SAP SE (NYSE: SAP) is expanding the tooling it offers enterprises to build, govern and operate AI agents across business workflows. The SAP TechEd updates to Joule Studio, Joule Work and SAP Business AI Platform focus on an enterprise concern that has become more urgent as agents gain access to systems of record: controlling what an agent can use, decide and execute.
The company is positioning its platform as a common layer for building and contextualizing AI across SAP and third-party environments. Joule Studio combines no-code and pro-code development tools, while the broader platform connects agents to SAP Business Data Cloud and SAP Knowledge Graph. SAP said customers can choose from more than 70 AI models and retain ownership of their data, which it says is not shared to train or improve large language models.
For IT teams, the more consequential change is the emphasis on agent orchestration and policy enforcement. SAP said new agentic-thinking capabilities in Joule Work will plan and coordinate the steps needed for complex, multistep workflows across SAP landscapes. The company also described interoperability features that let third-party agents invoke Joule Agents through a governed agent gateway, with enforced access control intended to preserve governance as agents work across providers and business domains.
SAP is adding a separate Enterprise AI Gateway to centralize authentication, policy enforcement, routing, monitoring, usage tracking and auditing for AI traffic. That design addresses a problem many enterprises now face: once multiple business teams connect agents to multiple models and data sources, identity, policy and cost controls cannot remain isolated inside individual applications. A common control point can give security and platform teams a clearer way to establish guardrails and inspect how AI services are being used.
The company also said Joule Studio is gaining observability and administrative controls that expose agent behavior, model usage and spend. SAP plans a time-bound promotion under which select customers can deploy and run custom agents on the SAP-managed Joule Studio runtime without charge through December 2026, while design-time tools for building and testing agents and n8n workflows will remain free under fair-use limits. The business value of that offer will depend on what happens after the promotional period, but it lowers the near-term barrier to testing governed agent workflows.
Data context remains central to SAP’s approach. The company said new Knowledge Graph capabilities will bring in customer-specific SAP and non-SAP data, allowing users to query their data in natural language and receive answers grounded in that information. SAP Business Data Cloud Ontology Manager is intended to help organizations model relationships among data from different teams, systems and applications so agents can reason across those boundaries with more context.
SAP is also extending its tabular AI work. SAP-RPT-2.0 is intended to add anomaly and outlier detection, embeddings and multi-table analysis, while SAP-RPT-TS targets forecasting uses involving time-indexed business metrics. SAP said the models can reduce manual engineering, though enterprises will still need to validate model performance, data quality and workflow controls in their own environments.
The update shows enterprise software vendors competing on more than model access. Legacy AI deployments often begin as isolated assistants bolted onto individual processes. SAP’s strategy is to make agent construction, data context, interoperability and audit controls part of the same business-application platform—an approach that could give organizations a more manageable route to production agents than disconnected tools with limited process awareness.