Last month, Anthorpic introduced a new open-source standard for connecting AI assistants, known as Model Context Protocol (MCP).

The the blog post “Introducing the Model Context Protocol”, Anthorpic state:

*“Even the most sophisticated models are constrained by their isolation from data—trapped behind information silos and legacy systems. Every new data source requires its own custom implementation, making truly connected systems difficult to scale.

MCP addresses this challenge. It provides a universal, open standard for connecting AI systems with data sources, replacing fragmented integrations with a single protocol. The result is a simpler, more reliable way to give AI systems access to the data they need.”*

At the highest level, MCP enables developers to build two-way connections between tools, services and data sources. The protocol consists of two parts, MCP servers and MCP clients.

  • MCP Server - The provider, which exposes specific functionality, like access to tools, services or data. The server knows how to receive a request and translate it to ensure compatibility with downstream systems.

  • MCP Client - The consumer, which runs on the application side, makes requests for tools, services or data. The client handles sending requests, receiving results, and passing them to the AI.

The MCP protocol manages this end-to-end process, ensuring everything is standardised and synchronised. It defines how the client and server communicate, what the messages look like, how actions are described, and how results are returned.

On the surface, this approach sounds like an Application Programming Interface (API), as an API and MCP both define how systems interact. However, an API is a general-purpose interface for various applications, whilst MCP is specifically designed to standardise how AI models, specifically Large Language Models (LLMs), interact with external tools, services and data.

Therefore, MCP does not replace the need for APIs. With this in mind, I expect we will see developers “wrapping” existing APIs for MCP.

Assuming widespread adoption, this could be a big step forward in how AI agents operate, unlocking the potential of Agentic AI.

enterprise architecture ea enterprise_architecture artificial_intelligence ai machine_learning ml generative_ai agentic_ai large_language_model llm natural_language_processing nlp transformer data prompt_engineering model_context_protocol mcp anthropic openai chatgpt gpt o1 microsoft google gemini gemini_pro imagen imagen_2 imagen_3 ai_studio project_mariner bard lifeinwork