extensibility, integration, data, and operations, using SAP’s gradual maturity model rather than a pass/fail lens. That is because fragmented cleanup surfaces later as upgrade friction and AI-readiness debt. SAPinsiders should, therefore, score all five dimensions before the next major release.

Put the PI/PO modernization decision on a clock. ERP program managers should scope the free SAP migration assessment now and decide which interfaces to modernize with event-driven and API-led patterns versus which to retire. A lift-and-shift forfeits the $500K-plus savings and 345% ROI Pietsch cited, and end-of-maintenance narrows the options. Thus, organizations should map interfaces to patterns before committing to a cutover date.

Build AI on governed data and hold announcements to the roadmap. AI leaders should route custom AI through SAP Business AI Platform’s governed LLM access and pair unstructured-data LLMs with structured models like RAPID where accuracy matters. Ungoverned APIs and hallucination-prone models cap enterprise AI reliability.

Your next steps:

Watch the full SAPinsider interview with Steffen Pietsch here

Then look at the next video in the Experts Exchange Series where we speak with Jesper Schleimann.