Automated identification of SAP roles and Fiori apps using AI agents

In complex SAP S/4HANA system landscapes, determining the correct business roles for specific tasks involves a significant amount of manual effort. Users need to know which Fiori or SAP GUI applications cover a particular activity, in which business catalogs they are included, and which roles must be derived from them.

This information is scattered across various systems and knowledge sources, such as the SAP Fiori Apps Library, authorization concepts, or customer-specific extensions. Manual research requires in-depth SAP expertise, is time-consuming, and carries the risk of incorrect assignments. Support teams, business departments, and new employees, in particular, face significant barriers to entry as a result.

We developed an AI-powered use case for automatically identifying suitable SAP roles and Fiori apps based on natural language queries. The solution is built on the SAP Business Technology Platform (BTP) and combines modern agent architectures with standard SAP services.

At the core of the solution is a ReAct agent that implements the Agent-to-Agent (A2A) protocol and is seamlessly integrated into SAP Joule. The agent accesses various data sources in a structured manner via Model Context Protocol (MCP) servers:

• An OData service in SAP S/4HANA provides information on applications, business catalogs, and assigned roles.
• A dedicated MCP server for the SAP Fiori Apps Library enables high-performance searches across approximately 14,000 Fiori apps using app IDs or business search terms.

Upon receiving a request, the agent analyzes the relevant business entities and actions, identifies suitable applications, evaluates hit lists, and derives the required business catalogs and roles from them. The entire infrastructure is deployed in production on BTP Cloud Foundry and fully embedded in the SAP context.