How to Use Make.com AI Web Search and MCP Features
The latest updates to make.com introduce a new AI web search tool plus major improvements to the MCP client and MCP server, helping you build safer and smarter automations with external tools and live web data.
This step-by-step guide shows you how to use the new AI web search feature, work with the MCP client in the Make AI agent, and update your MCP server setup so you can get the most from these changes.
Overview of the New Make.com AI Web Search
The AI web search feature in make.com lets your AI agent browse the public web to enrich answers with current information. It follows strict safety rules to avoid private or sensitive data and is disabled by default for existing agents.
Key points about AI web search:
- Searches only public, non-sensitive web content.
- Designed to avoid reading emails, calendars, docs, or files.
- Uses clear system rules to prevent data leaks.
- Must be explicitly enabled in your agent settings.
How to Enable AI Web Search in Make.com
For existing Make AI agents, AI web search will not be turned on automatically. Follow these steps to enable it inside make.com:
- Open your scenario.
Log in to make.com and open the scenario that uses your AI agent.
- Edit your AI agent.
Locate the Make AI agent module or configuration and open the agent settings.
- Find web search options.
Look for a setting related to AI web search or browsing the web.
- Turn web search ON.
Enable web search explicitly. New agents created in make.com may have it ON by default, but always verify this setting.
- Save your configuration.
Save the agent settings, then save and, if needed, republish your scenario.
Once enabled, the AI agent can call the AI web search tool automatically when it needs fresh information from the public web.
Safety and Privacy Rules in Make.com AI Web Search
The AI web search in make.com is designed to be safe by default. It follows clear internal rules that restrict the kind of data it can access and return.
Data That AI Web Search Can Use
The feature is optimized to read only public, non-sensitive information. Typical allowed sources include:
- Publicly accessible websites and blogs.
- Public documentation and knowledge bases.
- Non-personal, non-confidential content.
Data That AI Web Search Must Avoid
To protect you and your users, the AI web search tool inside make.com is restricted from accessing:
- Private emails or messaging content.
- Calendars, personal contacts, and internal documents.
- Stored files and any private or protected resources.
Built-in guardrails tell the AI agent not to use web search to retrieve internal or confidential data. If a request might lead to sensitive content, the agent should decline or provide a general answer instead of attempting a search.
Using the MCP Client in Make.com AI Agent
The AI agent in make.com works with a Model Context Protocol (MCP) client that lets the agent talk to external tools in a structured, auditable way. The latest update improves how these tools are discovered and invoked.
What the MCP Client Does in Make.com
The MCP client runs alongside the AI model and handles communication with MCP servers. In practical terms, this means:
- The AI agent sees a list of available tools defined by connected MCP servers.
- Each tool has a well-defined schema, making calls more reliable.
- The client passes parameters, receives results, and returns them to the AI agent.
This structure helps your make.com scenario stay predictable and debuggable, even when working with many external tools.
When the AI Agent Should Use MCP Tools
The AI agent should use MCP tools only when they are more appropriate than AI web search or a normal text response. Typical uses include:
- Reading or writing to a database via an MCP server.
- Calling an external API wrapped as an MCP tool.
- Accessing structured data or resources not available on the public web.
With the improved MCP client in make.com, the AI agent can more reliably decide when to call a tool and how to format requests.
How to Update and Use an MCP Server for Make.com
MCP servers provide the tool definitions the MCP client and AI agent use inside make.com. The latest release makes working with these servers more consistent and robust.
Steps to Prepare an MCP Server for Make.com
To make sure your MCP server works smoothly with the AI agent, follow these general steps:
- Review the MCP server specification.
Make sure your MCP server implements the current protocol expected by the MCP client.
- Define tools clearly.
Each tool should have:
- A clear name and description.
- A strict input schema (e.g., JSON fields and types).
- Well-defined outputs that the AI agent can understand.
- Handle errors gracefully.
Return clear error messages and codes if an input is invalid or a downstream service fails.
- Test locally before connecting.
Verify MCP server behavior using test clients before you rely on it in make.com.
Connecting an MCP Server to Make.com
Once your MCP server is ready, you can connect it to your AI agent configuration in make.com:
- Open your AI agent configuration.
In your scenario, access the settings of the AI agent module.
- Locate MCP or tools configuration.
Find the section that lets you specify or manage MCP servers and tools.
- Add your MCP server settings.
Provide the server URL, authentication details if required, and any other connection parameters.
- Confirm available tools.
Check that the tools exposed by the MCP server appear correctly in the AI agent or in logs.
- Run a test scenario.
Trigger the scenario and confirm that the AI agent can call the MCP tools and handle responses correctly.
Best Practices for Combining AI Web Search and MCP in Make.com
When designing AI workflows in make.com, decide when to rely on public web search and when to call MCP tools.
Use AI Web Search When
- You need up-to-date public information.
- You want general research or context from the open web.
- No MCP tool exists for the type of data you need.
Use MCP Tools When
- You need access to structured or internal systems.
- You must follow strict schemas and workflows.
- You want more control and traceability over external calls.
By following these guidelines, you can design robust AI-powered automations in make.com that use both AI web search and MCP servers effectively.
Where to Learn More About Make.com AI and MCP
You can read the full technical announcement and details about the AI web search feature, MCP client, and MCP server improvements directly on the official help article here: Make AI web search and MCP improvements.
If you need expert help optimizing complex workflows around make.com, you can also consult specialized automation and AI consultants such as Consultevo to design, audit, or scale your scenarios.
By enabling AI web search safely, configuring your MCP client correctly, and maintaining well-structured MCP servers, you can unlock a powerful and secure AI automation stack inside make.com.
Need Help With Make.com?
If you want expert help building, automating, or scaling your Make scenarios, work with ConsultEvo — certified workflow and automation specialists.