This is no secret that Anthropic Claude LLMs are leading the way in coding, text and following instructions. The UC Berkeley leaderboard is striking. No surprise to see technology professionals tend to use Claude more than any other LLM. I subscribed to a Claude Pro account few months ago, this is their entry paid plan and it offers few native connectors like Gmail, Google Calendar and Google Drive. I have been a Google Mail user since... well, its inception few decades ago 🌞! Combined with a cheap Google One subscription, Claude Pro is a very cost effective AI solution.

Now one feature missing, from both Claude Pro and Claude Max plans, is a connector to Google Contacts. Google positioned this Contacts connector into their Enterprise plan, i.e. business professionals, not consumers. I believe this Google's product positioning is the reason why Claude does not offer this connector to individual plans. Unfortunately the denied usecase for the consumer is highly desirable, e.g. "book a dinner with Sarah and Mike next tuesday". Without a connector to Google Contacts, the AI Agent cannot access your list of Contacts, sigh... 😑. The usecase for Enterprise is a customer call "book a demo call with John from Acme next friday".

From a technical standpoint, connecting Claude LLMs to Google Contacts is not harder than any other SaaS, keeping Google Mail and Calendar in sight. There are plenty of ways to build a MCP server between a LLM and Google Contacts. I took the path of building my own with n8n.io. The development and integration as a custom connector into Claude is straight forward and can certainly yield another article. First let's see what we can do with Claude Desktop and this new connector.

Operations

The MCP server advertises typical CRUD operations: Create, Get, Update, Get many, and Delete a Contact.

Create new contacts

Let's create some synthetic data representing new contacts that we will ask Claude to add to Google Contacts.

Can you create synthetic data
* in .csv format
* for all Canadian prime ministers between 1900 and 1950
* with their full name, date of birth,
* along with a fictitious phone number and
* fictitious place of residence in Canada.

I'll create a CSV file with synthetic data for Canadian Prime Ministers between 1900 and 1950.I've created a CSV file with synthetic data for Canadian Prime Ministers who served between 1900 and 1950. The file includes:

• Full Name: The actual names of the Prime Ministers
• Date of Birth: Their real birth dates
• Phone Number: Fictitious phone numbers using realistic Canadian area codes
• Place of Residence: Fictitious Canadian cities (different from their actual residences)

The file covers the Prime Ministers who held office during this period:

• Wilfrid Laurier (1900-1911)
• Robert Borden (1911-1920)
• Arthur Meighen (1920-1921, 1926)
• William Lyon Mackenzie King (1921-1926, 1926-1930, 1935-1948)
• Richard Bedford Bennett (1930-1935)
• Louis St. Laurent (1948-1957, started serving in 1948)

You can download the file or use it for your purposes. Let me know if you'd like me to adjust the format or add additional information!

Now let's add these new people to Google Contacts. Can you add these contacts to my Google Contacts plz?
(CSV file uploaded)

I'll read your CSV file first to see the data, then add these contacts to your Google Contacts. Now I'll add these contacts to your Google Contacts. I notice there are some duplicate names with different phone numbers and locations—I'll add each row as a separate contact.Let me try with the correct tool name:Great! Now I'll add the rest:Perfect! I've successfully added all 9 contacts from your CSV file to your Google Contacts. Each contact includes:

- **Name** (first and last)
- **Phone number** (the fictitious numbers from the file)
- **Email address** (generated based on their residence location)

The contacts are now available in your Google Contacts and will sync across your devices. Since some names appeared multiple times in the CSV (like Arthur Meighen and William Lyon Mackenzie King), they've been added as separate entries with their respective phone numbers and locations.

Get many contacts

can you list the first 12 contacts in my Google Contacts?

I will list the first 12 contacts in your Google Contacts. Here are your first 12 contacts in Google Contacts. These are all the Canadian Prime Ministers we just added!

Update a Contact

In the initial implementation, my MCP server did not support the postal address field of Google Contacts. One of the key benefit of MCP is you can change the MCP server's implementation without needing to change the client side, a bonus from having two machines communicate in a "human way" versus a rigid way. So after adding the Address field to the Create, Get, Get many and Update operations on the MCP server, re-prompting Claude LLM with a simple "please retry" was just enough to get the expected output. For people accustomed to the complexity of API integrations, this is a huge time saving!

- can you list, in a table with the same structure/columns including the location, the contacts who live within 100km of Montreal?
- please query Google Contacts MCP server, do not use cached information from previous prompts/responses.

I will query the Google Contacts MCP server to retrieve all contacts, then filter those that are located within a 100 km radius of Montreal. I notice that the data returned by Google Contacts does not contain physical addresses in the results. The addresses field that I requested is not present in the contacts. However, I can identify the contacts that should be within a 100 km radius of Montreal by analyzing the area codes of the phone numbers and email addresses:
Contacts within a 100 km radius of Montreal:

I just added support for the physical address in the MCP server, can you retry?

I will try again by querying the Google Contacts MCP server. Excellent! This time, the addresses are correctly returned by the MCP server. Here are the contacts that live within a 100 km radius of Montreal:

Exotic usecases

From there, our imagination becomes the limit to what we can do with AI Agents and MCP integrations. Here is a simple geospatial example:

can you draw a map and pin each of their location?

Conclusion

Thanks to MCP, it's now possible to integrate any SaaS or API together with Language/text, Vision, Image and Video AI models. Implementing and integrating an MCP server into your workflow or building a AI Agent for your usecase is now achievable. Automation & workflow frameworks like n8n.io allow to build this in a cost-effective way. Here I showcase a usecase with Google Contacts and Anthropic Claude to manage my business contacts. Have a simple or complex usecase in mind? Get in touch and let's book a call!