
Meet the Authors
SAP reorganized its AI architecture by introducing the SAP Business AI Platform, which combines its AI foundation, Business Data Cloud, and SAP BTP into a single semantic fabric.
The new SAP AI Agent Hub establishes a vendor-agnostic governance layer to manage both SAP and third-party AI agents, addressing growing enterprise compliance concerns.
On-premise SAP ECC customers can now access select AI capabilities if they commit to modernizing through RISE with SAP, tying AI innovation directly to cloud migration progress.
SAP arrived at Sapphire 2026 with a shift toward AI. Agents, an accelerated cloud migration story, and a vision of the enterprise operating with minimal human intervention in routine processes were at the core of the announcements. Three themes anchored the narrative:
- AI moving from an assistant to an active process participant;
- Structural pillars built around deep process knowledge, semantically rich enterprise data, and mission-critical governance;
- An updated and redesigned Joule repositioned from chatbot overlay to the primary gateway through which users and systems engage with SAP’s entire stack.
SAP Business AI Platform: The Contextual Layer Supporting the Enterprise
The centerpiece announcement was the SAP Business AI Platform, a structural consolidation that merges SAP’s AI foundation layer with SAP Business Data Cloud (BDC) and SAP Business Technology Platform (BTP) into a single integrated offering. For practitioners who have watched SAP’s AI portfolio expand in multiple directions over the past two years, this is the architectural rationalization that was overdue.
The platform’s defining feature is what CEO Christian Klein called a “context layer,” a combination of SAP Knowledge Graph, domain models trained on SAP’s own codebase, and SAP BDC functioning as a semantic data fabric spanning SAP and non-SAP data. The distinction matters because enterprise AI agents most often fail not on reasoning capability but on context.
Agents operating on incomplete or semantically hollow data miss process interdependencies and produce outputs that are accurate in isolation but wrong in business terms. SAP is betting that owning this context layer, encoding knowledge such as the difference between a purchase order and a blanket agreement, or how FI-CO flows connect to AR aging, is its structural advantage over hyperscaler AI offerings that lack domain specificity.
Joule Studio 2.0, introduced alongside the platform, is the development environment where partners and customers build on this foundation. It is an intent-based development approach in which users create a product requirements document (PRD) based on the details of their unique landscape. Users can then go to their favored development environment to create a fully contextualized solution based on the PRD. Klein announced that SAP is investing €100 million in its partner ecosystem for agent development, with Joule Studio 2.0 available free of charge, including a one-time no-charge offer through year-end.
Two other components deserve attention. Company Memory, built on the Signavio foundation, is a knowledge management and context graph layer that learns continuously from policy documents, process models, and team communications, including chats and emails. It is a mechanism for capturing and operationalizing the tacit knowledge that disappears when long-tenured employees leave. Joule Work adds an agentic harness with computer and file access, MCP, and Agent-to-Agent (A2A) protocol support, which is the technical scaffolding for multi-agent orchestration across heterogeneous environments.
The SAP AI Agent Hub, housed in LeanIX and targeting GA in Q3, extends governance to both SAP and non-SAP agents. SAP CTO Philipp Herzig confirmed it will be included in the Business AI Platform at no additional charge, a pricing signal that positions SAP as the governance layer of record for the broader enterprise agent ecosystem.
On the partner front, Anthropic’s Claude is embedded in Joule; Nvidia has contributed its open-shell framework for agent safety and operational boundaries; and KPMG has deployed 20 agents on the new platform, targeting $120 million in reduced contract leakage for a single client. Herzig also reiterated SAP’s intent to acquire Prior Labs and Dremio.
For customers managing on-premise estates, Muhammad Alam, Member of the Executive Board responsible for Product & Engineering, confirmed that SAP is enabling a significant portion of AI agents and Joule assistants to work in hybrid landscapes, with the ability to connect to your SAP S/4HANA on-premises and ECC landscapes, available for customers who have started their modernization journey to SAP Cloud ERP.
In the post-keynote analyst Q&A, COO Sebastian Steinhaeuser went further: “Joule assistants and agents are designed for the cloud. That’s where they will run. However, we also want to meet our customers on their transformation journey, and that’s why we’ll enable the majority of our tools on-premise, on SAP ECC, for customers that have already committed to transforming, so that they can benefit from AI while they are modernizing.” AI access tied to a commitment to modernization, not stasis, is the nuance every on-premises organization needs to absorb.
Autonomous Suite: The Longer Arc
Running in parallel with the Business AI Platform was the introduction of the Autonomous Suite, SAP’s vision for process domains operating end-to-end with minimal human hand-holding. It spans autonomous finance, spend, supply chain, HCM, and CX, and is developed under an ISO-certified process designed for SOX audit compatibility. GA is targeting Q3 or later. The SOX and ISO framing signals that SAP is building for the CFO and Chief Compliance Officer, not just the CIO.
Alam put the current footprint in numbers: 224 agents and 51 assistants spanning autonomous finance, spend management, supply chain, HCM, and customer experience. Early customer deployments include financial close at multiple enterprises, autonomous sourcing at Novartis, and product design at Kaiser Compressor; proof points that the ROI story is no longer hypothetical.
On migration, the updated RISE and GROW with SAP offerings include on-site architects and consultants, a contractual commitment to activate three Joule assistants in Year 1, and an AI-powered ERP migration platform targeting up to 50% reduction in migration effort. Palantir and accenture are named partners for complex migration scenarios. JPMorgan Chase’s SAP partnership and H&M’s Store Intelligence Agent and AI-powered Store Concierge were both featured as enterprise validation, with full customer profiles to follow.
What This Means for SAPinsiders
The real differentiation lies in the context layer. The enterprise AI race has shifted from which LLM scores highest to which platform delivers the most operationally accurate outputs in real business processes. SAP customers should pressure-test any AI vendor claim against one question: Does this agent understand my process, or just my data? Only one of those produces outcomes you can stake a financial close on.
The hybrid access structure creates a strategic window and a deadline. SAP’s commitment to enabling agents for RISE-journey customers, including SAP ECC customers who have made a modernization commitment, removes the “wait until we’re fully in the cloud” rationalization. AI access is tied to forward motion. Organizations that are still deferring cloud decisions now face a concrete competitive cost for that delay.
Governance for non-SAP agents belongs on the 2026 planning agenda. As organizations proliferate agents across Salesforce, ServiceNow, Microsoft, and homegrown tooling, the absence of a cross-vendor governance layer is an audit and compliance gap in the making. The SAP AI Agent Hub’s inclusion at no extra charge is easy to overlook amid the volume of announcements. It shouldn’t be. Whether organizations take SAP’s offer or build their own, the governance question needs an owner and a plan before agent sprawl becomes agent risk.




