Designing Ground Intelligence: Giving cities a real-time view of every pothole

We've all driven past the same pothole for months. Sometimes we’ve watched it grow over years, and wondered: Why hasn't the city fixed this? Do they even know it's there? That's the problem we set out to solve with Ground Intelligence.

Why potholes stick around

City governments have a laggy feedback loop for road infrastructure. A pothole gets fixed when enough citizens call it in, or when an expensive survey that runs every 2-3 years eventually catches it. Both signals are slow (biased toward the loudest neighborhoods) and the road keeps getting worse until a crew shows up.

Samsara has a vast network of commercial vehicles driving across America every day, all equipped with cameras and sensors. Ground Intelligence turns that network into the eyes on the road for city governments. When a Samsara-equipped vehicle encounters a pothole, it's logged instantly with video evidence and G-force readings. Over time, this builds trend data that cities can use to prioritize repairs by severity and frequency instead of who calls loudest.

Understanding how cities fix roads

Given that this was a 0→1 initiative, we had some meaty UX questions to answer. How do you give city officials a feedback loop that surfaces what matters most to them? How do you pack that much new data onto a map without overwhelming them? And how do you build a workflow that slots into the way road crews already operate, and makes it better?

We didn’t have the answers yet. So we started where we always do at Samsara, with our customers.

We dove deep into existing city government processes, getting on customer calls and gathering feedback. I used Claude Code and my personal second-brain system to synthesize insights from more than 20 customer discovery and sales calls. The result was a journey map of how public works departments process pothole requests, which became the foundation for our early ideation and exploration. What would have taken a week of dedicated domain research collapsed into a single day.

With a clearer understanding of the problem space, it was time to start building.

Summary of a city government’s road infrastructure defect discovery and repair process

Designing with real-world data

One of Samsara's unique advantages is the scale of real-world data flowing through our platform. Every year, we process 25+ trillion data points from vehicles and operations across the physical world, giving us the ability to design Ground Intelligence using real signals from real roads.

When the engineering team generated a dataset of pothole detections from our first trial customer, I jumped straight into the production codebase with Cursor and built a working, UX-focused prototype. It ran on real signals: over 2,000 road observations from Jackson, Tennessee, real dash cam footage, and live GraphQL queries against production endpoints.

For a designer, this was a fundamentally different kind of prototyping. Not "a thing that looks real" but the actual thing, built fast enough that I could stay in the design loop without waiting on engineering cycles.

The impact was immediate. Customers were exploring roads they recognized instead of static mocks. This led to higher-quality validation, faster stakeholder alignment, and a smoother engineering handoff. In fact, we took Ground Intelligence from the final design prototype to GA in a week and a half, a process that would have taken several weeks the old way.

Real-time ground-level infrastructure insights for every road segment

Learning in the field sometimes means patching potholes

With a working prototype in hand, we had the opportunity to go further: onsite to the City of Akron, Ohio. We accompanied road crews on ride-alongs, patched potholes, sealed cracks, and learned how they actually use GIS and 311 systems day-to-day. Most importantly, we put the prototype directly in front of them and gathered feedback from the people doing the work.

Every customer conversation, ride-along, and pothole we patched gave us new ideas for where Ground Intelligence could go next.

This was one of the most impactful design challenges I’ve got to work on at Samsara. Working on such an open-ended problem with a close-knit, high-agency team was incredibly rewarding. We also had the opportunity to experiment with the latest AI tools and workflows, allowing us to dive deep into customer processes and build high-quality experiences quickly and iteratively.

Ground Intelligence recently made its main stage debut at Go Beyond Public Sector, our public-sector customer conference, in front of thousands of customers. I'm so grateful to have had the opportunity to drive its design from inception to that moment with such a great team!

Interested in building at Samsara? Join the team.