The United States Tennis Association, known as the USTA, and technology partner IBM unveiled an expanded suite of artificial intelligence tools for the 2026 US Open. The deployment merges spatial computer vision with generative agent architectures to quantify match dynamics on court while delivering customized insights directly to fans. At the center of this technology push are two core capabilities: an automated serve evaluation metric and a conversational assistant inside the official mobile application.
To power the new Serve Quality metric, specialized high-speed cameras were deployed around Arthur Ashe Stadium. Operating at roughly 50 frames per second, computer vision algorithms track more than 21 joint and spatial coordinates across eight discrete phases of each serve. The tracking pipeline isolates variables such as knee flexion, shoulder rotation, wrist acceleration and racket head speed to map the player's full kinetic chain.
These inputs are transformed into a single efficiency score ranging from 0 to 100, which visualizes mechanical execution during live play. While television broadcast teams leverage the data to enrich visual graphics during replays, coaching teams can access the metrics to study opponent tendencies. This level of automated tracking allows analysts to spot subtle mechanical breakdowns that occur as fatigue accumulates during grueling matches.
Alongside physical tracking, IBM introduced an interactive assistant titled Match Chat, built upon the IBM watsonx Orchestrate framework. The underlying agent architecture is trained on extensive professional tennis statistics, rule sets and historical player records. Within the tournament's official mobile application, fans can query the system during live encounters to receive contextual updates regarding match momentum and head-to-head performance.
Complementing these conversational outputs, the infrastructure continuously recalculates win probabilities through a Likelihood to Win index after every single point. When significant momentum swings take place, the pipeline automatically generates contextual explanations and tags Key Moments across digital channels. The project underscores how major athletic organizations increasingly treat real-time inference as an essential element of both performance operations and consumer engagement.