Those colorful cards, the personalized stats, the shareability, it's brilliant. But as someone who works in product, I couldn't help thinking: what if we could do this for all the data we have in Pendo?

The "What if" moment

I'm a product marketer at Pendo, and I spend a lot of time thinking about how to make Pendo’s data more engaging. We have all this rich usage data, like who's using what features, how often, and which power users are crushing it, but it usually lives in dashboards and reports. 

Then, a customer asked “what if we could pull Pendo data and turn it into a Spotify Wrapped-style experience for their team?” For me, this felt like the perfect use case for Pendo MCP.

What Pendo MCP is, and why it’s such a big deal)

Before we look at how I built Pendo Wrapped, let’s look at the tool that made it possible.

AI tools are only as smart as the context you give them. Pendo captures the richest product context in the world: how users interact with your product, what they think and feel about it, and why they behave the way they do. Without context into behavioral data, AI systems don’t truly understand your business or users.

Pendo's Model Context Protocol (MCP) server solves this by serving as a bridge that makes product data accessible to both human users and agents working autonomously.

Whether you're asking Claude a question in chat, having Cursor generate usage-informed code, or running a nightly agent that flags at-risk accounts, MCP is the universal interface for AI systems and business data.

The build

My prompt had three key pieces:

What I wanted: A celebratory "Pendo Wrapped" report that showcases a customer's year of product engagement and success

What it should include: Top active users, total engagement stats, feature adoption milestones, and growth metrics

How it should feel: Fun and celebratory, using enthusiastic language and focusing on positive, achievement-oriented metrics

With my prompt ready and Pendo MCP connected, I opened Claude and got to work.

First, I specified the subscription and application IDs I wanted to analyze and hit enter.

Claude responded with, “I'll help you create a celebratory Pendo Wrapped experience! Let me gather the data. First, I'll pull the overall account metrics for 2025…” Pause while it called Pendo MCP. “Great! Now let me get the top power users…” Another pause. “Excellent! Now let me grab the most-clicked features…”

Behind the scenes, Claude was autonomously calling Pendo MCP multiple times through natural language queries that MCP translated into the right API calls.

After a few minutes, Claude responded again: “All set! I've created your Pendo Wrapped 2025 experience. The team logged 23.3M interactions across all 365 days of 2025. I've built a complete scroll-through experience with a hero section, stats cards, power user leaderboard, top features, and achievement badges.”

Then, it presented a complete, self-contained HTML file that included:

  • A hero section with animated emoji and bold title
  • Stat cards with fun comparisons ("that's like traveling to the moon!")
  • Power user leaderboards with creative titles ("The Unstoppable Force," "The Marathon Runner")
  • Feature cards with icons and click counts
  • Achievement badges that feel like unlockables

Pulling brand styling from Figma

Next, I told Claude: "This looks good, but it needs to match our brand. Look at our Brand Design Style Guide in Figma."

Claude immediately called Figma MCP and grabbed:

  • Our signature Pendo pink (#FF4876)
  • Our cream background (#F5F2ED)
  • Inter font with the right weights (400, 600, 700, 800)
  • Card styling with pink borders and subtle shadows
  • Hover states with pink glows
  • The exact spacing and border radius we use

Then, it rebuilt the entire experience in about 30 seconds, so it looked just like Pendo.

Deploying this to a Pendo Guide

The deployment process is straightforward with Guides. Here’s what I did to move this into Pendo:

  1. Copy the complete HTML file Claude generated.
  2. Create a new Pendo Guide (lightbox style for full-screen experience).
  3. Add a code block to the guide.
  4. Paste the HTML directly into the code block.

The guide becomes a container for the HTML experience, and Pendo handles all the rendering and targeting automatically. 

How to get started 

If you're a Pendo customer and want to try this:

Step 1: Enable Pendo MCP (Note: You must be a Pendo Admin.)

Go to Settings > Subscription Settings > AI Features and enable the MCP Server. Connect via OAuth in your AI tool (Claude.ai, Claude Desktop, Cursor, etc.). 

Step 2: Start simple

You don’t need to build a “Wrapped” experience on your first try. Start with something smaller:

  • A simple stat card showing this month's engagement
  • A top 10 power users list
  • A single feature adoption highlight

Step 3: Find your QA partner

Identify someone on your team who knows Pendo Guides well, like your Center of Excellence lead or a Pendo admin who understands custom code blocks. 

Step 4: Develop your prompt

Be clear about what data you need, what experience you want, and what the goal is. Tools like Prompt Cowboy can help structure this.

Step 5: Deploy to a test segment first

See what resonates. What questions does it raise? What would make it more useful? Update your prompt and regenerate.

MCP is part of a bigger shift

The best product ideas often come from customers and teammates, but we miss most of them because we’re strapped for resources, we don’t know where to start, or it’s just not a high-enough priority. 

With MCP, ideas that’ve been sitting in your backlog for months can now become full experiences in minutes.Now, your biggest bottleneck is imagination and taking the time to craft a good prompt.

And while MCP is about making existing workflows faster, you can also use it to enable entirely new workflows that were previously impossible. 

If you're a Pendo customer, start with one simple question: what data story are you not telling because it's too hard to build?

Then, connect to Pendo MCP and see what happens.

Get started with Pendo MCP today: https://support.pendo.io/hc/en-us/articles/41102236924955-Connect-to-the-Pendo-MCP-server-beta