Agentic AI’s power to change sport

For tennis fans, Roland-Garros 2025 was one for history books. It started with the very emotional farewell to Rafa, the King of Clay and an Infosys brand ambassador, ending in epic fashion as Carlos Alcaraz and Jannik Sinner battled through the longest match in tournament history, a five-set marathon that redefined endurance and brilliance. But, as the dust settles on another riveting edition of Roland-Garros, a deeper transformation is unfolding, not just in how the game is played, but in how it’s captured, narrated, and consumed. That transformation is being driven by agentic AI, bringing editorial judgment, contextual awareness, and real-time responsiveness to video content. As the long-standing digital innovation partner of Roland-Garros, here at Infosys we’re witnessing this evolution firsthand.

During this year’s tournament, agentic AI played a behind-the-scenes but vital role in how the event was experienced globally. From the commentary box to the production suite and immersive fan zones, intelligent systems shaped every layer of storytelling. Unlike traditional automation, which supports tasks, agentic AI makes decisions, asking “what content matters most?”, “who is it for?”, and “how should it be told?”

For instance, our AI Commentary engine generated human-like narratives for close to 70,000 points played across the tournament. The system leveraged Large Language Models and analysed more than 30 parameters per point to surface turning points, momentum shift and technical flourish. Tennis enthusiasts could choose between a neutral point of view or follow the match from the perspective of their favorite player. The AI aims to do more than simply describe the action, it looks to understand it, drawing connections across rounds of the tournament and tournament history, and then articulating the commentary.

Similarly, AI-generated highlight reels are changing the game for content teams. Traditionally labour-intensive and subjective, now, AI can scan hours of footage and select moments based on crowd reactions, player gestures, ball speed, and more. This capability was used at the Australian Open earlier in the year by the media team to pick and choose their AI Shot of the Day. These moments are focused on generating stories and narratives for media editors to pick and even tailor for various social media channels in appropriate formats and sizes.

This shift in agency, from human editors alone to human+AI teams, has wider implications. AI-driven infrastructure is starting to support broadcasters and content platforms in making editorial decisions on the fly. That means content can now respond to audience behaviour in real time, adapting how it’s packaged depending on viewer interest, device, language, or location. A single match might yield countless versions of the story, each tuned for resonance in a different corner of the world.

It’s not just about a game. It’s changing the game for business too. 

The play of agentic AI, and how it was leveraged at Roland-Garros, signals a vital capability that isn’t limited to tennis—or even sports. This same infrastructure has profound implications for any industrial or business application where video plays a central role. Think footage of shop floors or in-store traffic. The ability to automatically generate personalised, real-time commentary to then enable further decisions and action that’s near real time could unlock a completely new opportunity. 

As we look to the future, it’s clear that sports media, and content at-large will increasingly depend on embedded intelligent systems. Agentic AI won’t replace human intuition, rather it will expand its reach, offering tools that are as responsive as they are powerful.

At Roland-Garros, agentic AI didn’t just document the tournament. It decided what mattered, how to say it, and who needed to hear it. That shift from seeing AI as a tool to treating it as an ally isn’t just transforming sport, but redefining content and storytelling.

Sumit Virmani is global chief marketing officer at Infosys