Between SAP Worlds and AI Agents

The event focused on production-ready AI agents, the consistent opening of the platform for leading models, and the further deepening of the partnership with SAP. The SAP Garden in Munich's Olympic Center, which opened just last year, provided an impressive backdrop for the world tour's stop in southern Germany. This can also be seen as a statement of the intensification of the cooperation.

The keynotes by Daniel Holz, VP Central EMEA, Tanya Nitzsche, Director Enterprise Sales, and Samuel Bonamigo, SVP & GM EMEA, from Databricks focused on the latest developments in Agent Bricks, the framework that Databricks first introduced in June. The platform has been specifically expanded to make it easier for companies to move from the pilot phase to productive use.

Governance and transparency

The focus was on three key dimensions of trust: accuracy, governance, and openness and interoperability. With MLflow for Agent Quality and Observability, which is now generally available, organizations can for the first time seamlessly track every agent interaction, evaluate results according to individually defined criteria, and obtain structured feedback from subject matter experts.

Databricks has thus established a standard for qualitative agent evaluation that can be used in any environment. At the same time, governance has received a significant boost. With AI Gateway, all models and external MCP servers can be centrally managed and controlled via Unity Catalog. The openness of the platform has also been further enhanced. Agent Bricks now supports an even wider range of models, frameworks, and MCP servers, enabling the orchestration of multi-stage workflows across internal and external systems.

Databricks and SAP

In a stage discussion, Daniel Holz and Thomas Duschek from SAP reviewed developments surrounding SAP Databricks. The deeply integrated solution, which will become part of the SAP Business Data Cloud, enables companies to seamlessly merge SAP data with other company data and use it in both Databricks and SAP environments.

For many organizations whose business-critical processes are based on SAP, this is an important breakthrough. For the first time, rich, context-rich SAP data can be combined with modern AI and data intelligence, all while fully preserving its semantics. In addition to Agent Bricks, the event featured other product highlights. Lakebase and AI/BI Genie, two central components of the Databricks Data Intelligence Platform that are already generally available, were presented. The company also announced the native availability of models such as OpenAI, Google Gemini, and Claude 3.7 Sonnet on the platform.

In live demos, DataBricks experts showed how companies can use these models productively—for example, for document intelligence, BI automation, or process optimization. Partner companies such as You.com and Glean emphasized in their statements how seamlessly their services can be integrated into Databricks workflows thanks to the MCP catalog and how quickly agents can access current or unstructured information with it. However, the event was not only product-focused. German companies such as Aldi Süd and Bayer provided insights into their projects where AI is already in productive operation today.

They demonstrated how data platforms and agent technologies are becoming a real driver of innovation—especially where data from SAP systems converges with other sources of information. Overall, it became clear what strategic role Databricks intends to play in the SAP ecosystem and beyond in the future. The platform is growing into a central hub for enterprise AI: open, interoperable, fully controllable, and ready for productive use. The Munich event can be seen as a signal that the era of experimental AI prototypes is coming to an end—and that the transition to scalable, with trustworthy agent systems now within reach.

SAP BDC Connect

The SAP Business Data Cloud Connect to Databricks, released in October, also created a simple, bidirectional connection. Companies can use it to connect their SAP Business Data Cloud (BDC) to Databricks in just a few clicks—for bidirectional live and zero-copy data sharing based on Delta Sharing.