ENGLISH TEXTS
G. Live betting has just been greenlit: are the races ready?
D. D. There’ s still a delay of a few seconds between the race and the broadcast. We’ re working to reduce it, but full live betting throughout the race- like in football-isn’ t on the table yet. Initially, the idea is more about engaging the bettor with questions like:“ How many races will Éric Raffin win today?” or“ Will he win the next one?” Kind of like in tennis:“ Who will win the next set?”
G. Is the app going to evolve a lot?
D. D. It’ s been smart-connected for two years now. I’ d like it to be personalized- so a racehorse owner, upon launching the app, would get:“ Good morning, David, you have five runners today. Here they are.” We’ re testing this AI integration now. The experience could even be extended to PMU bars, where they could showcase their offerings.
G. Are you often approached by institutions?
D. D. Yes. Beyond the project with France Galop, Mr. Barjon( Le Trot) pushes us hard on video innovation. He wants to tell stories. The new Le Trot website is very data-driven. We’ re working on integrating data streams to offer an expert mode, like Formula 1 coverage on Canal +. The user could choose their viewing angle: drone footage, steward’ s cam... The Tour de France is a huge source of inspiration. Even if a race lasts just three minutes, being able to present infographics, tracking, and real-time data like that would be incredible. We want to enrich the customer experience.
G. What about the broadcasts themselves?
D. D. We currently stream 39 live feeds daily, and up to 20 more on weekends for PMH races. We’ re continuing to expand our coverage to small provincial racetracks. The goal is to broadcast all PMH meetings.
� PAGE 66
AT PMU
BACK TO THE ARTICLE �
By Serge Okey
AI MARKS A“ REAL CHANGE IN DI- RECTION”
We don’ t often realize just how much the historic operator of horse racing, PMU, is among the digital leaders in France. Yet it’ s hardly surprising when you handle billions of data points like they do. Four years ago, PMU made the shift to AI and plunged into a new business model.
Approaching its centenary in 2030, PMU likes to describe itself as“ an old but modern company.” The numbers are dizzying: 2 billion bets per year, 500,000 per hour, several hundred per second. On big race days, traffic peaks rival those of e-commerce giants on Black Friday. Internally,“ the first machine learning algorithms date back to 2018. Advanced data processing has been in production since then. Generative AI( autonomous content creation technology) has been in use for three years. We’ ve already filed two patents: one aimed at improving a computer’ s ability to identify objects in images, and another to speed up the performance of advanced AI systems, like those used for ChatGPT,” explain Anthony Hyvert, Chief Technology Officer, and Nicolas Ortiz, Chief Data Officer. All of this has led to“ a real change in direction over the past year and a half, with around ten innovative projects, both internally and aimed at our customers, partners, and bettors.” The two specialists emphasize this point:“ We’ re not working on projects that replace people, but ones that make life easier, that support the customer. AI is an assistant in a‘ Human in the loop’ process— where humans maintain a central role.” Looking ahead, information provided to bettors promises to be increasingly personalized. Internally, generative AI is already used daily by half of PMU’ s 1,100 employees. An entirely in-house version of ChatGPT continues to grow.“ We’ ve integrated all internal company documents
152