The SAP Business Research Retreat 2026 took place on 26 June 2026 at SAP Labs Munich, bringing together researchers from academia and industry for a day dedicated to exchanging insights on the latest developments in business AI research.

The event featured a broad range of topics, including tabular data reasoning, multimodal learning, knowledge graphs, large language model (LLM) agents, KV-cache compression, and agentic enterprise AI systems. The diversity of themes reflected the increasingly interdisciplinary nature of AI research in enterprise environments.

Throughout the day, poster sessions complemented the main program and encouraged continued exchange beyond the formal agenda. These interactions fostered discussions on early-stage ideas and emerging research directions, contributing to a highly interactive and collaborative atmosphere.

Talks of the day included:

• Different Flavors of Tabular Reasoning – Madelon Hulsebos (CWI)
• An Overview of Recent Research Output from the Business Foundation Models Team at SAP – Sam Thelin & Mojtaba Nayyeri (SAP)
• Leveraging and Understanding Multi-Modal Representations – Stefanie Jegelka (TUM)
• “Competitive Moat” or “Protocols”: How Corporate AI Strategy Can Steer a (Knowledge Graph) Research Agenda – Felix Sasaki (SAP)
• The More You Know: Towards Knowledgeable AI – Gerard de Melo (HPI)
• Benchmarking and Optimizing Tool-Calling LLM Agents for Enterprise Settings – Atin Ghosh (SAP)
• From KV-Cache Compression to LLM-Augmented Data Systems – Paolo Papotti (Institut Eurécom)
• Agentic Environment Engineering for Enterprise AI – Yipeng Li (SAP)