Autonomous Business: Joule Work, Joule Assistants, and the SAP Business AI Platform

Under the theme of the „autonomous enterprise,“ SAP showcased the integration of the new AI user interface, Joule Work, domain-specific Joule Assistants, and the integrated SAP Autonomous Suite. The consolidated SAP Business AI Platform now includes the following components: the cloud portfolio, the Business Technology Platform (SAP BTP), the Business Data Cloud (SAP BDC), and the AI Core Framework. To underscore its capabilities, SAP cites internal experience: With approximately 110,000 of its own employees, Joule Work is already in productive use and has led to demonstrable productivity gains averaging 20 percent in key administrative and logistics areas.

The transition from ad hoc assistance systems to autonomously acting AI agents is by no means merely a technical upgrade, but rather a fundamental shift in responsibility and cost structures. While SAP markets the Autonomous Suite as a way to free users from repetitive routine tasks, in operational reality, existing SAP customers face the challenge of reconciling probabilistic language models with the precision and audit-proof nature of a deterministic general ledger.

The technical core of the announced ecosystem is the Joule Work interaction layer, which allows users to initiate and execute cross-system tasks using natural language—both on mobile devices and on the desktop. To suppress stochastic hallucinations in business-critical operations, the assistant relies on the SAP Knowledge Graph. This knowledge graph maps over seven million data fields and relationship networks within the S/4 universe and provides the AI models with the necessary business context. When simple voice commands are not sufficient, specialized Joule Assistants step in to execute multi-step workflows directly within the system, based on established role and authorization models.

Low-Code and Pro-Code

To customize and expand this agent landscape, SAP provides Joule Studio, a low-code and pro-code development environment. Business users and developers can choose from more than 70 language and AI models and design complex multi-agent workflows using the visual orchestration platform from SAP’s integrated partner, n8n. To address the pressing threat of uncontrolled agent installations, SAP is relying on an open interoperability concept: The Agent2Agent (A2A) protocol and the integration of the Model Context Protocol (MCP) are designed to ensure that even third-party AI agents can be integrated into the system in a controlled manner.

The AI Agent Hub serves as the central control mechanism within the SAP LeanIX enterprise architecture solution, providing a centralized directory of all agents active within the company. This architecture is supported by a governance structure certified to the international standard ISO/IEC 42001, as well as the Business Transformation Management toolset—comprising SAP Signavio, LeanIX, and WalkMe—to analyze process bottlenecks and monitor AI usage within the company. In addition, SAP entered the payments market directly with the embedded payment solution SAP Pay (powered by Tereina) to automate invoice processing and payment reconciliation within the ERP workflow without the need for manual file exports.

Behind the facade of this orchestrated platform architecture, however, lie significant business pitfalls and architectural contradictions for existing SAP customers. The proclamation of the autonomous age is preceded by massive pressure from international financial markets: Because investors fear that autonomous AI agents will devalue traditional workplace licenses—and thus the traditional software subscription model—SAP is frantically searching for new sources of revenue.

„The autonomous company is in the business world
Here we are: with Joule and the SAP Autonomous Suite,
”which is based on the SAP Business AI Platform."

Christian Klein,
Chairman of the Board,
SAP

The result is a shift away from predictable flat rates toward usage-based billing models based on AI Units, BTP Capacity Units, and unpredictable AI token prices. Anyone who wants to use the new agents must purchase credit points in advance, which expire without exception at the end of the year if they have not been used up due to a lack of productive use cases. In practice, this principle of prepayment with an expiration date means that customers tie up significant portions of their budgets, while actual AI consumption in day-to-day operations often falls far short of marketing expectations.

Furthermore, the latest data from the German-speaking user association DSAG reveals a deep divide between corporate announcements and the reality on the ground at the operational level. Although 43 percent of DSAG members have already implemented specific AI projects, 77 percent of companies actively using AI rely on non-SAP solutions from hyperscalers such as OpenAI, Microsoft, or Anthropic. The reason for this is the vendor’s attempt to push hesitant existing customers into the expensive “Rise with SAP” cloud subscriptions by exclusively offering AI innovations through them—a move that the on-premises user base views as a breach of investment guarantees.

The Business Data Cloud’s data foundation is also derided by the community as „Business Data Complexity“ due to confusing license clauses and restrictions on OData data transfer. The situation is exacerbated by the restrictive SAP API Policy (Version 4.2026), which acts as a digital customs barrier by regulating direct bulk data extraction for external AI agents and forces customers to route their agent workloads through the fee-based BTP infrastructure.

Governance and the EU AI Act

From the perspective of sustainable IT management, the implementation of the SAP Autonomous Enterprise therefore requires the highest level of governance and financial safeguards. An autonomous agent that makes erroneous procurement or payment decisions due to “hallucinations” or insufficient master data quality triggers chain reactions in the core ERP system, for which the user company is ultimately and without exception liable under the European Union’s AI Act. Existing SAP customers should not be swayed by vendor narratives but should demand that every AI application be evaluated against business criteria. This can only be achieved through a consistently streamlined system core (Clean Core), adhering to human final decision-making (“Human in the Loop”) for business-critical processes, and utilizing open standards such as the Model Context Protocol (MCP) to preserve true digital sovereignty without creeping vendor lock-in.

„This is a groundbreaking moment for enterprise AI,“ said Christian Klein, CEO of SAP, in Las Vegas, USA. „The autonomous enterprise has arrived in the business world. With Joule and the SAP Autonomous Suite, which is based on the SAP Business AI Platform, we’re helping companies like Novartis, Morgan Foods, Nestlé, and PwC seamlessly integrate intelligent capabilities across all business functions to deliver trustworthy results.“

At SAP Connect, these customers shared how they are already reaping tangible benefits from the concept of the autonomous enterprise across all areas. „Together, we developed a pilot project for SAP’s Joule Sourcing Assistant and recently launched it,“ explained Christoph Bürki, Head of Procurement at Novartis. „Our vision is for AI agents to handle labor-intensive manual tasks so that procurement experts can focus on strategy, collaboration with suppliers, and value creation.“

„With our SAP S/4HANA transformation across 140 countries, we are creating a connected digital core with standardized processes, reliable data, and greater transparency across markets and business units,“ said Luca Dell’Orletta, Global Head of IT Innovation and Enterprise Architecture at Nestlé. „Building on this foundation, Joule and the Accounts Receivable Management Assistant can automatically prioritize debt collection tasks based on risk and the specific business context, recommend the best next course of action in each case, and help our finance teams respond more quickly to changes.“

Joule Work is now being made available to the first customers as a new AI user interface. This enables employees to interact with the SAP system in their own language and complete tasks across SAP applications and data sources, as well as third-party data sources. Joule leverages the business context of the SAP Knowledge Graph. „After 28 years as an SAP customer, our collaboration as part of the Early Adopter Program for Joule Work Desktop has generated more momentum in just a few months than in all the years prior,“ said Peter Wied, Director of Complexity Management at Maschinenfabrik Reinhausen. „What used to take up to an hour of manual effort in our SAP S/4HANA environment can now be accessed in seconds using natural language. This is fundamentally changing how our teams work with business data.“

Business AI Platform

With the SAP Business AI Platform, companies can centrally develop AI solutions, provide contextual information, and manage these solutions for both SAP and third-party environments. With Joule Studio, business users and developers can create and extend capabilities and agents in no-code and pro-code workflows, choosing from more than 70 AI models. SAP Business Data Cloud and the SAP Knowledge Graph provide these agents with trustworthy business data and business context. This enables companies to select the right model for every use case and better manage the benefits and deployment of AI.

SAP also announced at SAP Connect in Las Vegas that it would enter the payments market with SAP Pay, powered by Tereina, an SAP company. SAP Pay is directly embedded in SAP cloud ERP applications, processes payments within the same workflow, and reconciles them when an invoice is due. This reduces manual file handling and dependencies on external processing, while giving customers more control. (pmf)