At SAP TechEd 2026 in Berlin, SAP presented developments intended to take enterprise AI beyond question-answering and analysis toward autonomous execution of multi-step business processes. The announced features let AI plan, coordinate and carry out sequences of operations while preserving human oversight at key decision points.
How the AI agents operate in practice
A common corporate scenario is a supplier invoice whose total differs from the purchase order amount. Resolving such discrepancies today can require multiple employees and several sequential steps: finding related documents, verifying data, consulting stakeholders and obtaining approvals. SAP’s new AI agents aim to automate much of this work by gathering and linking relevant information, coordinating necessary applications and actors, and preparing the inputs needed for human decisions.
The Joule AI assistant has already been helping users with day-to-day tasks; the new capabilities go further than a chat interface. On the Joule Work surface, users can express desired outcomes in natural language and the system can orchestrate the applications, data sources and AI agents required to achieve them.
The orchestration is not limited to SAP products: agents from other vendors can participate in workflows, subject to access controls and security checks. SAP will ship preconfigured AI assistants and agents across five core enterprise functional areas, which can be adapted to industry-specific needs.
Joule Studio and developer options
SAP is expanding the Joule Studio development environment so customers can build AI agents tailored to their operations. Business users define tasks and rules; IT teams build and test the agents. Joule Studio offers developers a choice of more than 70 AI models, and early adopters can create automated workflows using n8n integration.
SAP will allow eligible customers to run their own AI agents in SAP-managed Joule Studio environments free of charge through the end of December 2026. Development and testing tools will remain available for free within specified limits after that date. The Joule Studio Developer certification, focused on building enterprise AI agents, is also available free until December 31, 2026.
Understanding business context is essential
A central challenge for enterprise AI is that necessary information often resides in separate systems: finance, procurement, manufacturing and sales may all use different databases, and the same business event can appear in different forms. Effective AI therefore requires not just data access but understanding how information relates and which business rules apply.
SAP’s Business Data Cloud and Knowledge Graph enhancements help AI form a more accurate picture of company data, processes and operating rules. For example, analyzing a stock shortage can incorporate warehouse levels, expected orders, supplier data and production plans.
SAP stresses that customers retain full control of their data: the company will not hand over customer data to third parties for training or improving large language models.
Integrations with partners support data linking: Dremio enables queries across systems without copying data, Databricks expands analytics and AI development capabilities, and Reltio supports unified management of customer, supplier and product data so AI can rely on more consistent information.
New models, space applications and document processing
SAP introduced AI models that broaden forecasting and anomaly detection capabilities. The SAP-RPT-TS model for time-series analysis is already available, while a new anomaly-detection model is planned for first-quarter 2027.
In a notable use case, SAP and Hewlett Packard Enterprise (HPE) announced that the SAP-RPT model could be run on the HPE Spaceborne Computer-2 aboard the International Space Station. Local analysis in orbit can help detect technical anomalies, predict failures, estimate missing sensor data and compile status reports without needing to send the data back to Earth.
In document processing, SAP Document AI enhancements aim to extract key information from diverse invoice formats, contracts and other documents, reducing the need for manual data entry and verification.
Oversight, compliance and physical automation
As companies entrust more tasks to AI, oversight becomes crucial: organizations must know what data AI can access, what actions it may take autonomously and which steps require human approval. SAP is adding tools to support secure, transparent operation.
The SAP Enterprise AI Gateway provides a single control surface for supervising enterprise AI systems: it centralizes access management, tracks AI-tool usage and monitors associated costs. SAP plans to introduce this gateway in the first quarter of 2027.
To support compliance with the European AI Act, SAP offers the AI Governance Assistant, which helps assess the risk categories of AI applications, identify required documentation and flag potential compliance gaps. The tool produces preliminary assessments that legal and compliance teams must review.
SAP is also exploring physical automation: in collaboration with Intel, the company is working to connect business IT systems with autonomous robots. Integrations with Docusign, Parloa and Redwood are intended to reduce manual handoffs and enable faster workflows with less human intervention.
Implications for Hungary and concluding notes
Hidvégi Péter, Managing Director of SAP Hungary, said: “The true value of enterprise AI is not shown by how cleverly it answers a question, but by how much real work it can reliably do. For that, AI must know the company’s data and processes, and leaders must clearly see what it does and under which rules. For Hungarian companies, this can be the path from impressive pilots to tangible business results.”
Overall, SAP’s announcements aim to make enterprise AI more practical and controllable — extending automation and analytics while strengthening governance, data control and developer readiness.
