SAP outlined a wide set of business Artificial Intelligence announcements at SAP TechEd, underscoring progress embedding Artificial Intelligence across its suite. By the end of 2025, SAP expects 400 sap business Artificial Intelligence use cases delivered in its solutions, including 40 Joule agents and 2,100 Joule skills. The company notes that its existing more than 300 use cases translate into 441 million EUR value add for a company with 10 billion EUR annual revenue.
Technical advances focus on data and model foundations. SAP HANA Cloud gained Model Context Protocol support to ground agents in enterprise data, plus knowledge graph engine enhancements planned for Q1 2026 and agentic long-term memory for continuous learning. SAP is integrating Snowflake with SAP Business Data Cloud via SAP BDC Connect, branded as SAP Snowflake, with general availability planned for Q1 2026 and BDC Connect for Snowflake in H1 2026. These moves are presented as part of an AI-native architecture that keeps data context and semantic richness central.
SAP introduced sap-rpt-1, a relational foundation model designed for structured business data and predictive tasks. The model family includes small and large editions for general availability in Q4 2025 in the generative Artificial Intelligence hub within Artificial Intelligence Foundation, plus an open-source release on Hugging Face and GitHub. SAP reports sap-rpt-1 offers up to 2x better prediction quality versus narrow models and 3.5x better quality versus large language models for tabular business outcomes. Artificial Intelligence Foundation additions include model orchestration, evaluation services, and Perplexity integration.
On agents and governance, SAP expanded Joule capabilities with Joule Studio for low-code and pro-code agent development, planned compatibility with an agent-to-agent protocol for cross-vendor orchestration, and centralized control via sap leanix agent hub and sap signavio agent mining. The company also emphasized regional digital sovereignty, expanding SAP cloud infrastructure into Deutsche Telekom’s Munich data center through the Industrial AI Cloud project, and previewed ABAP-focused models and tooling as well as early explorations of embodied Artificial Intelligence with robotics and quantum-ready business algorithms.
