The Rise of AI in Football: Transforming Scouting and Injury Management

The future of football isn’t arriving in a white lab coat. It’s turning up on laptops, in MRI suites, and on the touchline, whispering into the ears of scouts, physios and head coaches who are desperate for any edge they can find.

Clubs have money. The margins are brutal. If a machine can move those margins by even a fraction, it gets a seat at the table.

From Arsenal blog to global data department

For Nadav Bracha, the journey into that world started not in a boardroom, but on a blog about Arsenal.

From his home, he wrote long, detailed pieces on the Gunners, mixing what he saw with what the numbers said. He highlighted players he thought big clubs should sign. The audience grew. So did his day job in tech. To keep up, he built an AI model to spit out the skeletons of his posts.

Then he stitched the human and the machine together.

Suddenly, scouts were in his inbox.

“I started to get inbox requests from professional scouts and at clubs asking me, 'How do I know about that on a player?’” Bracha recalled. “I was like, ‘I don't know any of that about the player. Like, it's just ChatGPT.’”

If that level of insight impressed professionals, he wondered, what were they actually using?

The answer: a lot of data, not a lot of structure.

Over the last decade, football has lurched from scarcity to saturation. Wyscout exploded in the early 2010s. Rivals piled in. Now, clubs are drowning in numbers from competing platforms, each with its own language, its own metrics.

“In the past 10 years, this industry has moved from complete scarcity to data overload,” Bracha said. “There are so many different data providers.”

His company, Marquee, exists to cut through that noise. Working with clubs around the world, it pulls together scattered platforms, automates what he calls “glorified spreadsheet processes,” and acts as an outsourced analytics department.

This is not, he insists, Football Manager dressed up as enterprise software. Marquee builds bespoke player profiles for paying clients, suggesting potential signings based not just on quality but on tactical fit and club style.

Clubs can ignore it. They often do. But enough are listening. A handful of Premier League sides use the platform. Barcelona and MLS outfit Chicago Fire have publicly backed it. Whether it’s unearthing the next superstar is another debate. It’s already part of the ecosystem.

When the machine spots fatigue before the player does

The reach of AI isn’t limited to scouting. It’s creeping into the medical room, too.

During FC Cincinnati’s MLS clash with Nashville SC last year, the club’s tech flagged something odd in Matt Miazga’s movement. Five minutes later, the defender asked to come off.

The system had picked up an “irregular movement pattern.” It didn’t prevent the problem. The data wasn’t live, and no one forced him to play through pain. But the machine had seen trouble brewing.

The harder question comes next: when is he truly ready to return?

That’s where Springbok Analytics steps in. Their pitch to clubs is blunt: send us your most complicated injury. Almost every time, the answer is the same — hamstrings.

Football still hasn’t cracked them. A 2020 NIH study found hamstrings make up 12 percent of all professional soccer injuries. Re-injury rates ranged from four percent to a staggering 68 percent. Most issues arrive late in halves, when fatigue bites. Everyone has a theory on rehab. No one has a definitive fix.

“We’ve got all the new technology that exists every which way, all the new ways of testing people… how much force can you produce? What does running look like? Hamstring injuries have not gone down. They've gone up,” said Matt Brown, Analytics Director at Springbok.

He believes part of the problem is how clumsy the numbers are. Muscle strength, balance, atrophy — they’re measurable, but the process is slow and messy.

“You want to scan a player at the time of injury, two months later, six months later, to track atrophy and see if you're getting the stimulus and the changes that you're going after with muscle,” Brown explained.

Springbok’s roots lie in medicine, not sport. At the University of Virginia, researchers built hyper-specific MRI tools to help treat children with cerebral palsy, generating 3D images so surgeons could calculate tendon-lengthening procedures with precision. When that worked, the tech moved into elite sport.

The NBA signed on in 2023. MLS chose Springbok for its Innovation Lab this year.

Traditional MRIs, as Brown puts it, are “thousands and thousands of slices of [two-dimensional gray images].” Doctors have to layer them and mentally assemble a 3D picture. It’s painstaking.

Springbok uses AI to do that heavy lifting.

“We can now pre-process those images using AI… we can get all the crazy MRI images and the 3D space and time and all the stuff that exists there. We process through them, create the muscle boundaries, and we can give a very finalized, beautiful 3D digital twin,” Brown said.

They don’t treat injuries. They don’t claim to prevent them. What they do is compress a week of complex image analysis into hours and hand clinicians a clear, standardized set of measurements.

“We are the support system in that we can make imaging from an MRI way more impactful and actionable,” Brown said. “We are not the ones that actually actualize it for you. We are providing you the measurements.”

A 10-second scan for a 10-year career

Down the age ladder, another company is trying to solve a different problem: what, exactly, is a 14-year-old ready for?

At Philadelphia Union’s academy, staff constantly wrestle with the same questions. Which level can a kid handle? How do you balance physical development with technical quality? How much senior football can a prodigy like Cavan Sullivan take as a teenager — and will his body survive it?

MLS clubs test everything: strength, size, projected height, peak performance windows. It’s thorough. It’s also slow and inconsistent.

Fit:Match wants to turn that into a 10-second job.

The process sounds almost too simple. A coach or parent takes four photos of a player from different angles. The phone calculates height, body mass, wingspan and a long list of other measurements. Then it goes further, projecting likely height, growth maturation and a basic picture of what full physical development might look like.

Founder Haniff Brown calls it “ChatGPT for soccer,” half joking, half serious. A process that usually eats up time and staff is boiled down to a 30-second turnaround.

Brown’s first breakthrough wasn’t in football at all. It was in fashion.

“How can we allow [a user] to upload a body profile of himself so that he doesn't have to buy four shirts and return the three that don't fit? You'll just buy one and boom,” he said.

Retail noticed. Then hospitals. Then, in 2024, a European club asked him to scan its academy players. The scope suddenly looked much bigger.

One rule guided the product.

“I was very clear from the start that it had to take no more than 15 seconds,” Brown said. “I realized that coaches don't like assessments that take too long. They want the kids going back, doing their drills. The longer and more complicated the assessment is, the less likely they are to use it.”

The club bought in. Others followed. Another hurdle emerged: human inconsistency. Two coaches could measure the same player and come back with different numbers.

“What we saw was one coach would, for the same player, measure and get one result, and from the same team, another coach would measure that same player and come up with a different result,” Brown said.

Fit:Match strips that out. Four photos. Thirty seconds. One standardized digital profile. Clubs use it. So do families.

“When parents register their children to go into an academy, they can actually upload their photos. It generates their digital twin, and then on the back end, we tell MLS all these stats on that player,” Brown explained.

It gives clubs a clearer idea of which age groups and pathways might suit. Youth football is still heavily shaped by physical size. Early developers dominate. Late bloomers fall away.

“A player who is a 14-year-old but an early developer is far different from a player who's 14 and a late developer, and now MLS can scientifically tell that, and then make better pathways for those late developers so that they don't drop out of the ecosystem,” Brown said.

Ethics, egos and the “build or buy” dilemma

All of this sounds seductive, even revolutionary. But there’s a human cost to letting machines into the room.

Projecting a teenager’s future with an app carries obvious ethical baggage. So does outsourcing scouting and analytics to external platforms. People’s jobs are at stake.

“The first step was getting people comfortable,” Brown said of Fit:Match.

Marquee learned the same lesson. It couldn’t march into clubs and declare itself the oracle. It had to sit alongside existing staff, not above them.

“It's more about them, to be fair, to kind of feel comfortable with everything that we do together. And then once we create some successful stories together, we will definitely publish it,” Bracha said.

There’s also a cold financial calculation. Clubs spend enormous sums on salaries. Building an in-house AI and data infrastructure from scratch is expensive and slow.

“From an ROI perspective, it will always be faster, quicker, righter to go to us because we've already built something, and we're investing a lot to improve it. It's your only expense,” Bracha argued. “One of the largest expenses in football clubs today is salaries. So do they want to hire more to build such a thing or just buy externally? It's like the AI’s most common question nowadays: build or buy? In this case, I think buy.”

The answer on the pitch isn’t always flattering.

Wolfsburg were early adopters, loudly promoting how AI had helped them save €1 million a year on admin and injury prevention. The team then struggled. The PR blitz around tech jarred with performances, and the backlash was swift.

The club has doubled down on its AI commitment. Sevilla, meanwhile, use IBM WatsonX to manage data. The arms race hasn’t slowed.

When ChatGPT helps pick a back five

Not every use case is wrapped in enterprise jargon. Some are almost disarmingly simple.

Coaches have admitted to playing with ChatGPT, asking it to suggest matchups and formations. Most experiments go nowhere. One, though, made headlines.

In October 2025, Seattle Reign head coach Laura Harvey revealed on the Soccerish podcast that she had asked ChatGPT a direct question: “What formation should you play to beat NWSL teams?”

For two of the then-14 sides, the answer came back: play a back five.

Harvey didn’t blindly obey. She took the idea to her staff, weighed it, and eventually rolled out a five-defender system. The Reign finished fifth — eight places higher than the previous season.

Did ChatGPT transform Seattle Reign? No. But it gave a coach an angle she was willing to test, and that idea made it onto the grass. For those who champion AI, that’s a tangible victory.

There are plenty of failures, too. Models binned. Insights ignored. Data that never survives the jump from screen to training pitch. Maybe that’s the point. This isn’t a magic wand. It’s another tool in a sport where one extra percent can decide a season.

As one coach, Fraser, put it: “We’re all looking for any advantage we can get.”

In a game this tight, who’s really going to turn that advantage down?