The World Series of Poker returned to ESPN on Thursday for the first time since 2021. And Peyton Manning’s Ohama Productions, newly in charge of the event’s TV coverage, arrives with at least one trick up its sleeve: artificial intelligence.

Omaha has enlisted an AI model that can identify when players are likely bluffing—and when they might have the nuts hand. The tool, developed by independent AI engineer Luke Geel, analyzes past results and players’ activity—from posture to blink rate—spotting connections that might be imperceptible to the average viewer but visible to top pros.

Omaha will be cautious with how it uses the feature on the broadcast, focusing only on players after they’ve been eliminated. As Omaha expands its live production portfolio, the company aims to marry character-driven and analytically minded coverage. 

“We’re trying to bring back some of what was great about the World Series during the [Chris] Moneymaker boom and some of what’s been great about poker in the last 5-to-10 years,” Omaha head of content Dan Gati said. “A lot of people think that because poker is not on mainstream television that it’s not doing as well. Live poker is as robust as it’s ever been. And I think the next step is bringing it back to ESPN for the mainstream to see it again.”

The WSOP main event, with a $10,000 buy-in and a potential winner’s prize of $10 million, began on the ESPN app Thursday. Final table coverage will air on ESPN during the first week of August.

Geel spent roughly six months developing his technology. “It was significantly more difficult than I had initially hoped,” he said. “I can’t just, like, upload a YouTube URL and say, ‘find their tells.’” After posting about his work online, Geel said multiple players inquired about leveraging it themselves, either to study upcoming opponents or identify their own potential giveaways. 

“Some people have asked, like, ‘Hey, why don’t you run this on a top professional poker player, like Daniel Negreanu?’” Geel said. “He’s not going to have any tells. He’s probably worked hard to remove them. I just don’t see any reason to do that yet. Maybe in the future, when AI becomes even better.”

With more advanced cameras and larger data sets, involuntary behavior such as heart rate and skin flushness could also be tracked. And Geel sees no reason computer vision-based prediction models can’t find other use cases, from car dealers studying what features get shoppers excited to World Cup goalies trying to guess which way a penalty kick taker might aim. 

Already, researchers have built prediction models that outperform goalies at identifying the direction of penalty kicks. David Freire-Obregón, an associate professor in Spain who has studied that problem, said that future models will help goalkeepers identify potential tells in shooters’ run-ups while also identifying individual players’ specific patterns. His team is currently studying just how far in advance of the boot meeting the ball that a direction might be tipped. 

“That said, I don’t think AI will ever make penalty shootouts predictable,” Freire-Obregón added. As with tipping pitches in baseball, evolved surveillance technology will instead force competitors to better mask their intentions—or use their tendencies against an opponent by switching things up in a key moment. 

Within the poker community, artificial intelligence has already changed the way pros prepare; it has also triggered countless conspiracy theories around online competitors potentially aided by bots. Now, viewers will get a glimpse at just what today’s computers can see.