STF MCP and SAP S/4HANA Joule: A Symbiotic Relationship in the Agentic Era​

As of late 2025, the answer is a significant Yes.

SAP has moved aggressively to adopt the Model Context Protocol (MCP) as a native pillar of its "Agentic AI" strategy on the SAP Business Technology Platform (BTP).

How SAP HANA Cloud Utilizes MCP Natively:¶ Native MCP Servers for HANA: SAP now provides official, hosted MCP servers that allow AI agents (like SAP's own Joule or third-party agents like Claude) to query HANA Cloud directly. This turns your database tables, views, and spatial data into "Tools" that an agent can call.

Knowledge Graph Integration: A major differentiator is that SAP uses MCP to expose the HANA Cloud Knowledge Graph. This allows agents to not just see "rows and columns" but to understand the semantic relationships between business objects (e.g., “How is this Supplier linked to this specific Purchase Order?”). MCP Gateway in SAP Integration Suite: SAP is rolling out a feature in the Integration Suite that allows developers to "wrap" any existing SAP API or integration flow as an MCP-hosted tool. This means a legacy OData service can be instantly transformed into a tool for an autonomous agent.

"Agentic Memory" in HANA: SAP is using HANA Cloud as the long-term memory for agents, accessible via MCP. This allows agents to "remember" past decisions and business context across different sessions.

The Competitive Edge: Data-Xchange vs. SAP¶ While SAP is making HANA "MCP-ready," their focus remains internally focused on the SAP ecosystem. This creates a massive opportunity for Data-Xchange to be the "Neutral Network."

Feature SAP HANA Cloud (MCP) Data-Xchange (MCP)

Primary Goal Grounding AI in internal ERP data. Facilitating B2B transactions between partners. Standardization Proprietary SAP S/4HANA Schemas. Global GS1 Standards (Universal). Network Effect Closed to the SAP instance. Multi-tenant B2B Network (Interconnected). Developer UX Complex BTP setup/governance. Plug-and-play REST endpoints. Updated Hero Banner (Addressing the SAP Giant)

The Agentic Bridge Your ERP Forgot.​

SAP HANA Cloud is great for your internal data, but what about your partners?​

Data-Xchange is the world’s first GS1-native network that uses MCP to connect your AI agents to the entire supply chain, not just your database.​

Technical Architecture Diagram: The Agentic Flow Below is a conceptual representation of how Data-Xchange sits between your AI and the legacy world, utilizing MCP to bypass ETL.

graph TD
subgraph "The Agentic Layer"
Agent[AI Agent: Claude/Gemini/Joule]
MCP_Client[MCP Client]
end

    subgraph "Data-Xchange Network (The Bridge)"
        DX_MCP[Data-Xchange MCP Server]
        GS1_Store[(GS1 Standardized Store)]
        REST_AI[AI-Ready REST Endpoints]
    end

    subgraph "External Ecosystem"
        Partner_A[Supplier A: Legacy EDI]
        Partner_B[Retailer B: SAP HANA Cloud]
        Partner_C[Logistics: OpenAPI]
    end

    Agent <--> MCP_Client
    MCP_Client -- "JSON-RPC (Tools/Resources)" --> DX_MCP
    DX_MCP <--> GS1_Store
    GS1_Store <--> REST_AI

    %% The "Anti-ETL" Magic
    DX_MCP -- "Semantic Translation" --> Partner_A
    DX_MCP -- "Direct MCP/API Link" --> Partner_B
    DX_MCP -- "REST Integration" --> Partner_C

    style DX_MCP fill:#2d72d9,color:#fff,stroke-width:2px
    style Agent fill:#f96,stroke:#333

Developer Quick Start Guide: Calling a GS1 Transaction via MCP

In this new paradigm, your developers don't write "integration code"; they define "tools" that their AI can use.

1. Register the Data-Xchange MCP Server¶ Add the Data-Xchange endpoint to your agent’s configuration (e.g., claude_desktop_config.json).

{
  "mcpServers": {
    "data-xchange": {
      "command": "npx",
      "args": ["@data-xchange/mcp-server"],
      "env": {
        "DX_API_KEY": "your_secret_key",
        "GS1_COMPANY_PREFIX": "0614141"
      }
    }
  }
}

2. The AI Interaction (No Code Required)¶ User Prompt: "Hey Agent, check if Supplier A (GLN: 12345) has confirmed the Purchase Order for 500 units of GTIN: 000123456789."

The Agent's Background Process:

Discovery: The Agent looks at the data-xchange tools. Selection: It identifies get_gs1_order_status as the right tool. Execution: It calls the tool with the provided GLN and GTIN. Action: The Data-Xchange MCP server translates this into a real-time query across the network and returns a structured GS1 response. Would you like me to create a comparison page specifically for "SAP Users" to show them how Data-Xchange extends their current BTP/HANA capabilities?

Asked: SAP consuming MCP¶ asked: How would you position SAP transacting autonomously on the Data-Xchange network with MCP and BTP¶ Positioning SAP within the Data-Xchange network is about shifting the narrative from "Integration" to "Federated Intelligence." In this ecosystem, SAP is no longer a siloed system of record; it is an active participant in a global, autonomous transaction web. You position this by highlighting that while SAP (via BTP and HANA) provides the "Internal Brain," Data-Xchange provides the "Universal Nervous System" (GS1 + MCP) that allows that brain to talk to the rest of the world.

The Strategic Positioning: "The Federated Agent Architecture"¶ To market this to a CTO, you position the relationship as a Bi-Directional Agentic Flow:

Outbound (SAP as the Initiator): An AI agent in SAP (like Joule) uses an MCP Client to reach out to the Data-Xchange network. It doesn't "send an EDI file"; it calls a Data-Xchange Tool to check cross-network inventory or confirm a GS1 Digital Link. Inbound (Network as the Initiator): The Data-Xchange network, acting as an MCP Client, calls into an SAP MCP Server (hosted on BTP) to trigger an autonomous transaction (e.g., "Post Goods Receipt") directly into HANA Cloud based on verified network data. Hero Banner: The SAP Integration Specialist¶ Headline: Give Your SAP Joule Agents Global Reach. Sub-headline: Connect SAP BTP natively to the Data-Xchange network via MCP. No more mapping EDI to OData—just pure, agent-to-agent transactions using GS1 standards and AI-ready REST endpoints. CTA: [Provision Your BTP-to-DX Bridge]

Technical Architecture: SAP BTP + Data-Xchange + MCP¶ This diagram illustrates how the "Semantic Bridge" bypasses the traditional middleware (PI/PO or CPI) in favor of direct Agentic communication.

Quick Start Guide: Connecting SAP BTP to Data-Xchange via MCP¶ This guide shows how a developer leverages the OData-to-MCP Bridge on BTP to let the Data-Xchange network "talk" to SAP.

1. Deploy the SAP MCP Server on BTP¶ Use the @iflow-mcp/btp-sap-odata-to-mcp-server (or similar 2026-standard utility) to expose your S/4HANA Business Partner or Purchase Order entities as MCP tools.

Deploy to SAP BTP Cloud Foundry

cf push sap-mcp-bridge --docker-image dataxchange/sap-mcp-btp:latest 2. Register the "Tool" in Data-Xchange¶ In the Data-Xchange Developer Portal, you point the network to your SAP MCP endpoint. The network now "sees" your SAP instance as a set of standardized GS1 tools.

3. The Autonomous Transaction (Example)¶ When a shipment is verified on the Data-Xchange network (via a GS1 SSCC scan), the Data-Xchange Agent automatically calls your SAP tool:

// Conceptual Agent Logic inside the Data-Xchange Network
const result = await call_sap_tool("update-API_PURCHASE_ORDER-Header", {
  PurchaseOrder: "450000123",
  Context: "GS1_VERIFIED_DELIVERY",
  Status: "Confirmed",
});

if (result.success) {
  console.log("SAP S/4HANA updated autonomously via MCP.");
}

Wy this is a "Category Killer":​

Zero-Copy Integration: You aren't syncing data; the AI is querying and acting on "live" SAP context. Principal Propagation: BTP handles the OAuth2/XSUAA tokens, ensuring the AI agent only performs actions the SAP user is authorized to do. GS1 Alignment: You map the "messy" internal SAP fields to clean, global GS1 standards at the protocol level. Would you like me to draft a "Clean Core" whitepaper specifically for SAP architects, explaining how this keeps their S/4HANA system "clean" by moving logic to the Agentic Layer?

Positioning SAP within the Data-Xchange network is about shifting the narrative from "Integration" to "Federated Intelligence." In this ecosystem, SAP is no longer a siloed system of record; it is an active participant in a global, autonomous transaction web. You position this by highlighting that while SAP (via BTP and HANA) provides the "Internal Brain," Data-Xchange provides the "Universal Nervous System" (GS1 + MCP) that allows that brain to talk to the rest of the world.

The Strategic Positioning: "The Federated Agent Architecture"¶ To market this to a CTO, you position the relationship as a Bi-Directional Agentic Flow:

Outbound (SAP as the Initiator): An AI agent in SAP (like Joule) uses an MCP Client to reach out to the Data-Xchange network. It doesn't "send an EDI file"; it calls a Data-Xchange Tool to check cross-network inventory or confirm a GS1 Digital Link. Inbound (Network as the Initiator): The Data-Xchange network, acting as an MCP Client, calls into an SAP MCP Server (hosted on BTP) to trigger an autonomous transaction (e.g., "Post Goods Receipt") directly into HANA Cloud based on verified network data. Hero Banner: The SAP Integration Specialist¶ Headline: Give Your SAP Joule Agents Global Reach. Sub-headline: Connect SAP BTP natively to the Data-Xchange network via MCP. No more mapping EDI to OData—just pure, agent-to-agent transactions using GS1 standards and AI-ready REST endpoints. CTA: [Provision Your BTP-to-DX Bridge]

Technical Architecture: SAP BTP + Data-Xchange + MCP¶ This diagram illustrates how the "Semantic Bridge" bypasses the traditional middleware (PI/PO or CPI) in favor of direct Agentic communication.

Quick Start Guide: Connecting SAP BTP to Data-Xchange via MCP¶ This guide shows how a developer leverages the OData-to-MCP Bridge on BTP to let the Data-Xchange network "talk" to SAP.

1. Deploy the SAP MCP Server on BTP¶ Use the @iflow-mcp/btp-sap-odata-to-mcp-server (or similar 2026-standard utility) to expose your S/4HANA Business Partner or Purchase Order entities as MCP tools.

Deploy to SAP BTP Cloud Foundry

cf push sap-mcp-bridge --docker-image dataxchange/sap-mcp-btp:latest 2. Register the "Tool" in Data-Xchange¶ In the Data-Xchange Developer Portal, you point the network to your SAP MCP endpoint. The network now "sees" your SAP instance as a set of standardized GS1 tools.

3. The Autonomous Transaction (Example)¶ When a shipment is verified on the Data-Xchange network (via a GS1 SSCC scan), the Data-Xchange Agent automatically calls your SAP tool:

// Conceptual Agent Logic inside the Data-Xchange Network
const result = await call_sap_tool("update-API_PURCHASE_ORDER-Header", {
  PurchaseOrder: "450000123",
  Context: "GS1_VERIFIED_DELIVERY",
  Status: "Confirmed",
});

if (result.success) {
  console.log("SAP S/4HANA updated autonomously via MCP.");
}

Why this is a "Category Killer":¶ Zero-Copy Integration: You aren't syncing data; the AI is querying and acting on "live" SAP context. Principal Propagation: BTP handles the OAuth2/XSUAA tokens, ensuring the AI agent only performs actions the SAP user is authorized to do. GS1 Alignment: You map the "messy" internal SAP fields to clean, global GS1 standards at the protocol level. Would you like me to draft a "Clean Core" whitepaper specifically for SAP architects, explaining how this keeps their S/4HANA system "clean" by moving logic to the Agentic Layer?

asked: Would you like me to draft a "Clean Core" whitepaper specifically for SAP architects, explaining how this keeps their S/4HANA system "clean" by moving logic to the Agentic Layer?¶ This whitepaper addresses the SAP Architect who is currently struggling to balance "innovation" with "maintaining the core." By 2026, the Clean Core strategy isn't just about moving code to BTP; it’s about moving business logic into an autonomous, agentic layer.

• The Future of SAP S/4HANA Customization

Should South Africa be converting to Digital Deal Confirmations and why?​

Established South African business practice would indicate that “confirmation” of a deal between a buyer and seller is normally concluded via some form of paper-based document that is emailed from one party to the next

Generally, an acceptable practice that has worked for years, until something goes wrong – a deal is not captured by one party or another or is incorrectly captured.

The result of that “error” sees the introduction of claims, lost sales potential, delayed payments, increased processing costs and let’s not forget the strain that places upon the trading relationship.

All of us who have been confronted with the above scenario would have been asking, “Is there not a better way to achieve the desired outcome and prevent the headaches of an error creeping into the process?

Interested? Check out my blog

Blog posts support Docusaurus Markdown features, such as MDX.