What is the Model Context Protocol (MCP)?

The Model Context Protocol (MCP) is an open standard through which AI models and agents access external data sources, tools, and systems. Instead of building a separate interface for every model-and-system combination, an MCP server exposes its capabilities once in a standardized way, and any MCP-capable client can use them. For process data this means an agent can query the process model, metrics, and case histories without a bespoke integration.

Why does this matter?

The limiting factor for agentic AI is rarely the model but access to reliable context. An open access standard decouples the tool landscape from the model in use and prevents organizations from permanently locking themselves to a single vendor.

What does this look like in practice?

Instead of building a separate connector from each of two models to SAP, Jira and the data warehouse — six integrations — each system exposes an MCP server and every model uses the same client. Six integrations become four, and each additional model costs none at all.

Not to be confused with

Not to be confused with an ordinary REST API: MCP standardises how a model discovers and calls tools and data sources, not what any individual system offers. Nor with one vendor's function calling: MCP is open and works across providers.