Why Business AI Is Becoming the New ERP Paradigm

SAP uses the term „Autonomous Enterprise“ to describe a new generation of enterprise software in which AI not only supports processes but increasingly orchestrates, prioritizes, and executes them independently. The key message is this: ERP systems are evolving from „systems of record“ to „systems of execution.“ SAP views the ERP as the brain of the enterprise—SAP AI agents, like users, have access to this brain and use it more efficiently and effectively. For companies, this represents a paradigm shift. Business AI is not merely changing individual functions within SAP systems; it is fundamentally transforming the role of ERP—and, with it, the way companies will manage, make decisions, and operate in the future. SAP’s role is also evolving from that of an ERP provider to that of a business AI provider.

Business AI Corporate Culture

The first wave of AI in businesses was primarily characterized by assistants and co-pilots. These systems helped generate text, summarize information, or streamline processes. SAP is now taking things a step further. With Joule, AI Agents, and the Business AI Platform, an architecture is increasingly emerging in which AI is no longer viewed as an add-on feature, but rather is embedded in the company’s end-to-end processes. This is precisely what makes Business AI strategically relevant. The decisive change does not lie in ERP systems becoming „smarter.“ Rather, the key point is that processes can in the future be executed, prioritized, and orchestrated autonomously by agents. For companies, this represents a fundamental transformation and a paradigm shift: In the past, users processed tasks manually, step by step. Today, they are increasingly defining rules, governance, and objectives, while the system (with the help of agents) carries out operational tasks independently. This changes not only how ERP systems are used, but also role definitions, governance structures, and operating models.

BTP as an Orchestration Platform

At the heart of this development is the Business Technology Platform (BTP). It is increasingly becoming the technological foundation for intelligent business processes. The role of the BTP extends far beyond traditional extension or integration scenarios. It is becoming the central orchestration and governance layer for hybrid AI architectures. The platform not only bundles SAP Joule and the AI Agents, but also data and integration services as well as knowledge graphs. In addition, external AI models and non-SAP systems can be connected. This is particularly strategically relevant for midsize companies. In practice, ERP landscapes rarely consist exclusively of SAP systems. Production systems, logistics platforms, CRM solutions, and industry-specific applications must remain integrated. The BTP thus becomes the key development and operations platform for processes and embedded AI services. It is particularly important to note that AI only works reliably in an enterprise context if it can access semantically understandable data structures. If the data foundation is not sound, even the best AI models will not deliver reliable results. That is why SAP is investing heavily in areas such as the Business Data Cloud, where data is consolidated, standardized, and thus made usable for AI. In addition, the focus is on concepts such as „Company Memory“—that is, a company’s digitally available knowledge—as well as knowledge graphs, which serve as an intelligent map of this knowledge, making connections and relationships visible. The innovation lies not only in the language model but in the ability to interpret business objects, relationships, and processes in context. Looking ahead, SAP will consolidate the currently separate elements—Business Technology Platform, Business Data Cloud, and Business AI Foundation—into a unified SAP Business AI Platform, in which the BTP will serve as the cornerstone for development, integration, and governance.

Joule Transforms ERP Interaction

This shift in strategy is particularly evident in Joule. The AI assistant is increasingly becoming the central control layer for business processes. In the future, users will no longer primarily control processes via transactions or user interfaces, but will instead issue commands using natural language. At first glance, this sounds like a convenience feature. In reality, however, it represents a profound architectural shift. Interaction is shifting from clicking to issuing commands. AI agents access corporate data, utilize system functions, and can influence operational workflows. Companies must therefore clearly define which tasks autonomous systems are permitted to take on, which data they can use, and where human oversight remains necessary.

Traceability becomes particularly important in this context. Decisions and process steps must remain explainable even when they are triggered by AI agents. Auditability, clear role models, and controlled access rights thus become fundamental prerequisites for productive use. Governance is therefore evolving into an ongoing management task. „Human in the Loop“ is increasingly becoming „Human above the Loop“: Humans no longer execute every single step, but rather monitor, evaluate, and control autonomous systems. Monthly closings, supply chain analyses, purchase approvals, or cash flow simulations can in the future be initiated via Joule and orchestrated by various AI agents in the background—across both SAP and non-SAP systems. As a result, the traditional user interface is becoming increasingly less important. The actual logic is shifting to process and agent layers.

This opens up enormous potential for companies. Repetitive tasks can be automated, the workload on business units is reduced, decision-making is accelerated, and IT teams gain the freedom to innovate. The impact is particularly significant where high process volumes intersect with clearly defined rules—for example, in finance, procurement, or the supply chain. When selecting suitable use cases, it is helpful to collaborate with external partners, whose role in the AI era is also evolving beyond that of mere implementers: What’s needed are orchestrators, integration partners, and governance advisors who work with companies to define business goals, build operating models, and continuously refine AI scenarios.

Data Quality as a Prerequisite

As ambitious as the vision of the Autonomous Enterprise is, one key insight from Sapphire is equally clear: No AI agent can compensate for poor data quality. This brings an issue back to the forefront that has been a challenge for many companies for years: data quality and governance. Historically evolved SAP landscapes, custom extensions, data silos, and heterogeneous data models have so far prevented the scalable use of AI in many companies. That is why Clean Core is becoming the strategic foundation of modern AI architectures. Companies must harmonize their ERP landscapes, standardize processes, and consolidate data structures before autonomous process models can function effectively. „Garbage in, garbage out“ takes on new significance in the AI era. After all, poor processes remain poor processes even with AI—they just run faster.

So the motto shouldn’t be: “We need better AI,” but rather, “We need better data and more context.” Today, data comes from a wide variety of sources in different formats—such as SAP, supplier portals, Excel spreadsheets, or emails. If AI accesses contradictory or duplicate data without sufficient context, it cannot make reliable conclusions. A shared understanding—or a common language—across all data is needed to build AI properly.

Especially among small and medium-sized businesses, there is currently a certain discrepancy between vision and reality—and not just when it comes to data quality. This is because many companies are still grappling intensively with the S/4 conversion, migration to the cloud, or the shortage of skilled workers. Cybersecurity also remains high on the agenda for many companies: All for One’s „Innovation Compass 2026“ shows that IT security is now perceived as the most important driver of impact for transformation projects, even ahead of investments in AI and the cloud.

The challenge, therefore, is not so much to introduce as many AI tools as possible as quickly as possible. Rather, the key is to make your company’s architecture AI-ready.

Successfully Developing an AI Roadmap

For many companies, the question is no longer whether AI will be part of their ERP strategy, but how they can develop a robust AI roadmap.

In practice, it has been shown time and again that successful AI strategies do not start with the agents, but with the fundamentals. These include harmonized processes and clean data models—the so-called “Clean Core.” Equally crucial are clear governance structures with defined roles, rules, and responsibilities, as well as stable integration architectures and a sustainable operating model. Only once this foundation is in place can scalable AI scenarios be developed. At the same time, companies should avoid viewing AI solely as a technology project. The actual transformation involves processes, roles, and organizations. Iterative approaches are therefore particularly successful. Instead of betting on a “Big Bang” approach, companies should start with clearly defined use cases that quickly create operational value and foster acceptance among the workforce. Afterward, additional AI projects can be rolled out step by step along specific business processes.

At the same time, companies must establish governance with clear structures and, above all, clear authorizations: Who sets the rules? Who monitors agents? How are compliance, security, and transparency ensured?

This also changes the role of IT and SAP partners. Companies need partners who not only implement technology but also orchestrate hybrid environments, establish governance, and ensure a sustainable capacity for innovation.

Operational Readiness: A Topic for the Future

The most important insight is not that SAP is massively expanding its AI capabilities. The real change runs deeper: ERP systems are evolving from documentation platforms into intelligent execution platforms. For companies, the competitive advantage of the coming years will therefore not stem solely from access to AI models. What will be crucial is the ability to organize processes, data, and operating models in such a way that AI can function reliably. The future, therefore, does not belong to the companies with the most AI tools or pilot projects. It belongs to the companies that implement the Autonomous Enterprise in a company-wide, end-to-end operating model.

To the partner entry: