MCP (Model Context Protocol)
An open protocol from Anthropic (late 2024) for connecting AI models to external tools and data sources in a standardized way, like USB-C for AI.
Quick answer
What is MCP (Model Context Protocol)?
An open protocol from Anthropic (late 2024) for connecting AI models to external tools and data sources in a standardized way, like USB-C for AI.
Last updated:
MCP solves a fragmentation problem: every AI tool needs to talk to dozens of systems (Salesforce, Slack, GitHub, calendars, databases), and historically each integration was custom. MCP standardizes the conversation. An "MCP server" exposes a tool/data source through a protocol; any "MCP client" (Claude Desktop, Cursor, increasingly other LLM apps) can use it without custom code.
Salesforce announced support for MCP in 2025, with bi-directional MCP servers for Salesforce data going GA in 2026. This means Claude, ChatGPT Enterprise, and other LLM clients can query Salesforce records, take actions, and write back, all through a single standardized protocol.
For gptfy and similar platforms, MCP is an accelerator: instead of building dozens of custom integrations, they can expose gptfy capabilities via MCP and let any LLM-based client use them.
Related terms
Browse all terms- AI AgentA software entity that takes goal-directed actions autonomously, calling tools, retrieving data, and producing outcomes without step-by-step human guidance.
- LLM (Large Language Model)A neural network trained on massive text corpora to predict and generate text, the foundation behind ChatGPT, Claude, Gemini, and modern AI assistants.
See it in your Salesforce org
See MCP (Model Context Protocol) running in GPTfy
Book 30 minutes with a GPTfy engineer to see how MCP (Model Context Protocol) actually works inside a Salesforce org like yours.
Book a demo