MCP is Model Context Protocol. It is an open standard protocol that enables seamless integration between LLM applications and external data sources and tools. MCP integration standardizes and streamlines the interaction between AI models and tools / services (e.g. Dataverse data).
MCP was introduced by Anthropic and is an open-source framework. It is an open protocol that standardizes how apps provide context to LLM. It enables AI systems to seamlessly access and utilize external tools, systems and data sources
MCP provides a standardized way to connect AI models and connect agents that can perform more complex tasks and access real time information
MCP acts as a common language for AI applications and utilizes a client-server like model.
For example:
Model Context Protocol makes it more efficient and effective for developers to apply Dataverse’s rich data capabilities
The MCP architecture is shown below:
Model Context Protocol Architecture diagram consists of following components:
- MCP Hosts: Programs like Claude, AI Tools, IDE, Copilot Studio
- MCP Client: Protocol clients that maintains connection with servers
- MCP Servers: Lightweight programs delivering specific capabilities
- Local Database: Internal database services in your local network
- Remote Service: Services available on the Internet
Why do u need MCP?
With the advent of AI and proliferation of numerous AI tools there is a need of integrations that Large Language Model (LLM) can plug in to consume various services and provide an accurate and timely response to the prompt requested by users.
Also, with MCP servers, It is easy to switch between LLM vendors / providers once the agent is provisioned. The solution design become technology agnostic and also it is best way to secure your data
Read more about MCP (Model Context Protocol) at this link: https://modelcontextprotocol.io/
MCP Server in Copilot Studio
Dataverse MCP Server allows you to connect Dataverse with Microsoft Copilot studio and thereby consume data residing within Power Platform ecosystem for answering queries without a need for writing complex flows / automations.
Example of Tools that are available for Dataverse MCP Server are:
- List tables
- Describe Table
- Read Data
- Create Record
- Update Record
- List Prompts
- Execute Prompt
- List Knowledge source
- Retrieve Knowledge
