The Most Important AI Use Case in Endurance Sports Has Nothing to Do With Coaching

A runner died at mile 78 of the 2026 Cocodona 250. Randi Zuckerberg’s been thinking about what the right tracking infrastructure could have done. So have I. Here’s the thesis.

On Tuesday, May 5, a participant in the Cocodona 250 collapsed at the Groom Creek trailhead, somewhere between miles 75 and 80 of the course. She was a woman in her 40s. She died on the trail. The race organization, at her family’s request, has not released her name.

Randi Zuckerberg was on that same course. She finished, ten hours faster than her time the year before. She also knew the runner who died. When she came on Long Run Labs last week, she asked permission to bring the conversation somewhere dark, and then she said the thing I’ve been thinking about ever since:

“If we had had technology like Laurel out on the course, and people knew that she stopped moving for a certain amount of time, maybe we could have prevented that. AI might even know before we know that you’re struggling. It should be able to identify patterns of heat stroke and heart attack before we can as humans. I can’t think of a more important use for technology than that.”

I want to spend the rest of this piece explaining why she’s right.

---

Disclosure

I’m an investor in Laurel Innovations. I’m also an advisor to the company. Phil Dumontet is a good friend and the CEO. I introduced Phil to Randi who then became an investor and board member herself.

So when I tell you why I think Laurel matters, take it as the conviction of someone who has put time and money (twice now) into the thesis, not as neutral commentary. I’m writing this because the thesis deserves to be in front of more race directors, more timing companies, and more people thinking about what the next generation of race infrastructure should look like. Not because I’m pretending to be objective.

---

What’s Wrong With Current Race Tracking

The status quo in race timing has been roughly the same for 20 years. RFID chips on bibs. Physical timing mats at fixed intervals. Heavy equipment that requires a crew, a truck, and cell service to deploy. You learn where a runner is at the moment they cross a mat. You learn nothing in between.

For most road races, that’s fine. The mats are close enough together that a missing runner is detected reasonably quickly. The course is paved, well populated, and patrolled.

For trail and ultra events, that infrastructure breaks down. Mats can’t be deployed every quarter mile across 250 miles of remote desert. Cell service disappears. The gaps between checkpoints are measured in hours, not minutes. A runner who collapses between aid stations isn’t visible to the race organization, to their crew, or to medical staff until someone physically walks past them or until they fail to show up at the next checkpoint.

That’s the gap. It’s not a theoretical gap. It’s the gap that determines whether a medical emergency in remote terrain becomes a tragedy.

---

What Laurel Is Actually Building

Laurel rebuilt race timing from the software up. Bluetooth instead of RFID. Satellite-capable in remote terrain where cell service doesn’t reach. Lightweight enough that checkpoints can be placed multiple times per mile, hidden in traffic cones and inconspicuous spots along the course, without crews or trucks.

I ran a race timed by Laurel recently. The thing I noticed most was what I didn’t notice. No mats to trip on. No beeping. No equipment hauled in by a truck. The timing was invisible during the race, but obvious in the data afterward

What I did notice afterwards: multiple checkpoints per mile. Segments timed across the biggest hill and across a sponsor’s stretch of course. A QR code on my bib that pulled up my full results the moment I crossed the line. A complete and color-commentary filled summary in my inbox the next morning.

That’s what happens when you rebuild a category from the software up instead of bolting tech onto hardware that hasn’t meaningfully changed in two decades.

But the timing experience is the obvious benefit. The less obvious benefit, and the one that matters most for the thesis Randi articulated, is what becomes possible when you have data this granular.

---

The Safety Use Case

If a runner is being pinged multiple times per mile instead of every five to ten miles, the race organization knows much more, much sooner.

It knows when a runner stops moving. It knows for how long. It knows if that runner is slowing at a rate inconsistent with terrain and conditions. It knows if the slowdown is happening in a stretch where heat and elevation make medical emergencies more likely.

Layer AI on top of that data, and the system can do something humans on a crew cannot: it can detect patterns of distress before a human notices them. A runner whose pace is degrading 30% faster than the field average in a heat-affected section. A runner who has been stationary for 12 minutes in a location where there’s no obvious reason to stop. A runner whose movement signature is starting to look like someone in early heatstroke instead of someone just managing fatigue.

None of this requires medical instrumentation. It requires high-frequency location data and the software intelligence to interpret it.

The runner who died at the Groom Creek trailhead between miles 75 and 80 may or may not have been savable with better tracking infrastructure. Nobody knows. What we do know is that the current infrastructure didn’t give her crew, the race medical team, or the race organization the chance to find out in time. That’s the gap that better technology can close.

Randi said it plainly on the show: she can’t think of a more important use for AI in this sport. Neither can I.

The thesis isn’t just theoretical. At the 2025 Mesa Marathon, Laurel’s tracking infrastructure was deployed end-to-end. Lieutenant Ryan Stokes, who led traffic operations for Mesa Police Department at the event, said it directly: “Laurel was vital to safety at the Mesa Marathon, providing real-time tracking of the final participants’ locations for effective monitoring and swift response. This level of visibility was crucial in ensuring a secure and well-managed event.”

That’s a public safety official, on the record, after the event. The use case isn’t speculative. It’s already working at scale.

---

Why The Broadcast Use Case Follows

The same infrastructure that surfaces safety signals also unlocks what major marathons and elite trail races have been waiting for: real broadcast coverage that doesn’t require helicopters or motorcycle camera crews.

Imagine watching a major marathon where the booth has reliable splits every quarter mile for the entire elite field. Where the leaderboard updates continuously instead of every 5K. Where the commentator can say with confidence “Jess McClain has lost six seconds in the last 800 meters (JK, that’d never happen!), and here’s how that compares to her pattern at the same point last year.” That requires data the current timing stack can’t produce, and Laurel’s architecture can.

The same goes for sponsor activations. If a brand is sponsoring a specific stretch of course, they want to know who ran fastest through their stretch, who slowed down, who walked, who finished strong. Today that data either doesn’t exist or requires a separate timing setup. With Laurel, it’s already in the system.

---

What Race Directors Should Do With This

If you’re running a trail or ultra event right now, you are exposed. The current timing standard is inadequate for the geography you race in. Your medical infrastructure depends on someone noticing a problem in time. The next medical emergency on your course could be the one where your community asks why your tracking infrastructure didn’t see it coming.

You don’t have to wait for that conversation. You can adopt better infrastructure now.

If that’s of interest, Phil Dumontet is the CEO of Laurel. Tell him I sent you.

---

What Investors Should Know

Laurel is at the intersection of three things that almost never overlap: a real product working at real races today, a category that hasn’t seen meaningful technology investment in 20 years, and a moment where the industry has just been forced to confront what the existing infrastructure can’t do.

The TAM conversation usually starts with race timing as a category and ends there. That undersells the company. The architecture Laurel has built is the foundation for safety, broadcast, sponsor measurement, and athlete data, in a sport where mass participation continues to grow and where every adjacent layer is undersolved.

Randi Zuckerberg has invested. I’ve invested. We didn’t invest because we wanted a piece of the race timing market. We invested because we believe this is the architecture the next decade of endurance sport will be built on.

---

The Episode

Randi and I covered this and a lot more on Long Run Labs recently. She talked about how she runs a 250-mile race like a startup, what she looks for when she invests in entrepreneurs, why brand activations like Mount to Coast’s Double Boston actually work, and the version of herself she’d want the 2010 Randi to meet.

The Laurel conversation runs about ten minutes and starts around the 17-minute mark.

Listen on Spotify, Apple Podcasts, or wherever you listen.

---

Jon Levitt is the founder of Long Run Labs, an endurance and outdoor podcast network, the host of For The Long Run, and an investor and advisor to Laurel Innovations since 2024. This newsletter covers the business of creator and athlete partnerships, sponsorship strategy, and what the data actually shows.