This quarter, Pendo launched our very first MCP Hackathon: a challenge for users to build something creative, innovative, and fresh. The question we posed was simple: What would you make if your AI could actually understand how people use your product?

The Hackathon community has grown to over 100 members, and the event also featured a live workshop.

From across the community, Pendo’s team hand-selected three projects that showcase the very best of what’s possible with Pendo data.

Let’s say hello to (and congratulate) our winners.

🏆 1st place: Michelle Green @ Cohley | Finn, the product usage agent

Taking home the gold is Michelle Green, Director of Product at Cohley. As the head of product, she needs to know exactly what’s going on in her product (but doesn’t always have the time to comb through behavioral data).  

To fix this, Michelle built a product usage agent, Finn, to generate a morning report that tells her exactly what happened in the product the previous day. 

This agent was trained using knowledge from Pendo’s MCP server, Snowflake, Guru, Google Drive, Planhat, business logic, and SQL queries to help it understand different values in the data tables. 

With Finn, Michelle gets a clear, concise overview of activity—grouped by customer support managers—to help her answer questions, like: 

  • What should we celebrate? What were yesterday’s top activities? 
  • What should we be concerned about? 
  • Which session replays are worth watching?

Integrating all of these business systems means her agent delivers insights ready for action. Instead of spending her morning triaging data, Michelle now starts the day knowing exactly where to focus. Congrats, Michelle! 

🏅 2nd place: Lindsay Frank @ Clinical Ink | Persona Pulse, a design critique agent

In second place is Lindsay Frank, UX designer at Clinical Ink. Gathering design feedback is usually slow, subjective, and disconnected from behavioral data. To fix this, Lindsay connected Pendo MCP with Figma MCP to create Persona Pulse: a design critique workflow grounded in actual behavioral insights. 

To use Persona Pulse, Lindsay uploads a design file, and Claude synthesizes Pendo usage data (feature adoption, behavioral patterns, segment insights) with her team's existing user personas to simulate realistic feedback.

This is far more useful than the generic feedback Claude usually shares. Lindsay now has access to: 

  • High-level user feedback, its impact on user workflows, and specific concerns for each persona
  • Suggestions on how to improve the design, with data behind the “why.”
  • Critical metrics to track when rolling out a new design, like feature click rates and average time on page 

🏅 3rd place: Stephanie Tanzar @ Auto Titling Company | Customer health dashboard

Stephanie Tanzer, VP of Product at ATC, built a customer health dashboard that helps account teams answer big questions: “How healthy are my customers? Are they at risk? Are they doing the behaviors we expect?” 

To do this, Stephanie had Claude review behavioral data for every customer to understand what healthy customers do and how company size affects behavior. How often they log in, total events in the last 90 days, and how they’re using key events. 

Her Customer Health Dashboard pulls Pendo data—login frequency, feature usage, event engagement, and weekly activity trends—and calculates a health score from 0 to 100. Accounts are categorized as Healthy, At Risk, or Critical. Account managers can filter by status, see which new customers aren't adopting key features, and drill into detailed engagement metrics.

By using Pendo MCP, Stephanie can easily answer, “What actions can I take to improve customer health? How can this customer get better and healthier over time?” ATC’s account managers can also use this tool to understand, “who is at risk?” This is a super cool use case, and we love seeing customers bring Pendo data into everyday work. 

💫 Most creative: Masahiko Kashiwazaki @ Tokyo Community | Yuki’s Magic Lamp

We would be remiss if we didn’t mention Masahiko Kashiwazaki, Principal Infrastructure Engineer @ Tokyo Community, for their wildly creative MCP submission. 

With 8,000 employees relying on Box as their content platform, Masahiko had no visibility into who was using the tool effectively and who was struggling. Standard data reporting felt dry and controlling, leading to user resistance and inefficient "spray and pray" training.

To fix this, Masahiko built Yuki's Magic Lamp: a system that transforms Box usage data into personalized astrology-themed reports that encourage them want to improve. Using Pendo MCP, Masahiko aggregated feature usage data to identify each user's "Top 10 Features" across three contexts:

  • Their last 30 days
  • Year-over-year comparison
  • Company-wide benchmarking

The reports assess user proficiency while embedding direct Pendo Guide URLs for immediate learning, and it’s all delivered via the fun, personalized lens of astrology reports!

By pairing this with segmentation (dividing users into four skill-based groups) and using embedded guides, Masahiko achieved a 24% guide usage, 24.5% adoption rate, and measurable behavioral changes across multiple features. Custom Collections usage grew 1.2x, and the Sidebar Toggle feature went from occasional use to a regular habit—usage jumped 3x.

Most importantly, this approach shifted IT culture toward a data-driven strategy across all business processes. Creative thinking meets hard metrics—exactly what we love to see. Congrats, Masahiko!

Show us what you can do with Pendo MCP

The hackathon was our first experiment in seeing what the community would build. Based on these submissions, we're just scratching the surface.

If you missed the hackathon but want to see what's possible with Pendo MCP, join us at Pendomonium 2026. March 24th includes a dedicated Hackathon session where you can learn from what the community built—and start building something yourself.

Congrats again to Michelle, Lindsay, and Stephanie. And thanks to everyone who joined, submitted, and pushed the boundaries of what Pendo + AI can do.

We can't wait to see what you build next! Learn more about MCP, and learn how to connect Pendo and Claude.