TL;DR — Key Takeaways
-
SAP doubles down on agentic AI: New Joule capabilities and support for MCP and A2A aim to bring hundreds of AI agents into enterprise workflows.
-
Governance takes center stage: SAP Enterprise AI Gateway introduces centralized controls for AI access, security, monitoring and compliance.
-
Autonomous enterprise ambitions grow: Expanded data integration, AI models and partnerships with NVIDIA, Intel and HPE push AI beyond conventional business applications.
SAP this week added a bevy of extensions to its portfolio that collectively embed artificial intelligence (AI) capabilities deeper into workflows.
Announced at its TechEd event, organizations can now use Joule Studio and Joule Work to invoke more than 70 AI models via an SAP Knowledge Graph that has been extended to third-party data sources. There is now an SAP Business Data Cloud Ontology Manager in SAP Business Data Cloud (SAP BDC) that will provide an AI-native authoring experience for building ontologies across SAP and third-party data.
At the same time, SAP is also adding support for a framework that makes its more than 300 AI agents developed by SAP and its partners discoverable via the Model Context Protocol (MCP) and the Agent2Agent (A2A) protocol. Additionally, SAP is previewing support for n8n workflows with its SAP Joule Studio tool for securely building agentic AI workflows using the OpenShell framework developed by NVIDIA. Joule Studio has also been extended to add observability and administrative controls to give developers and administrators visibility into agent behavior, model usage and spend.
SAP also unveiled an SAP Enterprise AI Gateway to enable organizations to centrally manage access to AI services, including authentication, policy enforcement, routing, monitoring, usage tracking and auditing. There is also now a Joule Assistant and SAP AI Agent Hub and an AI Governance Assistant to make it simpler to use natural language to enforce policies.
The company also revealed it is allying with Dremio to add native support for Apache Iceberg and Apache Polaris data frameworks and extending its relationship with Databricks to enable that platform to run natively on the SAP data lake platform without any restrictions. The company’s latest tabular AI model, dubbed SAP-RPT-2.0, which it gained with the acquisition of Prior Labs earlier this year, is being extended to add support for anomaly and outlier detection, and richer AI-powered analysis through embeddings. SAP also revealed it has, in collaboration with Hewlett-Packard Enterprise (HPE), deployed SAP-RPT on the HPE Spaceborne Computer-2 aboard the International Space Station (ISS).
Finally, SAP revealed it is working with Intel to extend the reach of its AI platforms to physical AI applications involving, for example, robots developed by Boston Dynamics.
SAP CTO Philipp Herzig told conference attendees that these capabilities will make it possible to add probabilistic AI technologies to ensure more accurate output is generated within business workflows that are largely deterministic in the sense they need to be performed the same way each time. That level of integration is not going to be possible by relying solely on general-purpose AI models that lack the context needed to generate outputs that organizations can trust, he added.
The overall goal is to leverage AI to create a truly autonomous enterprise by not just adding AI agents but also reinventing the machinery under the hood, noted Herzig. “The bar has never been higher, he said.
Just as importantly, organizations do not necessarily always have to build and maintain those AI agents themselves, noted Herzig.
It’s not clear at what rate enterprise IT organizations are moving to embed AI agents into workflows, but at this juncture it’s more a question of when rather than if.
