When to use Model Context Protocol (MCP)
Key Takeaway
Use MCP whenever an agent needs to call an external tool or API. Unlike MCP, A2A connects agents to other agents.
Model Context Protocol (MCP) is an open standard for how AI models connect to external systems, including their tools, data sources, and APIs. Agentforce has a built-in MCP client, so an agent can call a third-party tool through a registered MCP server without a custom, hand-built integration.
What MCP is
MCP is a vendor-neutral standard that Anthropic developed.
Think of MCP as a USB-C port for AI. USB-C gives devices one standard way to connect to many accessories. In the same way, MCP gives AI models one standard way to connect to many data sources and tools.
Without MCP, connecting an agent to an external system means a manual integration. In a manual integration, you handle all of this work yourself:
- API connectivity
- Authentication
- Data access
- Code
MCP acts as a translator between the systems, so that work is done for you.
For example, through MCP, an agent can use Salesforce data to look up a contact's loan risk score from a third-party banking service. The agent then uses that score to decide its next step in the conversation.