AI EngineeringInterview preparation
Intermediate·Concept·2 min read

MCP versus REST, function calling, and plugins

Practice this question

Overview

  • REST is an HTTP API design style for application services.
  • Function calling is a model interaction mechanism for proposing structured operations.
  • MCP standardizes integration discovery and invocation/context exchange; a plugin is a packaging/integration mechanism whose exact meaning depends on its host.

Why this matters

Understanding these layers prevents rebuilding a backend or expecting a protocol to orchestrate the agent.

An everyday example

A menu, ordering language, kitchen process, and restaurant app are different layers of the same service.

A technical example

Model function call → MCP client → MCP tool server → existing REST claim API.

How it works

Use model tool calling for decisions, MCP for standardized integration, and domain APIs for authoritative operations.

Architecture and flow

LLM tool proposal → host policy → MCP invocation → REST/business API → result.

The insurance assistant example

The insurer keeps existing claim APIs and adds a narrow MCP facade for approved AI hosts.

When to use it

When multiple AI hosts need consistent discovery and tool/context access.

When to avoid it

Do not replace all internal REST APIs with MCP or describe MCP as a model or agent framework.

Interview answer

“Function calling is the model's decision interface, MCP is the integration protocol, and REST can remain the business-service interface behind it.”

Benefits, trade-offs, and production considerations

Added protocol layers introduce network latency, authentication flows, and operational ownership. Keep the MCP facade narrow and independently testable.

Adapted from 13_AI.MD · “MCP versus REST, function calling, and plugins”. Code examples are study fragments. Difficulty and core labels are editorial estimates; version-specific behavior should be checked against your project.