MCP integrations allow AI Agents to do something that was, until now, reserved to humans: use tools. They let Hugo go beyond simply answering questions from training or documentation, and instead interact with real systems, fetch live data, and take action — just like a human agent would.

Hugo is a next-generation AI support agent, fully integrated with Crisp.

While conversational at its core, Hugo goes beyond answering questions. He understands customer intent and context to proactively guide users toward resolution — just like a human support agent.

Built to autonomously handle a large share of incoming conversations, Hugo can also perform real actions through your integrations, such as checking account details, tracking orders, or issuing refunds, always within the rules you define.

By engaging proactively and providing concrete guidance, Hugo helps resolve issues faster and frees up your team to focus on complex cases.

Haven't met Hugo yet? Check out official resources

What are MCP Integrations

MCP (short for Model Context Protocol) servers can connect to nearly any service — including your own internal tools.

They expose tools that allow AI Agents to perform tasks that were, until now, only possible for human agents.

In practice, most support request don’t require any particular reasoning, they require checking.

For example, a customer might ask: “Can you check my subscription status?”

A human agent would open their back-office, look up the customer, and reply.

This is often a very simple task in itself, but time-consuming for both your users and your team when put in perspective.

With MCP integrations, Hugo can do now directly handle that for you and:

1. Call a tool connected to your internal system
2. Fetch the real subscription data
3. Respond with an accurate, personalized answer
4. And now be able to reason on this information if the user follows-up

No escalation needed, no time wasted, and a better experience for your customers 🚀

There is a wide range of information look-ups, tasks and actions that AI Agents can perform thanks to MCP integrations.

• Check a customer’s subscription status or plan
• Verify account information, billing details or invoices
• Check order status and shipping progress
• Retrieve tracking numbers or lookup order history
• Trigger actions like resetting access, updating stocks, performing a refund
• and as many more use-cases as you could come up with

How to connect MCP Integrations

Connecting MCP can be done relatively quickly. You will find two different ways of integrating MCP servers with Hugo:

• 1-Click Integrations → Natively-exposed tools with some of our core integrations, which can be enabled in the blink of an eye
• External MCP Servers → Allowing you to connect Hugo to your own custom-built MCP server

To get started, head over to Crisp, find AI Agent on the left, and navigate to Automate > MCP & Integrations.

1-Click Integrations

Hugo natively ships with pre-configured MCP servers for some of our popular integrations. These are ready to be connected in a heartbeat.

To connect them to Hugo, simply hit the Connect button and follow the instructions.

You can also manage them by hitting the Manage button once connected, to review the tools available with that integration, or decide whether Hugo should always verify the user's identity before using a specific tool.

Among the available native integrations, you will find the following MCP:

Shopify

• Get user → Fetches a user profile information (account details, order history...)
• Get order → Retrieves an order and its status, items, and delivery information

Prestashop

• Get user → Fetches a user profile information (account details, order history...)
• Get order → Retrieves an order and its status, items, and delivery information

WooCommerce

• Get user → Fetches a user profile information (account details, order history...)
• Get order → Retrieves an order and its status, items, and delivery information

Stripe

• Get user → Fetches a user profile information (account details, payment history...)
• Get invoice → Retrieves a user's invoice

External MCP Servers

If you are a SaaS with your own back-office, an e-commerce business with an internal database, or if your use case relies on a custom-built system that can’t be integrated natively, don’t worry.

Hugo allows you to build and connect your own MCP servers.

Already half-way there

For most teams, building an MCP server doesn’t mean starting from scratch.

You likely already have a database or a back-office your agents use in order to access data and perform actions.

In most cases, this also means you already have a REST API in place — ready to be reused within your MCP server. This is most of the work already tackled.

MCP servers act as a bridge between Hugo and the tools you already use, much like your back-office does for human agents. Their role is to expose access to your internal tools in a structured, controlled, and secure way.

In practice, this usually means:

• reusing existing REST API endpoints
• exposing them as MCP tools
• defining what the AI is allowed to read or do

Your logic is already in place, and just needs to be ported to your MCP server.

Building your own MCP server to integrate with Hugo

We recommend using dedicated MCP libraries to get started, such as @modelcontextprotocol, which handle much of the boilerplate for you.

Your server should define a set of tools, each with clear descriptions, so Hugo understands when and how to use them.

When a tool is called, your MCP server executes the underlying logic and returns the result to Hugo.

Security also extends to tool usage itself:

Once your server is configured, you can require user identity verification before any tool is executed by enabling the Require user identity verification setting on your MCP tools.

When enabled, Hugo automatically verifies the user by sending a one-time code via email or SMS before proceeding.

To help you get started, we provide a ready-to-use demo MCP server. You can use it to explore MCP capabilities or as a baseline for building your own.

Updated on: 26/01/2026