How to Use Notion MCP in Make - Exploration (Automate Notion with AI)
Why Notion MCP - Use Cases and Value
By using the Notion MCP server, you can allow LLMs to access your Notion workspace and retrieve data or take actions on your behalf. This is particularly applicable to AI Agents, a paradigm of LLMs that are not merely conversational, but can also interact with software tools as a user.
MCP lowers the barrier of access to building automation by providing a standardized surface to instruct the AI. There is a goal—what you wish to achieve—and the steps to get to the objective are determined by the AI Agent, also using your connected MCP servers as needed.
When connecting an MCP server to any LLM that supports it (e.g., Cursor AI, ChatGPT, Claude, n8n, Make, Zapier), you can decide which tools the LLM can use. You connect your Notion workspace, and then define what actions the LLM can take within the workspace. MCP tools are defined by the developers of the MCP server. They are akin to API endpoints, each doing a specific thing (e.g., fetch content, create, update, …). They are most often based on the API specifications of the software, though sometimes there are different tools available that are optimized for AI and hence more open, less structured, and less fixed than a traditional API.
Traditional API calls are still very valid and the most widely used method of automating workflows. MCPs are a new paradigm worth exploring and understanding because it has the potential to become ubiquitous like APIs have been so far. That’s because AI is being adopted very widely at so many levels of society that it seems difficult to remain static in thinking about the world of software (and broader sectors) like we have been thinking so far. This is natural, and everything is always changing, so there is nothing transformational or incredible about it, as far as I can tell.
We make the future by creating and reinforcing constructs and narrations around those constructs, following the lead of the people we admire. Silicon Valley is one of those “people we admire”, and the narrative about AI and its ubiquities there appears established and prominent, a matter of fact rather than a mere prediction. The world seems to be following. In such cases, it is not a question of liking or not liking. It is or it is not.
Reality can often clash with our wishes, hopes, comfort, etc. Change is always happening, and it is an opportunity for renewal, because if change is always happening, everything is always new. This is a mindset shift that many of us can benefit from. There is no static, no great past or wishful future or ugly present. There is this, and then there is that. Everything is in a continuum that connects past, present, and future. This is exciting and offers great opportunities.
The good thing is that we don’t have to play every game. We do get to accept reality and the games people play, but that doesn’t mean that we don’t make up our minds on these changes and derive conclusions that apply to us personally. The ability to open up and welcome change provides the chance and spaciousness to play the game fully, play it just in the necessary amount, or also withdraw from playing the game or make up a new game. This is “strategy” in business.
How to Use Notion MCP in Make (AI Agents or Tool calls)
In Make, there are two primary ways to use MCPs:
As a tool for an AI Agent: There is a dedicated section for “MCPs” in the Make AI Agent configuration screen. There, we can connect pre-defined MCPs by selecting them from the dropdown menu, or connect our own. To connect an MCP server in Make, we need the MCP URL (https) and (depending on the requirement of the MCP) the authentication key/token. In the case of the Notion MCP, we can find the URL here, and we don’t enter any authentication token (Notion uses OAuth2). Once we save the connection in Make, we will be asked to authenticate in Notion and select the workspace we wish the MCP to access. Finally, we can select which MCP tools are allowed for use by the AI Agent. The AI Agent will determine the most suitable tools to call (if any) during each run.
As a module (like any other module, as part of a scenario) - in any Make scenario, we can use the “MCP” module to call a specific MCP server tool as part of our automation. For example, as part of a knowledge retrieval automation, we could use the MCP module to call the Notion MCP “Fetch” tool and provide a dynamic query to get answers we can then process in our Knowledge Management System.
The Notion MCP server currently offers these tools. As the documentation reads: “These tools work seamlessly together through prompts, and their real power comes from combining them”.
Search - input a query and get a list of results from your Notion workspace and connected apps.
Fetch - input a page/database ID or URL and get information about it (properties, page content, other metadata)
Create pages
Update page
Move pages
Duplicate page
Create database
Update database
Get comments
Create comment
Get users - all paid members in the connected workspace
Get user - a specific user by passing ID or email address
Get self - get your bot user information (connected workspace, file size limit, etc.)
more coming soon…
