What Is MCP?

MCP (Model Context Protocol) is an open standard, originally introduced by Anthropic, that allows AI agents and large language models (LLMs) to automatically discover and interact with external tools and services. Instead of requiring manual API documentation or custom integrations, an MCP-enabled platform exposes its capabilities in a format that any compatible AI can read and act upon directly.

The protocol has been widely adopted across the software industry. Atlassian (Jira, Confluence), GitHub, and GitLab all support MCP. VELOX is now part of this ecosystem.

‍

The VELOX MCP Server

The VELOX MCP Server is a microservice module that acts as a translation layer between VELOX's REST APIs and AI agents. It allows any MCP-compatible LLM to connect to a VELOX-powered webshop and perform actions through natural language, without custom integration work.

The module is available now. It is optional and can be activated independently, like any other VELOX module.

‍

Customer-Facing Capabilities

Once connected, an AI assistant can:

• Search and browse products using natural language
• Retrieve product details, attributes, and availability
• Complete an order on behalf of the user, end-to-end

‍

This enables conversational commerce: a customer can instruct their AI assistant to find and order a product without visiting the webshop directly. The AI handles the search, selection, and checkout via the MCP Server.

The same capability powers on-site AI chatbots. Rather than redirecting a customer to a product page, a chatbot can take the customer through the full checkout process within the conversation.

‍

Back-Office and Integration Use Cases

The MCP Server also enables AI-driven automation of internal processes. Any operation supported by VELOX's APIs can be triggered and managed by an LLM. Practical applications include:

• Querying product catalogues and inventory data via AI
• Automating order management tasks through natural language commands
• Integrating RAG (retrieval-augmented generation) tools with shop data
• Connecting a single preferred LLM to all VELOX-powered workflows, without adopting separate AI tools for each platform

‍

This last point is significant for operations teams: rather than managing multiple platform-specific AI tools (each with different interfaces, pricing, and learning curves), teams can use one LLM of their choice across all connected systems via MCP.

‍

For more information, contact René Hämmerli to learn more.