Kaizen 265 - Working with Zoho CRM Using Zoho MCP - Part I
Over the past few weeks, we have been exploring the zoho-crm skill and how it can help you work with your Zoho CRM account using natural language. We started with an introduction to how the skill works and how to use it, and then explored different Zoho CRM features one by one, looking at how the skill can help you work with each of them.
While the Zoho CRM Skill makes it easier to work with CRM, there is another way to work with your CRM account : Zoho MCP.
The Skill and MCP play different roles when you work with an AI client. The Skill provides the context and guidance that helps the AI understand a CRM requirement and determine how it should approach the task. MCP provides a standardized way for the AI client to access and invoke the tools that can interact with your Zoho CRM account. In other words, the Skill can help with understanding and planning the task, while MCP makes the relevant capabilities available for the client to act on that task. They can also be used together, depending on what you are trying to accomplish.
In this week's Kaizen, we will look at how you can configure a Zoho MCP server, connect it to an MCP client, and use it to perform operations in Zoho CRM. We will start with simple record operations and gradually move towards metadata, customization, and automation.
What is MCP?
Model Context Protocol (MCP) is an open protocol that provides a standardized way for applications to interact with Large Language Models (LLMs). In simple terms, MCP allows an AI agent to interact with external applications and services through a defined set of tools.
Instead of manually performing every operation in an application, you can give an AI agent an instruction in natural language. The agent can understand the context of the request, identify the tools it needs, and use those tools to perform the required operations.
For example, instead of manually navigating through CRM to find a record and update it, you could ask an AI agent to find the relevant record and make the required change.
The important thing to understand is that MCP itself is not the AI. MCP provides the standard way for an AI agent to access and use tools exposed by an application.
What is Zoho MCP?
Zoho MCP is a service that lets you create and configure your own MCP servers. You can configure your server with MCP tools from Zoho applications and supported third-party services. These tools define the actions that the connected MCP client can perform.
For example, if you configure your Zoho MCP server with Zoho CRM tools for creating and retrieving records, your MCP client can use those tools to perform those operations based on your instructions.
Zoho MCP also provides pre-configured servers for specific use cases. These servers come with a set of relevant tools already configured and can be further customized by adding or removing tools as required.
This gives you a simple model:
You → MCP Client → Zoho MCP Server → MCP Tools → Zoho CRM
The MCP client interprets your request, while the tools configured in the Zoho MCP server determine what actions can actually be performed.
Configuring Zoho MCP
Setting up Zoho MCP involves three main steps:
- Create or select a Zoho MCP server.
- Configure the MCP tools required for your use case.
- Connect the server to your preferred MCP client and authorize access.
You can create a custom MCP server from the Zoho MCP Console or use one of the available pre-configured servers.
Once the server is created, you can add the tools required for your workflow. Zoho MCP supports tools from native Zoho services as well as supported third-party services. For this week's Kaizen, we will focus primarily on Zoho CRM.
One important recommendation is to avoid adding an unnecessarily large number of tools. We recommend keeping the count to 300 or fewer. Keeping it limited to the tools actually required for your workflow helps the MCP client identify and use the appropriate tools more effectively.
Connecting the server
After configuring the tools, you can connect the MCP server to an MCP client such as ChatGPT, Claude, Cursor, Windsurf, or VS Code. The exact steps depend on the client you are using. Zoho MCP provides the required connection details and configuration options from the MCP Console.
For example, when connecting to a supported client, you provide the MCP Server URL and complete the required authorization.
Note: Treat your MCP Server URL like a password. The URL contains the information required to access the MCP server and invoke the tools configured on it. In the videos in this post, we use VS Code with GitHub Copilot in Agent mode as the MCP client.
The following video walks you through configuring the Zoho MCP server and connecting it to an MCP client.
Once the authentication is complete, we are ready to start working with Zoho CRM.
Working with Zoho CRM
One of the biggest advantages of configuring MCP is that you can start with simple operations and gradually build more complex workflows.
We will look at this through a series of short videos.
Working with CRM Records
Let us start with some of the most common CRM operations: working with records.
In this example, we ask the MCP client to create a Contact, but only if a Contact with the same email address doesn't already exist. We then link the new Contact to an Account and fetch its details.
Points to note:
- One request can combine several operations. The duplicate check and the create happen from a single instruction. You describe the outcome, and the client decides which tools to use.
- A missing tool shows up as a clear message. When we ask the client to link the Contact to an Account, it tells us no update tool is configured, instead of working around it. Adding the tool to the server fixes it, with no other changes.
- The client asks before it changes data. Before running a tool such as create, it asks for your permission.
Exploring CRM Metadata
Once we know how to work with CRM records, we can also use MCP to understand the configuration behind those records.
In this example, we ask the MCP client to retrieve the metadata of the Deals module. The response provides details about the fields in the module, including their names, data types, and properties such as whether a field is mandatory or unique.
This can be very useful when working with an existing CRM setup. Instead of manually going through the CRM configuration to understand how a module or automation is set up, you can ask the MCP client to retrieve the relevant information and present it in a way that is easier to understand.
Conclusion
We have only scratched the surface of what you can do with Zoho MCP. In the next part, we will explore more use cases and see how we can use different tools to handle more complex CRM requirements.
If you have a specific implementation you would like help with, or something you would like us to cover, leave a comment below or email us at support@zohocrm.com. We would be happy to help you.
Previous Kaizen: Using zoho-crm Skill to Write Functions | Kaizen Collection: Home