The snow gods: How a couple of ski bums built the internet’s best weather app

The best snow-forecasting app for skiers and snowboarders isn’t from any of the federally funded weather services. Nor from any of the big-name brands. It’s an independent app startup that leverages government data, its own AI models, and decades of alpine-life experience to offer better snow (and soon avalanche) predictions than anything else out there.

Skiers in the know follow OpenSnow and won’t bother heading to the mountains—from Alpine Meadows to Mont Blanc, Crested Butte to Killington—unless this small team of trusted weathered men tells them to. (And yes, they’re all men.) The app has made microcelebrities of its forecasters, who sift through and analyze reams of data to write “Daily Snow” reports for locations throughout the world.

“I’m F-list famous,” OpenSnow founding partner and forecaster Bryan Allegretto says with a laugh. “Not even D-list.” 

The app has proved especially vital this year, which has been one of the weirder winters on record. The US West saw very little daily snow, despite an intense storm cycle that led to one of the deadliest avalanches in history. That storm was followed by one of the fastest melts in memory, and several resorts in California are already shutting down for the season. Meanwhile, in the East, the ongoing snowfall has offered a rare gift: a deep and seemingly endless winter.. 

MIT Technology Review caught up with Allegretto, better known as BA, in the Tahoe mountains to talk about the weather, AI, avalanches, and how a little weather app became the closest thing powder-hounds have to a crystal ball: a daily dump of the freshest, most decipherable, and most micro-accurate forecasts in the biz. And how two once-broke ski bums—Allegretto and his Colorado counterpart, CEO Joel Gratz— managed to bootstrap a business and turn an email list of 37 into a cult following half a million strong. 

This interview has been edited for clarity and accuracy. 

You grew up in New Jersey. Middle of the pack as far as snowy states. What were your winters like as a kid?

I was always obsessed with weather. Especially severe weather. Nor’easters. There was the blizzard of ’89, I believe, that hit the East Coast hard—dropped two to three feet of snow, which was a lot for the Jersey Shore. My dad worked for the highway authority, so he had tools other than the evening news. He was in charge of calling out the snowplows whenever it snowed, so I just remember chasing storms with my dad. I wasn’t allowed to ride in the snowplows. I’d watch them. When I got older, I was the one shoveling the neighbors’ driveways. I just liked being out there. In it. In college, I used to go around and shovel all the girls’ sidewalks. That was fun. 

When did you start skiing?

We would cut school and take a bus to go skiing, unbeknownst to our parents. It was the ’90s, and the surfers decided snowboarding would be fun, so the local surf shop started  running a bus and all these surfers would show up and hop the bus to Hunter Mountain. We’d drive to the Poconos, go night skiing, turn around. It wasn’t uncommon for me in high school to get in the car by myself, either —and just drive. Me, my dog, my backpack. I’d sleep in gas stations and ski. Storm-chasing around the Northeast. 

What were you really chasing, you think?

Natural highs. Happiness. I’ve always been a soul-searcher. I grew up in a crazy house situation, a broken home. My dad left. My mom became a drug addict. I just wanted to be gone. I’m the oldest. I was always trying to help my mom and make sure she was okay. No one was telling me to go to school and have a career. I just wanted to do something that fulfills me.

How’d you go about figuring out what that was? 

For me, to go to school was a big task, given where I was coming out of. There wasn’t any money. I could get grants and scholarships because my mom was so poor. I wanted to go to Penn State but didn’t have the grades. I ended up at Kean, a public university in New Jersey. It had a meteorology program. We got to go to New York City, to NBC, and practiced on the green screen. In meteorology school, I started thinking: How do I work in the ski and snowboard industry and use weather at the same time? I went to Rowan [University] for business, in South Jersey, and in between moved to Hawaii to surf and spent a year teaching snowboarding. My goal the whole time was to not work in a career I hated.

I imagine you weren’t like most meteorology students. 

Us punk rockers, skaters, snowboarders—we were a little different than the typical meteorology nerds. I was the radical storm chaser. A big personality. I still am.

You didn’t quite fit the traditional weatherman mold.

Back then, there were no smartphones or social media. If you were a meteorologist, you either worked in a cubicle for the government or at an insurance company assessing weather risk.  Or you were on the local news. That wasn’t my thing. They didn’t want Grizzly Adams up there with his big beard.

Beards belong in the mountains?

Meteorologists live in cities because that’s where the jobs are. They don’t live in small mountain towns.  That’s what was missing in the industry. When I moved to Tahoe, in 2006, I realized nobody had any trust in the weather forecasts. It was more like a “We’ll believe it when we see it” old-fashioned mentality. If you’re a forecaster in flat areas, you just look at the weather model and regurgitate the news. Weathermen in Sacramento or Reno didn’t give a crap about the ski resorts! They’d just say “We’ll see three feet above 6,000 feet” and go on to the next segment. And skiers were like: “Wait a minute. Is it going to be windy at the top?” I thought: Let’s home in and give skiers what they’re looking for.

So you were living in Tahoe, skiing and forecasting?

I was working in the office at a resort, snowboarding, and doing weather on the side. I’d get up at 4 a.m. and do it before my 9 a.m. day job. Forecasting, figuring out: How the heck do these storms interact with these mountains? I started emailing everyone in the office what I’d see coming, and people kept saying “Add me! Add me!”  Eventually, resorts around Tahoe started asking to use my forecasts.

How were you actually forecasting, though? 

The NOAA, the GFS [Global Forecasting System], the Canadian model, the Euro model, German, Japanese—all these governments make these weather models to forecast the weather. And share it. Anyone can access it. But you can’t just look at a weather model and go, Yep, that’s what’s going to happen. That’s not how it works in the mountains. It’s way harder. You can’t rely on model data. It’s low-res, forecasting for a grid area that’s too big. It can’t understand what’s going on. It’s going to generalize the weather. You can try that, but you’re going to be wrong. A lot of people are going to stop listening. I was able to forecast more accurately than most people because I was living there; I could fix a lot of these errors. Around 2007, I started my own website, Tahoe Weather Discussion.

Snazzy.

Meanwhile, I heard about this guy Joel out in Boulder, Colorado. People were telling us about each other, saying: “You guys are doing the same thing!” He was sleeping on his friend’s couch, running a site called Colorado Powder Forecast. And then there was Evan [Thayer, who would later join the company], in Utah. I think his website was called Wasatch Forecast. 

Great minds!

He actually grew up outside Philly, only about an hour from me. We both were obsessed with storms and snow and moved west to the mountains and started similar websites. We would’ve been best friends as kids! Anyway, Joel called me in 2010 and was like, “Hey. I’m building this site, forecasting skiing in ski states.” And wanted me to join. He knew I had big traffic. He was like, “Let’s do it together, not against each other.” I asked, “What’s the pay?” He said, Zero. Give me your company. 

And you just said: Yeah, sounds good?

I just really trusted him. He’d asked Evan too—but Evan was like, Give you my site and my traffic for free?? No, I built this.

A normal response.

I was the knucklehead that was like, okay. Evan was still single. I already had a wife and two kids. I’d just had my son. I was working two jobs. I was so overwhelmed. So busy with my day job, as an account manager at the Ritz at North Star. Vail had just bought them and we all thought we were going to lose our jobs. My site was struggling. I was desperate for somebody to do it with. I think I thought it was a good opportunity. I was scared, though. For sure.  

That was 15 years ago. How’d OpenSnow work in the old days? 

We were just using our brains. That’s how it started: with us using our brains.Looking at all the weather models—all the data from the government models and airplanes, satellites, balloons. A million places. Building spreadsheets and fixing all the errors in the forecast models. We’d take the data and reconfigure it—appropriate it for the mountains. It was all manual for a really long time.

How manual? 

It was old-school. All the resorts had snowfall reports on their sites, and I was the one hand-keying it in: “three to six inches.” That was me on the back end, typing it in every single morning for every single ski resort. It’d take me hours. 

And then?

Around 2018, we built our own weather model to do what we were doing. We called it METEOS. It’s an acronym—I can’t even remember what it stood for!  METEOS was just us using our brains and our experience to create formulas. It automated everything and allowed us to create a grid across the whole world and forecast for any GPS point. It took all this data, ingested it, fixed some of it, and then spit out a forecast for any location. In the world. 

Were you guys making any money? 

It was crap in the beginning. Advertising-based. We stole Eric Strassburger from The Denver Post —he doubled our ad revenue in his first year full-time with us. Still, Google Ads had chopped our ad rates in half; it wasn’t a good long-term strategy to rely just on ads. We had to pivot to plan B so we didn’t go out of business. 

Subscriptions.

When all the newspapers started charging to read articles, Joel was like: We are meteorologists writing columns every day. Journalism weather is not sustainable! We need to be a weather site. We need to be a weather app. 

What happened when you moved from ads to subscriptions? 

The money took off.  We could quit our day jobs and work full time on OpenSnow. The company exploded. We were like: Are people gonna really pay for this? They did! Although they could still access the majority of the site for free. 

At the end of 2021, you put in a pay wall?

That’s when we panicked! We’re gonna lose 90% of our customers! But 10% will stay loyal and pay. Since the beginning, there’s been only two times our traffic went down: the paywall and covid. Otherwise, every year it’s gone up. People were like, Okay I can’t live without this.

I admit, I’m one of those people. So is my editor. Any other weather app is useless for skiers.

When it comes to ski towns, everyone uses OpenSnow. When the Tahoe avalanche happened, we were up early on search-and-rescue calls, helping the rescuers with forecasts. We’re now the official lead forecast providers for Ski California. Ski Utah. Head of Forecasting for National Ski Patrol. Professional Ski Instructors of America. US Collegiate Ski & Snowboard Association. Dozens of destinations and ski resorts. Joel doesn’t like to talk about it publicly, but our renewals and retention and open rates blow away the industry standards. 

I bet. OpenSnow is like a benevolent cult. 

People connect with a small company with underground roots. We’re independent. Fourteen full-time, plus seasonal. About half have meteorology backgrounds, from bachelor’s to doctoral degrees. Our very first employee was Sam Collentine,  a meteorology student in Boulder, who started as an intern in 2012 and is now our COO and does everything. 

Sounds like employees and subscribers sign on and just … stay.

Everyone stays! Our cofounder Andrew Murray, Joel’s friend and OpenSnow’s web designer, left around 2021. But yeah, people feel like they know us. They’ve been reading me in Tahoe with their coffee for 20 years! I get recognized everywhere I go. For example, I broke my binding, and went into a ski shop and asked if I could demo. And the guy was like, ARE YOU BA? Just take it! Sounds fun—until you just want to have dinner with your family, or buy a glove. Joel gets the same thing—people make Joel shrines in the slopes that look like Catholic candles.

You guys are like modern-day snow gods. Gods of snow.

People are weird.

How weird?

Someone once sent me a photo, saying: “Look, my friend dressed up as you for Halloween!” People are always inviting me over to dinner, to PlumpJack with Jonny Moseley. I guess they want to hang out with the “Who’s who of Tahoe.” There was an executive from Pixar who had me to his multimillion-dollar home on the west shore of Lake Tahoe. He had a photo of me over the fireplace in the bathroom. I thought: That’s weird, he has a photo of me over the fireplace. What was even weirder, though: It was autographed. I’ve never autographed a photo in my life! This guy just signed it—himself. I didn’t say anything. I just left.

Do you get a lot of hate mail? Mean DMs? 

Thousands. People think I can make it snow. I think they think I’m to blame when it doesn’t. The other day, someone messaged me on Instagram with a picture I’d posted over California of the high-pressure map—somebody had shared it, and wrote “Fuck Bryan Allegretto” over the high pressure.

Hilarious.

People were yelling at me during covid: You’re encouraging people to go out skiing! It wasn’t March 202o, it was January 2022. I’ve since deleted my personal social media. I never wanted to be in the spotlight. That’s the whole reason signing off my forecasts with “BA” became a thing— I didn’t want to use my full name. I just do it because it’s good for the company. Joel realized years ago that people come to us for forecasts —and forecasters. That’s why we still have forecasters. Even though AI can do what we’re doing now.

Is AI doing what you do now? 

We were using METEOS until this season. In December, we launched PEAKS. We built our own machine-learning model. The AI is taking what we were doing—and doing it everywhere, faster. The whole world instantly, in minutes. It can go back and actually ingest decades of government data—estimated weather conditions over the entire US from 1979 to 2021—and correct the errors. 

What makes it so accurate?

Before PEAKS, it wasn’t very specific. The data used to be what Joel calls “blobby”—like giant blobs, just big splotches of color over a mountain range. It’s like, if you take a pen and press into a piece of paper, the ink will spill out. The AI is like if you just tap the paper. A dot versus a blot. Now we can know how much it will snow, say, in the parking lot at Palisades and how much at the summit. It’s less blobby, more rigid and defined. 

Defined how?

All weather models output forecasts on a grid. The gridpoints are essentially averaged data over the grid box. So a model with a 25-kilometer grid resolution averages data over 25 kilometers, or around 16 miles. This is far too large an area, especially in mountainous terrains where a few miles can make a massive difference in experienced conditions. The AI is downscaling the models into smaller and smaller grid boxes. We are able to train a model to transform lower-resolution data from the same period into this high-resolution “ground truth” data. Then the model can generalize this training to global real-time downscaling. PEAKS is learning wind patterns, thermal gradients, terrain, and weather patterns and connecting all these factors to learn how to transition from coarse resolution into high, three-kilometer resolution—leading to more precise forecasts. We’ve basically taught the AI how to forecast like us. Except 50% more accurate. Now, when I wake up at 4 a.m., PEAKS has already done it.

So … then what are you doing at four in the morning?

Oh, I’ll still do the forecasting. I like to double-check it—but I don’t really need to. PEAKS has allowed me to spend more time on writing. Now instead of spending four hours forecasting and then rushing to write it,  I’ve been able to make my forecasts more interesting, more entertaining. Yeah, AI could probably write it—but I want to. It’s all about the personal connection. 

How did last year’s federal funding cuts for the NWS and NOAA affect your business? Are you guys concerned about that going forward?

We had those discussions when it first happened. In forecasting, you still need humans: to launch the weather balloon, staff the weather stations, collect the initial data. Some people in our office panicked—they had spouses or friends getting laid off. We were wondering if we’d have less data coming in, if it’d make the models less accurate. But the backlash in the weather community was swift. I think they were like, There are important things you can’t cut. It was pretty short-term. Are we worried going forward?  No, not as long as the data keeps coming in! We won’t survive without the government publishing data.

What’s next? 

We recently bought a small company called StormNet that tracks severe weather, probability of lightning, hail, tornadoes. We just launched it. Used to be like, “The storm is an hour away.” Now we can say, “In seven days there might be a tornado here.” And next winter, we’re working on a feature that can help forecast avalanches using AI. Right now, it’s still manual—people going out testing the snow layers. Forecasting is limited. This wouldn’t replace the avalanche centers, but it will be able to look at everything, including slope angle and previous weather and current conditions, and forecast further out, give people more advance—and location specific—warning. Help alert the public sooner.

Help save lives. 

I talked to one of the guys who left the Frog Lake huts on Sunday, before the storm. Before the group that was caught in the Tahoe avalanche. He told me: “People are always like, Oh, it’s never as bad as they say. But I read OpenSnow. I could tell by the language you were using, that we should get the heck out of there. I wanted no part of that.” We don’t hype storms. Or sugarcoat. Our only incentive is to be accurate.

True that it was the biggest storm in Tahoe in four decades?

In 1982, we got 118 inches over five days, and this one was 111 inches—two storms of similar size created the same level tragedy. It’s too much, too fast. It was snowing three to four inches an hour. That was the fastest we’ve seen. I don’t know what’s the bigger story—the fact that we’ve had the biggest storm in over four decades or the fact that all that snow disappeared in five days.

Do you worry about the future of OpenSnow given, you know, the future of snow?

We’ve had the second-warmest March in at least 45 years. We’re just getting these wild swings now. The seasonal snow averages are almost the same, but we’re seeing more variability than we did in the 1980s and ’90s. We’re either getting really cold and really warm, or really dry and really wet.

Bad years can affect our business, for sure.  It’s certainly affecting the industry—I know Vail, Alterra took big hits this year. Usually we’re okay, because if it’s dry in Tahoe, it’s snowing in Utah or Colorado. Our three biggest markets. I don’t recall a season where the whole, entire West was in the same boat. It’s been the worst year in the West. Yet our traffic keeps going up. Everything is up. The East Coast had a good year, Japan, BC. We’re slowly expanding in those places. It happens to be the first year in 15 years we started marketing. Marketing works!

Amazing.

Joel and I have had this repeat conversation for years—we just had it again two weeks ago: “Can you believe what we’ve done? This was never the goal.” I’m still blown away daily. We’ve never borrowed from investors. No series A, B, C. We’ve gotten offers to sell, but no. We’re still having too much fun. All I know is: Joel and I didn’t come from money. We’ve never chased money or fame, and got both. I think it’s because we never chased them. We’ve always chased the joy of skiing and forecasting powder, and doing that for other people.We were just trying to create something that made us happy.