Fellow’s MCP Server: Bring Your Meeting Data Into Your LLMs

If you use AI tools like Claude or ChatGPT, you already know how powerful they are for brainstorming and planning, but they’ve only ever known part of the story. While they can search the internet, they’ve never had access to the nuanced context that lives inside your meetings. That changes with Fellow’s MCP Server. Your preferred AI chatbot can now tap directly into your meeting recaps, notes, and action items, and even connect that information with tools like HubSpot, Linear, Google Drive, or Gmail, to give you better answers, richer context, and insights that actually reflect how your team works.

In this post, we’ll explain what an MCP Server actually is, how Fellow’s MCP Server works, and how you can use it to make every meeting more useful; whether you’re preparing for your next sync, following up on action items, or connecting the dots between your meetings and your projects.

What is an MCP Server?

Before we dive into everything this connector can do, you might be wondering: What is an MCP Server? What does MCP stand for? Let’s break it down in non-technical terms:

MCP stands for Model Context Protocol

Let’s start simple: MCP stands for Model Context Protocol. The protocol portion means it’s a standard way for how LLMs (large language models) like Claude, ChatGPT, or Cursor talk to other tools or sources of data. If you think about how USB-C is seen as a universal port to connect two pieces of technology, like a phone to a laptop, an MCP Server is the digital equivalent USB-C for AI. Instead of each app speaking its own language and forcing you to copy and paste between them, an MCP Server acts as a bridge that allows your LLMs to access the right context directly from the source. That means Claude, for example, can look into tools like Fellow, Asana, or HubSpot and instantly find the information it needs, like the decisions from last week’s team sync or the open action items from your last project meeting. In practice, it turns your LLM of choice from a simple AI chatbot into a connected workspace; one that understands your meetings, projects, and priorities across all the tools you rely on.

How does Fellow’s MCP Server work?

Let’s dive into how Fellow’s MCP Server works. 

Fellow securely provides your meeting data and LLMs use it to answer your questions

Because Fellow records and summarizes your meetings, there’s a wealth of information and knowledge living within your Fellow workspace. And, Fellow only stores data and information from the meetings you have access to, like transcripts, notes, and action items. This means when you ask your AI chatbot a question, it sends a quick request to Fellow, Fellow searches for the relevant meetings, and then sends that information back so the LLM can build an answer. You don’t have to worry about moving files, sharing links, or copying and pasting transcripts or summaries; the two tools handle that communication for you. 

Fellow’s MCP Server makes your meeting data instantly available in your LLM

With the MCP Server connected, Fellow becomes a powerful data source your LLM can tap into. Every meeting recap, transcript, and decision that you have access to in Fellow is now something you can ask your AI chatbot about directly; like giving your it access to a living memory of your meetings. Plus, because your LLM has access to other inputs, having your meeting data side by side with those sources allows your AI chatbot to unlock insights and automations. It can pull answers from your Fellow workspace in seconds, giving you instant clarity right where you’re already working.

LLMs only grab what they need, when they need it

A big question people have is: does this mean Claude now has access to all the meeting data in my workspace? The answer is no. Claude only has access to the meetings you have access to. It looks at what you ask for, when you ask for it. So if you say, “Summarize my last 1:1 with Sarah,” Claude will access that meeting in Fellow, and generate an answer based on that specific data. Your data stays safe inside Fellow, protected by the same enterprise security standards.

What can I do with Fellow’s MCP Server?

Just like building a bridge from one place to another, an MCP Server unlocks a treasure trove of possibilities in terms of how you can use it and what you can do with it. Here are just a few things that the MCP Server enables you to do, using Claude as an example:

Use meeting context alongside other connectors

One of the most powerful things about Fellow’s MCP Server is that it doesn’t exist in isolation. Your meetings are connected to everything else you do, like your projects, your documents, and your communication. With multiple connectors enabled, Claude can pull all that context together in a single answer.

Use your AI to analyze and summarize across all your meetings

You can get Claude to analyze and summarize data from all your meetings, and build out artifacts like documentation, slideshow presentations, colour-coded pie charts, or bar graphs based on your meeting data; the sky really is the limit. Once Fellow is connected, Claude can start recognizing patterns across all of the meetings that you have access to, and you can build on them directly within the chatbot. Instead of manually reading through multiple recaps, you can ask broad questions and get summaries that reflect everything your team has been discussing. Your LLM can scan your stand-ups from the past month and highlight the themes that keep coming up, like recurring blockers, workflow issues, or repeated goals, and then turn those insights into visual reports or shareable summaries in seconds.

Prep for upcoming meetings in a fraction of the time

With Fellow’s connector, Claude can pull everything you need to prepare for upcoming meetings, instantly, and let you build off of it. Claude can combine your meeting notes with up-to-date CRM information from HubSpot to show what’s changed since your last customer call, like new deals, pipeline updates, or follow-up opportunities. Then, you can get Claude to build you a meeting agenda, generate creatives that reflect any meeting insights, or predict any objections that may come up next time based on the context of your previous call.

Make meeting notes part of your creative work

Once your meeting data is accessible inside Claude, you can use it in all sorts of creative ways. You can turn meeting discussions into new content, like blog posts, FAQs, internal summaries, or social media updates. You can even ask Claude to build short recaps or draft project briefs based on what your team talked about. Because you can generate images or apps, you can take the context of your meetings and get really creative with what you can generate, without any coding or developer experience required. For example, you could create an interactive flashcard quiz using quotes from a meeting to help boost your memory skills. This is just a fun example that shows how flexible this connector can be. When your meeting data is easy to access, it becomes fuel for creativity.

Linear

When connected to Linear, for instance, your AI chatbot can cross-reference what was said in meetings with what’s happening in your backlog. It can show you which issues should be prioritized based on recent conversations, identify blockers mentioned in stand-ups that tie to specific tickets, or compare your planned project timeline to the real progress happening in Linear.

Google Drive

Paired with Google Drive, your LLM can link discussions from your meetings to your working documents. It can point out which decisions from yesterday’s sync impact your Q4 strategy doc, or surface the project files that need updates based on your latest 1-on-1s.

Gmail

With Gmail, your AI chatbot can close the loop between meetings and communication. It can surface your most recent email threads with clients discussed in a meeting, or help you draft a personalized follow-up based on the conversation and your previous exchanges.

Together, these connections turn your AI chatbot into something far more powerful: a workspace that understands your entire workflow. With Fellow’s MCP connector, LLMs become a place where your meetings, projects, and communication all come together.

How can I enable Fellow’s MCP Server?

Now that you know what you can do with Fellow’s MCP Server, let’s discuss how you can get access to the connector for your own organization.

Admins enable MCP Server for the workspace

Before anyone can connect Fellow to an AI chatbot like Claude or ChatGPT, your workspace admin needs to turn the MCP Server on. This happens inside Workspace Settings → Security, where there’s a toggle labelled “Allow users to create MCP Server connections.” Once that setting is turned on, the admin can see and manage all active MCP Server connections for the organization. It’s a quick change, but it’s what unlocks the ability for everyone else to start linking their AI chatbot to Fellow.

Connect Fellow to your LLM of choice

After the MCP Server is enabled, connecting Fellow to Claude only takes a minute. Inside Claude, you (or your admin) add a new custom connector, just like you would connect any other app. You name it “Fellow” and use the URL https://fellow.app/mcp as the server address.

From there, you’ll see Fellow appear in Claude’s connector list. When you toggle it on for the first time, Claude will ask you to sign in with your Fellow account and confirm that you’re okay sharing your meeting data. Once you approve, you’re done; your meetings are now accessible in Claude.

You can start asking questions like “What were the key takeaways from last week’s product sync?” or “Show me my open action items from this week’s meetings,” and Claude will automatically pull that information from Fellow.

Keep your data secure and under your control

Every connection between Fellow and Claude is protected by the same security and permission controls that already exist in your workspace. Claude can only access the meetings and notes that you personally have permission to view, nothing more. Admins maintain full visibility into which users have connected and can revoke or disable connections at any time through Workspace Settings → Security. That means you get all the power of connected AI, without giving up control of your data.

Final thoughts

AI tools are at their most powerful when they understand the full context of your work. Fellow’s MCP Server connector for Claude makes that possible. By giving Claude secure access to your meeting history, decisions, and action items, you turn it from a standalone AI chatbot into a real teammate; one that can remember what was said, spot patterns across meetings, and help you prepare, plan, and follow up faster. This is just the beginning of what’s possible with MCP technology. As more apps adopt it, your LLM will be able to work across every part of your day; connecting meetings, documents, and communication into one seamless workflow. And with Fellow as your meeting memory, that connected future is already here.

FAQ

What does MCP stand for?

MCP stands for Model Context Protocol. It’s a new standard that lets AI tools like Claude or ChatGPT securely connect to other apps like Fellow, HubSpot, or Google Drive, and pull in real-time context when you ask questions.

Do I need to be technical to use Fellow’s MCP Server?

Not at all. Once your admin turns on the setting in your Fellow workspace, connecting to Claude takes less than a minute. There’s no coding or configuration required; just a quick sign-in and approval.

Is my meeting data safe when connected to Claude?

Yes. All access follows the same security and permission controls already built into your Fellow workspace.

What can I actually do once Fellow and Claude are connected?

You can ask Claude to summarize meetings, list open action items, recap past discussions, or help prep for upcoming ones. You can also combine Fellow data with other connected tools, like using HubSpot CRM info for better sales prep or linking meeting insights with Google Drive documents.

Who can turn on the MCP Server for my organization?

Only admins can enable the MCP Server at the workspace level. Once it’s turned on, everyone else in the workspace can connect their AI tools to Fellow and start using meeting data right inside Claude.

Does this replace Ask Fellow?

No, it complements it. Ask Fellow works inside your meetings to help you catch up, summarize, and follow through. The MCP Server extends that same context into the AI tools you already use every day, like Claude, so you can work with your meeting data wherever you are.