- SAP HANA Cloud now features integrated triple store for knowledge graphs.
- New Python package enables seamless LLM integration for natural language queries.
- Automated Sparkle query generation simplifies complex data retrieval.
At SAP TechEd 2025, a groundbreaking developer demo showcased how SAP HANA Cloud, fortified with a knowledge graph engine and a new Python package, is revolutionizing data interaction through large language models (LLMs). This innovation allows users to query complex data using natural language, dramatically simplifying access and analysis.
The demonstration highlighted the activation of a triple store within an SAP HANA Cloud instance, transforming it into a robust knowledge graph engine. This capability enables the creation of intricate ontologies, exemplified by mapping SAP developer advocates and their diverse demo presentations, all stored as readily available triples within the knowledge graph.
A key component of this advancement is a newly developed Python package, designed for seamless integration between SAP HANA and popular LLM frameworks like Longchain. This package facilitates a sophisticated Q&A mechanism, connecting to the knowledge graph and leveraging LLM models from SAP AI Core's GenAI Hub. The demo vividly illustrated how a natural language question, such as "What demo will show cloud application programming model and who can show this?", is processed by the LLM.
The LLM intelligently interprets the query, generating the necessary Sparkle code automatically—a task traditionally requiring specialized expertise. The system then returns precise answers, identifying specific demos like "SAP Cap MCP and AI agents" and the presenters, Nico and DJ. Furthermore, the underlying Resource Description Framework (RDF) allows for linking these internal knowledge graph resources to external web content, creating a richer, more interconnected data landscape.
This integration marks a significant leap forward in making complex SAP data more accessible and actionable, empowering developers and business users alike to extract insights with unprecedented ease and efficiency.
“LLM is getting this question generating the sparkle query. I know all of you would love to write this by yourself, but let LLM do this and then it returns the answers.”
- Witalij Rudnicki, Developer Advocate, SAP
