7 AI-powered features you’ll find on Prime Video's ‘Thursday Night Football‘ this season

“Prime Vision with Next Gen Stats” has reinvented its passing-game illustrator into a new offering called Prime Targets. This feature uses a machine learning model to highlight which receivers have created separation from defenders within their routes, and are in position to convert a first down. This all unfolds on the screen in real time, allowing fans at home to read a developing play just like a quarterback.

Meet the on-air lineup for Prime Video's 'Thursday Night Football,' from Charissa Thompson to Richard Sherman

Meet the broadcast team guiding fans through all the ‘Thursday Night Football’ action, including legendary announcers, renowned reporters, and more.

Read more

The 4th Down Decision Guide allows fans to place themselves in the shoes of NFL coaches when it comes time to make pivotal decisions. Should a team go for it on fourth down, or punt? The feature uses machine learning to not only calculate the likelihood of success for both options, but also how the decision will affect the team’s game-winning probability. What would you do?

“Prime Vision with Next Gen Stats’” new Defensive Alerts feature employs a neural network driven by generative AI that grows stronger and more accurate as more and more plays unfold. It tracks the movements of defensive players before the snap, reads their acceleration, and identifies “players of interest” that are likely to rush the quarterback. When a potential blitzer is identified, a red circle appears under them to give viewers a heads up. The statistic came out of a unique collaboration between TNF producers, engineers, and Amazon’s computer-vision machine learning team.

“With Defensive Alerts, we've been able to develop something that’s as good, if not better, than NFL offensive linemen and quarterbacks when it comes to identifying those players of interest,” says Sam Schwartzstein, TNF analytics expert at Prime Video.

AI and machine learning also provide in-game options for viewers to tailor their experience. Key Plays is a new feature that allows fans to view in-game highlights and key moments on demand while watching TNF on Prime Video. A bank of pivotal moments builds as the game progresses, and offers viewers immediate access to a full rundown of the action.

Rapid Recap assists those who join the game in-progress and want to catch up on the action they missed before entering the livestream. Accessible after selecting “Watch,” Rapid Recap employs AI to compile up to 13 highlights, with the full recap lasting no longer than two minutes. While viewing the rundown, viewers have the ability to either skip or rewatch big plays within the queue and jump to live action at any time.

Prime Video customers are already familiar with X-Ray, an exclusive feature that allows viewers to explore layers of information, such as actor bios, photos, filmographies, and character backstories. While watching TNF, fans can use X-Ray to track advanced Next Gen Stats powered by AWS, watch on-demand replays enhanced with player-performance metrics, get team and player info, shop official NFL merch, and even predict the outcome of the game with Fan Polls.

Alex Strand, coordinating producer for live sports at Prime Video, says that “Prime Vision with Next Gen Stats” is an example of the advantage that Amazon holds as a technology company that presents NFL games.

“As live game producers, most of us come from traditional networks. Now we are at Amazon, and able to engage in-house AI specialists and ask, ‘What storytelling opportunities can we create with machine learning?’” he says. “It’s an amazing opportunity for us at Prime Video that’s opening up endless possibilities. This is just the beginning.”

Pressure Probability, a new Next Gen Stat that debuted Week 3 of TNF, captures the pressure on a quarterback and how it evolves after the snap. 13 different metrics can be derived from an individual’s pressure probability estimates, offering new insights that were previously inaccessible. The statistic was inspired by submissions from the NFL Big Data Bowl—a collaboration between the NFL and AWS designed to spur innovation in football. Some 250 teams of engineers competed to develop machine learning models.