MCP Server

The built-in MCP server of Xtract Universal.iQ enables SAP data integration with AI platforms that support the Model Context Protocol. A typical use case is creating chatbots that query live SAP data.

How It Works

By using the Xtract Universal.iQ MCP endpoint, an AI agent / chatbot can execute Xtract Universal.iQ web services to access real-time SAP data, such as customer master data or sales order information. This integration allows users to request and receive current SAP data in natural language while the system handles all connectivity and authorization tasks in the background.

When a chatbot uses the Xtract Universal.iQ MCP endpoint, the following process applies:

1. A user sends a request to a chatbot, e.g., “Show all open sales orders."
2. The chatbot identifies the user's intent to access SAP data.
3. The chatbot calls Xtract Universal.iQ's MCP endpoint at https://<hostname>.com/mcp.
4. The MCP server identifies and executes the corresponding service.
5. Xtract Universal.iQ connects to SAP, executes the request, and retrieves live data.
6. The MCP endpoint returns the retrieved SAP data to the chatbot platform.
7. The chatbot formats the response data and presents it to the user in text form.

Example

Activate MCP

To make Xtract Universal.iQ services accessible to AI platforms, the MCP endpoint of Xtract Universal.iQ must be exposed to the internet. To enable the MCP endpoint:

• Set up Transport Layer Security to ensure secured communication to the MCP server.
• Open the Settings menu and activate the checkbox Enable MCP Server endpoint in the Designer/MCP section.
• Open the MCP HTTPS port (default is 8177), e.g., using open port services.

Once the MCP endpoint at https://<hostname>.com/mcp is accessible, you can configure your chatbot / AI agent to consume the MCP endpoint. The following Knowledge Base Articles contain integration manuals for AI platforms:

Langdock

Microsoft Copilot

n8n

Expose Services via MCP

Create Xtract Universal.iQ services that access relevant SAP data. The following settings ensure that the service is recognized and executed when called via the MCP endpoint:

• In the service settings, activate the checkbox MCP enabled. This option exposes the service to the Xtract Universal.iQ MCP server.
  
• MCP Descriptions:
  Provide clear and meaningful MCP descriptions when creating a service. The MCP server uses this description to match user intents to the correct service. A precise description improves recognition accuracy and response quality.
• Parameter Naming:
  Assign clear, self-explanatory names to all custom service parameters, e.g.,for input parameters in SAP transactions. Clear parameter names ensure consistent mapping between user input and service fields.

Use Cases

The following examples provide SAP objects and their corresponding service types for MCP-powered chatbot use cases. Start with common business tasks, expose only the MCP services the chatbot should call, and keep service input clear so the AI can map user requests correctly.

Sales Assistants

Sales representatives spend a lot of time searching for information in SAP. An AI agent, embedded in tools like Microsoft Teams, can answer these queries in real time.

Document Upload instead of Manual Data Entry

Upload a document to an AI agent to automatically extract and validate the relevant data. The Xtract Universal.iQ MCP server then transfers the data to SAP in the background, eliminating transfer errors.

Maintenance and Inventory Management

Employees in warehouses or production can describe issues to an AI agent that identifies and triggers the appropriate SAP process.

Controlled Customer Self Service

After authentication in a customer portal, the verified customer ID can be used for SAP queries in a customer chatbot. This setup in combination with exposing only relevant Xtract Universal.iQ services to the AI agent, ensures that the customer can only access a limited data set of their own data.