Tech in Five - What is MCP?
John Savill gives a quick, practical overview of MCP (Model Context Protocol), explaining why it matters for integrating AI with external tools and services. He frames MCP as a standard way for models to discover capabilities and interact with “target services” through local or remote MCP servers.
Overview
Integration challenges with AI
- The video starts from the common problem of integrating AI systems with external services and tools.
- It positions MCP as a way to reduce friction and inconsistency when connecting models to different “target services.”
What a target service can do
- The concept of a target service is introduced as the thing an AI-enabled solution wants to interact with.
- The key question becomes: what capabilities does that service expose, and how can an AI system discover and use them reliably?
MCP as “the USB-C for AI”
- MCP is described as a standard interface (an analogy to USB-C) intended to make connecting AI systems to tools/services more consistent.
Capability discovery
- A core idea highlighted is capability discovery: enabling an AI client to determine what actions/capabilities are available via MCP.
MCP server capabilities
- The video discusses the role of an MCP Server and the capabilities it exposes.
Local or remote MCP server
- MCP servers can be hosted:
- Locally
- Remotely
What MCP changes
- MCP is presented as a shift toward a more standardized way for AI systems to integrate with external capabilities, reducing bespoke, one-off integrations.