SAP-LLM Integration Showcase with Model Context Protocol

Experience seamless AI-enterprise systems integration: This showcase demonstrates how Large Language Models can intelligently interact with SAP Business Partner data through the Model Context Protocol (MCP). Users can effortlessly create new business partners directly through natural language conversations with their LLM of choice. Built on Java foundation with Spring's MCP implementation.

Architecture Overview ⚫ Prerequisites ⚫ Setup & Configuration ⚫ Building from Source ⚫ Usage ⚫ Troubleshooting ⚫ License ⚫ Contributing & Support

Architecture Overview

This showcase leverages several key technologies:

• Spring AI MCP Framework - Provides the MCP server implementation
• Natural Language Processing - Intelligent parsing of conversational input using AI and LLM
• SAP OData APIs - Allow direct integration with SAP Business Partner services

Prerequisites

Before diving into this novel integration experience, ensure you have the following components ready:

• LLM with MCP Support - A Large Language Model that supports the Model Context Protocol (see supported clients)
• SAP System - SAP environment with an published Business Partner creation service (details see below)
• Java Runtime - Java 17 or higher for optimal performance
• Maven - For building the project from source (version 3.6+ recommended)

Setup & Configuration

1. Configure your SAP Server

Ensure your SAP system has the Business Partner creation API properly activated and accessible. The API endpoint must be reachable from your application environment, and you'll need a user with valid credentials and appropriate authorization for Business Partner creation operations.

2. Choose your LLM

Choose and install an LLM capable of interacting with the MCP. A comprehensive list of LLMs can be found here.

3. Download the Application

Download the latest JAR file from the releases section of this repository, or build from source using the instructions below.

4. Configure the Application

Copy and edit the application configuration file environmentSetup.properties.template to match your environment:

# SAP Connection Configuration
connection.endpoint.createpartner=https://mysapserver.com:8000/sap/opu/odata/sap/API_BUSINESS_PARTNER/A_BusinessPartner
connection.user=userName
connection.password=secretPassword

5. Configure Your LLM

Add the MCP server to your LLM configuration to enable integration:

{
	"mcpServers": {
		"sap-bp-mcp-showcase": {
			"command": "java",
			"args": [
				"-Dspring.ai.mcp.server.transport=STDIO",
				"-Dspring.config.additional-location=file:Path\\To\\environmentSetup.properties",
				"-jar",
				"Path\\To\\sap-bp-mcp-server-1.0.0-SNAPSHOT.jar"
			]
		}		
	}
}

See the MCP documentation for detailed guidance on connecting to MCP servers.

Building from Source

For developers who want to customize the showcase and build it on their own:

# Clone the repository
git clone https://github.com/yourusername/sap-bp-mcp-showcase.git
cd sap-bp-mcp-showcase

# Configure your environment
cp environmentSetup.properties.template environmentSetup.properties
# Edit environmentSetup.properties with your SAP credentials

# Build the project
mvn clean install

This compiles the source code, runs tests, and packages the application into a deployable JAR file in the target directory.

⚠️ Important Note: During the build process, an integration test will execute that creates an actual Business Partner record in your configured SAP system. Ensure your test environment is properly set up to handle these operations.

Usage

Once configured, simply start a conversation with your LLM and begin creating SAP Business Partners using natural language. The MCP integration seamlessly handles all SAP API interactions behind the scenes, providing an intuitive conversational experience.

Example conversation:

"Create a new business partner Jane Doe in SAP"

Troubleshooting

Check log log file: For all errors, please check the log file sap-bp-mcp-server.log

Debug Mode: For a detailed analysis, please consider using the MCP debugging capabilities.

License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details. It does not come with any kind of warranty and it is not intended for productive usage.

Contributing & Support

This is a showcase project demonstrating the powerful potential of MCP integration with enterprise systems. Please consider that it is not intended for usage in production environments. However, please feel free to open a GitHub issue if you have questions, suggestions, or encounter any challenges while exploring this integration approach.

---

Transforming interaction with enterprise systems through intelligent conversation.