SAP argues that enterprise AI agents must be grounded in a company’s own context using knowledge graphs and vector-embedded data to move beyond generic assistants toward agentic systems capable of executing real business processes.

Cloud revenue rose 24% in SAP’s Q2, helping push total revenue up 11% to $11.24 billion, even as services revenue dipped as the company reallocates resources to AI initiatives and enterprise AI offerings.

Strategic partnerships with Google Cloud, Microsoft, Snowflake, and Nvidia aim to enable agent-to-agent interoperability, data integration, and security for SAP’s AI platform.

SAP is investing in its AI transformation by hiring AI talent and reskilling employees, aiming to train more than 90% of its workforce in the coming months.

To align its own knowledge with customers’ diverse landscapes, SAP is acquiring and integrating tools like LeanIX and Signavio and embedding automation platforms like n8n to map and understand non-SAP systems within Joule Studio.

SAP launched the SAP Business AI Platform at Sapphire in May, featuring the Autonomous Suite to add AI agents to existing SAP applications for end-to-end process automation.

Governance, identity, and security are central to scalable autonomous agents, leveraging SAP’s process governance, anomaly detection, and ML-based validation to monitor agent behavior and enforce access controls.

The strategy includes modernizing older on-premises systems to avoid throughput bottlenecks and ensure agents can operate effectively as enterprises scale autonomous capabilities.

Klein highlighted concerns about AI models understanding business data and governance, stressing AI sovereignty and reducing vendor lock-in amid high AI token spend.

Acquisitions like Dremio and Reltio are pivotal for unifying and governing diverse data landscapes to support AI agents across workflows.

Klein asserted SAP’s goal to become an autonomous enterprise by shifting from traditional software development to building AI at scale across operations.

Joule, SAP’s generative AI assistant embedded in cloud apps and the Business Technology Platform, requires proper provisioning of access rights for users and agents to prevent permission circumvention.