Artificial Intelligence
What is an LLM and how is it used in SAP processes?
Learn how large language models work with SAP data through RAG, controlled tools and enterprise authorization boundaries.
What is an LLM and how is it used in SAP processes?
An LLM, or large language model, can understand and generate natural language. In an SAP scenario, the model does not connect to the system by itself. It operates through a secure integration layer, authorized data sources and explicitly defined tools.
How is SAP context provided to an LLM?
Three common methods can work together:
- RAG retrieves relevant passages from enterprise documentation and supplies them as context.
- Query tools retrieve SAP data within the user’s authorization.
- When an action is required, only the permitted tool is invoked under a defined access mode.
The model handles interpretation and language generation, while the SAP connection layer preserves identity, target-system and authorization controls.
Which use cases benefit?
Ready-to-run SAP AI workflows can cover dump and request analysis, functional and technical specification drafting, support consulting and controlled actions. Enterprise-specific processes can combine specialist agents, data sources and human approvals in Flow Builder.
What does a secure implementation require?
- Ground answers in accessible SAP data or approved documentation
- Use per-user SAP identities and least privilege
- Manage read and write tools separately
- Add human validation for consequential output
- Audit tool calls and redact sensitive fields