Two days and four tracks. First day is about the technology, computer vision and the AI infrastructure behind it. Second day is about what the industry does with it, how it is commercialized and how it is governed.
Computer vision has left the laboratory. At the 2026 World Cup the semi-automated offside system runs on about a dozen cameras per stadium, tracking 29 points on every player fifty times a second, and the same technology now feeds broadcast, officiating and performance analysis. In research, markerless motion analysis has reached the point where the question is no longer whether it is accurate enough but which applications justify it and how it moves from a capture setup onto a training ground.
The track looks at that gap: what a club can run week to week from broadcast feeds or a few fixed cameras, where the data is reliable enough to act on, and what the next models make possible.
Every club now holds tracking, event, physical and medical data from several providers, and the recurring finding is that it sits in systems never designed to talk to each other. That architecture, rather than any single model, decides whether AI is useful to a coach on a Tuesday. Language models have started to enter the workflow too: recent work from Uppsala University and Twelve Football turns expected-goals models into plain-language explanations for coaching staff.
The track is about what a club has to build first: who owns the data, where it lives, how it moves, and what a language model can and cannot be trusted to do inside a performance department today.
The money in sports technology is real and uneven. SportsTechX counted about 12.5 billion dollars in disclosed deal flow through October 2025, the second biggest year on record, and 28 new funds with 9.5 billion dollars earmarked for sport. Most of it went to fan and media businesses; the share for athlete and performance solutions fell to 18 percent against a five-year average of 31 percent.
For a company taking research-grade AI into clubs and federations, the path to a paying customer is harder than the headlines suggest. The track puts founders, buyers and investors in one conversation about that path: what gets procured, why pilots stall, and what investors still active in performance and data are looking for.
The EU AI Act applies a risk-based classification that reaches biometric monitoring and automated decisions about people, with human oversight and documentation obligations. Player unions have issued their own data protection guidelines, and a 2025 paper in Frontiers in Sports and Active Living describes a gap in which athlete data is at once personal, professional and commercial while most agreements never define who owns it.
The closing track asks whether the tools discussed over the two days are trustworthy, fair and interoperable at scale, what standards they will have to meet as regulators catch up, and who is responsible when a model gets it wrong. It is placed last as a check on the other three.
One facilitated session for the whole room, working across the four tracks on a shared problem so that the researchers, the practitioners and the commercial side leave with the same picture of what is possible now. The theme and the facilitator are published with the speaker list.
Partners get fifteen minutes on stage inside the programme, in front of a room that is small enough to follow up with everyone afterwards. The partners page explains the ways to partner.
Ten to fifteen minutes each inside the Commercialization block, in front of the buyers and investors in the room, for companies taking research-grade AI into sport.