How AI is discovering athletes that human scouts miss | Richard Felton-Thomas

Episode Summary

Executive Summary: Richard Felton Thomas argues that AI and biomechanics can make sports talent identification more equitable by replacing limited, biased, geography-dependent scouting with smartphone-based drills analyzed by computer vision. His system benchmarks athletes by age and role, expands access in remote and underserved regions, and is already being used by clubs, federations, and development programs worldwide.

Main Topics: Talent exists everywhere, but scouting access is limited (Priority: 5/5): Thomas opens by challenging the assumption that elite athletes come only from a few countries or wealthy systems, arguing that many talented young people are never seen because traditional scouting is scarce and unevenly distributed. AI-powered talent ID through smartphone drills (Priority: 5/5): He explains how AI.io's app lets athletes record standardized drills on a phone, then uses computer vision and inferred 3D analysis to extract movement metrics like speed, jump height, coordination, and symmetry. Designing a trustworthy and sport-specific scoring system (Priority: 4/5): The product was built with scouts and clubs to ensure data is comparable, reliable, and transparent, while also being tailored to what each team values most in a player. Evidence from club partnerships and athlete outcomes (Priority: 5/5): Thomas describes pilot work with Premier League clubs and a standout player who was discovered through the app, earned trials, scored for Chelsea U18s, and later signed elsewhere and represented his country. Expanding equitable access in India and beyond (Priority: 4/5): The same model is used with Reliance Foundation to identify talent across India, including kids in hard-to-reach places, and to augment scholarship scouting with data-driven preselection. Global scaling and broader applications (Priority: 4/5): The technology is expanding into multilingual, multi-cloud deployments and is being adopted by MLS Next, while the underlying movement library could also support other sports and even healthcare use cases.

Key Arguments: Traditional scouting is inherently limited because only a small number of scouts can see only a fraction of the available talent. Geography, cost, and access create major bias in who gets noticed, so many elite athletes are overlooked before they ever have a fair chance. Standardized smartphone drills can create a fairer, more consistent way to evaluate athletes anywhere in the world. Computer vision and deep learning can extract meaningful biomechanical data from ordinary video and turn it into comparable talent metrics. Scouting must be customized to the needs of each club or federation rather than using one generic score. The system is meant to augment human scouts, not replace them, by surfacing candidates for in-person evaluation. The approach can help identify talent in both highly connected cities and remote communities, broadening participation across gender, geography, and socioeconomic background.

Data Points: Premier League scout reach: about 2,000 players per year - Thomas says each Premier League scout can only see a small number of players despite millions playing. Body segments analyzed: 22 key body segments - AI Scout uses computer vision to analyze movement from uploaded phone videos. Youth age benchmark example: 13-year-olds compared to 13-year-olds - He stresses age-specific benchmarking so older athletes are not unfairly compared with younger ones. Pilot recruitment: 50 college kids - Early testing of the app in the UK included a group of college players. Scholars selected in India: 11-year-olds - Reliance Foundation scouts talent at age 11 for five-year scholarships. Scholarship duration: 5-year scholarships - Selected children in India receive education and sport support. Scale in India: tens of thousands of kids - Children trial annually through the WhatsApp/app-based process. Youth Olympics preparation: 40 kids - After testing in Senegal, forty children were identified for training ahead of the Youth Olympics. MLS Next usage: 45,000 kids - In the U.S., MLS Next participants use the app three times per year. MLS Next frequency: 3 times per year - Players are assessed preseason, midseason, and postseason to track change over time.

Pivotal Quotes: "Talent exists everywhere. It's finding the talent that can be the challenge." — Richard Felton Thomas: Core thesis of the talk, reframing the problem from talent supply to talent discovery. "What if we took all this lab protocols, all the data, all the equipment and put them into a set of standardized smartphone drills so any kid anywhere in the world could be tested fairly and equitably?" — Richard Felton Thomas: Describes the founding vision behind AI Scout. "The system didn't see him, but we did." — Richard Felton Thomas: Refers to the overlooked player discovered near Chelsea's training ground, illustrating the value of the platform.

Implications: AI-based, transparent talent identification could democratize sports access, reduce scouting bias, and help federations and clubs discover overlooked athletes. The same movement data may also extend into healthcare and cross-sport performance analysis.