MyFitnessPal Bought CalAI. Here's Why That's Telling.

MyFitnessPal just acquired CalAI for somewhere in the eight-to-nine-figure range, and tech YouTuber Matthew Berman says the whole thing proves that legacy software companies fundamentally misunderstand where their industry is headed. His evidence? He rebuilt CalAI's core functionality in 20 minutes.

That's the story. Now let's figure out what it means.

The Setup

CalAI is simple: point your phone at food, it estimates calories. That's it. The company apparently had $30 million in annual recurring revenue with seven employees. MyFitnessPal—the venerable calorie-tracking app from 2005—spent nearly a year negotiating to acquire them.

Berman's argument is straightforward: when anyone can spin up CalAI's functionality with an AI coding assistant in under an hour, what exactly did MyFitnessPal buy? Not technology. Maybe distribution, maybe revenue, but neither of those is durable when the ground is shifting.

"Anybody can vibe code that specific functionality. And I would bet under an hour, not even a day," Berman said in his video.

He then proceeded to prove it, using OpenClaw to build a working calorie tracker. Started at 11:18 AM, finished at 11:35 AM. Seventeen minutes. He showed it working: snap a photo of an In-N-Out burger, get back a calorie breakdown. Off by about 130 calories from the actual nutritional info, but close enough for most use cases.

The Counterargument

Jason Lemkin—SaaS investor and someone who actually knows how software businesses work—pushed back on Berman's claim. Sure, you can replicate the basic version quickly, Lemkin said, but "not the 30 million in ARR though. or community."

That's the reasonable objection. Revenue and distribution are real. CalAI had paying customers. That's worth something. Building a product is one thing; getting people to pay for it is different.

Berman's response: that ARR is "completely not durable revenue." His reasoning is that AI agents are about to collapse the entire vertical SaaS category. Why pay for a standalone calorie-tracking app when your AI assistant—the one you're already paying for and using all day—can add that functionality for free?

Here's where it gets interesting. Even CalAI's founder, Zach Yadagari, responded to Berman's "could have vibe coded this" claim with "True" and a laugh emoji. The guy who just sold his company for millions acknowledged that the core technology isn't the moat.

The Pattern

MyFitnessPal's history makes this acquisition more telling. Founded in 2005. Acquired by Under Armour in 2015 for $475 million. Then sold to a private equity firm in 2020 for $345 million—a $130 million loss in five years.

Berman's take: "Anytime that you have non-technical people, non-product people trying to run a software product, it never works out."

That's harsh but not exactly wrong. Under Armour is an athletic wear company. Private equity firms manage from spreadsheets. Neither background suggests deep product intuition for consumer software. And now they're buying an AI-powered calorie tracker from a seven-person team of young founders who, in Berman's words, "got out at the perfect time."

Maybe even "a little late."

What's Actually Happening Here

The question isn't whether Berman can replicate CalAI in 20 minutes. He did. The question is whether that matters.

If you're CalAI's founders, it doesn't matter at all. They built something people wanted, got distribution, generated revenue, and sold before the music stopped. That's a win by any measure.

If you're MyFitnessPal, it might matter a lot. You just spent significant money on technology that's increasingly commoditized, betting that you can integrate it into your existing user base before AI agents eat your entire category.

Berman's broader argument—that "the entire software market is collapsing down into agents"—is either prescient or premature. Probably both. This exact pattern has played out before. Remember when every company needed a mobile strategy? Or when everyone was pivoting to the cloud? Technologies that seem like features today were products yesterday.

The difference this time might be velocity. If AI agents can replicate app functionality in minutes instead of months, the window for defensibility gets narrower. Distribution still matters—getting users is always hard. But when your AI assistant already knows your health goals, dietary restrictions, and grocery list, why would you download a separate app to track calories?

"When my agent can have essentially any functionality that I want, why am I paying all of these external services?" Berman asks.

It's a reasonable question without a clear answer yet. We're in the territory of what-ifs and maybes. Berman isn't wrong about what's possible. The question is what actually happens.

The Missing Pieces

What this analysis doesn't capture: consumer behavior is sticky. People who already use CalAI might keep using it even if alternatives exist. Switching costs aren't always about technology—sometimes they're just about not wanting to think about it.

Also missing: the integration challenges. Berman built a working prototype in 20 minutes. That's not the same as a production app with reliability, security, customer support, and all the boring infrastructure that makes software actually work at scale.

And here's what I keep circling back to: if the core functionality is so easy to replicate, why did MyFitnessPal take almost a year to negotiate this deal? Either they saw something Berman doesn't, or they moved too slowly to recognize what Berman sees clearly.

Berman says he'll check back in 18 months to see how CalAI is doing under MyFitnessPal's management. That seems about right. The thesis is testable. If AI agents do collapse vertical SaaS, we'll see it in the revenue numbers. If they don't, we'll see that too.

For now, we have a data point: a seven-person startup sold for millions, a 20-year-old app bought them, and someone rebuilt the core product in 17 minutes. Draw your own conclusions about what that means for the next year of software.

—Mike Sullivan, Technology Correspondent