Get Your Weekly Digest

Receive the best podcast summaries and insights delivered to your inbox every week.

We respect your privacy. Unsubscribe anytime.

TED Talks Daily

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

with Richard Felton-Thomas

Published November 5, 2025
View Show Notes

About This Episode

Sports scientist Richard Felton-Thomas explains how his team is using AI, computer vision, and biomechanics to make youth sports scouting more equitable and data-driven. He describes the AI Scout smartphone app, built with clubs like Chelsea and Burnley FC, which analyzes standardized movement drills to identify talent regardless of geography or background. Through examples from the UK, India, and Senegal, he shows how the technology is uncovering overlooked athletes and scaling across sports and regions.

Topics Covered

Artificial intelligence Data-driven decision-making Entrepreneurship Pattern recognition Smartphones and mobile technology Sports talent identification Biomechanics Computer vision Youth sports development Sports equity and access

Disclaimer: We provide independent summaries of podcasts and are not affiliated with or endorsed in any way by any podcast or creator. All podcast names and content are the property of their respective owners. The views and opinions expressed within the podcasts belong solely to the original hosts and guests and do not reflect the views or positions of Summapod.

Quick Takeaways

  • Traditional scouting is constrained by human bias, geography, and limited capacity, leaving many talented athletes unseen.
  • AI Scout uses smartphone-recorded movement drills and computer vision to analyze 22 body segments and infer 3D motion from 2D video.
  • The system tailors its scoring algorithms to each club or federation, aligning with what specific scouts value in players.
  • Age- and gender-specific benchmarks allow fair comparisons among peers rather than across vastly different physical maturities.
  • The technology has already surfaced overlooked talent near elite clubs and in remote or underserved regions like rural India and Senegal.
  • Partnerships with organizations like Chelsea FC, Burnley FC, Reliance Foundation, and the IOC help integrate AI into existing scouting pipelines.
  • In India, tens of thousands of children now trial annually via the app, with some earning multi-year sport-and-education scholarships.
  • A single shared community phone was enough for one previously unorganized player to secure a five-year scholarship.
  • MLS Next uses the app three times per year with tens of thousands of players to track performance changes over time.
  • The same movement primitives underpinning the system can translate across multiple sports and potentially into healthcare and medical applications.

Podcast Notes

Podcast and talk introduction

Host introduces show and topic

Elise Hu introduces TED Talks Daily and frames the question of what AI has to do with sports equity[2:00]
She previews that sports scientist Richard Felton-Thomas will explain how AI and biomechanics can level the playing field in athlete discovery[2:17]

Context of the talk

The talk was given at TED Sports Indianapolis in 2025[13:24]

Reframing sporting greatness and where talent is found

Common image of sporting greatness

Richard asks the audience to visualize "sporting greatness" and lists examples of famous athletes such as LeBron James, Caitlin Clark, Saquon Barkley, Simone Biles, Ronaldo, and Messi[2:44]
He notes that people tend to think of athletes from a relatively small subset of countries[2:49]
He says it would be understandable to think the best talent in the world just comes from those places[2:55]

Talent versus opportunity and visibility

Richard emphasizes that talent existed long before the athletes became globally great[3:00]
Many things have to happen along the way for athletes to realize their potential[3:04]
A commonly overlooked factor is that elite athletes got an opportunity and were visible to decision makers[3:11]
He states that talent exists everywhere, and the challenge is finding it[3:17]

Limitations and biases of traditional scouting

Scouting as a traditional pathway

Richard describes the classic scouting fantasy: a young athlete notices an older scout in the bleachers taking notes, then later gets a call inviting them to try out[3:21]
He notes that this is a thrilling fantasy for many, but there are only so many scouts[3:32]

Capacity constraints in elite football scouting

Richard, from the UK, cites Chelsea Football Club as having one of the most prestigious and well-funded youth academies in the world[3:39]
He explains that each Premier League team, or each Premier League scout, can see about 2,000 players per year[3:43]
He contrasts this with the reality that millions of people play the game[3:48]
Even among those 2,000 viewed players, scouting is already heavily limited by factors like geography, cost, and access[3:51]

Social media as an imperfect alternative

Richard notes that many young people now upload social media clips of their best plays to take exposure into their own hands[4:02]
He points out that this effectively replaces a human scout with an algorithm not designed for talent identification deciding who gets seen[4:09]
He poses the question of what to do about this situation[4:12]

Questioning whether elite clubs can see all global talent

Richard says it is simply not possible for Chelsea Football Club to see every talent in the world, then asks whether that might actually be possible with new technology[4:14]
He introduces technologies like computer vision, AI, and deep learning as tools that can help bridge this gap[4:23]

Richard's background and genesis of AI Scout

Biomechanics background

Richard clarifies that he is not a scout or a coach; his route is biomechanics, the science of motion[4:29]
He describes how biomechanics professionals typically work in sports laboratories or with clubs to improve athlete performance or reduce injury risk[4:34]

Meeting founder Darren Perry and identifying the problem

Richard recounts working in his lab when Darren Perry, who would become his founder and CEO, arrived with his son, Reef, who had an injury[4:42]
They began analyzing Reef using the lab's high-end equipment[4:46]
During this process, they started talking about the scouting problem[4:49]
Darren observed how unfair and biased scouting could be and said he had seen this firsthand[4:51]
Richard quotes Darren saying that entire futures could be decided in one day by one person with an opinion[4:57]
Darren also noted how devoid of data youth scouting often is[5:03]

Vision for a smartphone-based, equitable testing platform

Darren proposed taking the lab protocols, data, and equipment concepts and translating them into standardized smartphone drills[5:15]
The goal was that any kid anywhere in the world could be tested fairly and equitably via these drills[5:20]
Richard describes this as a brilliant vision addressing a genuine problem[5:23]
He joined Darren's team at AI.io, a company building AI-based solutions across all of sport[5:28]

Creation of AI Scout

The first product they built was AI Scout, designed for the youth talent identification problem[5:31]

How AI Scout works: technology and measurement

User experience and data capture

Richard explains that a kid downloads the app for free and records themselves performing predefined drills directly on their phone[5:39]
The recorded video is analyzed in the cloud using computer vision AI[5:43]
The system analyzes 22 key body segments in the video[5:47]
The AI can also infer 3D motion from 2D video[5:50]
From these analyses, they extract metrics such as running direction, turning, jump height, speed, symmetry, and coordination[5:59]

From raw data to meaningful scoring

Richard states that collecting raw data is only one part of the problem[6:02]
They must interpret the data and create a score for each athlete[6:08]
Most importantly, the scoring has to be specific to what talent seekers (clubs, scouts, coaches) care about[6:13]

Need for sport- and club-specific metrics

In football (soccer), different teams, scouts, or coaches look for different attributes depending on their current needs[6:26]
Some may prioritize power and pace, while others focus on coordination, technique, and high-quality body movement[6:36]
Therefore, AI Scout must be tailored on a club-by-club basis to match those differing requirements[6:42]

Collaborating with elite clubs to build and validate the system

Partnerships with Burnley FC and Chelsea FC

To answer key questions and build out the product, they partnered with two Premier League teams: Burnley Football Club and Chelsea Football Club[6:48]

Defining what scouts need from the metrics

Richard asked the clubs what they needed to know from football-specific metrics to make the data relevant and usable[6:56]
Clubs said they needed comparable, benchmarkable, and reliable data[7:04]
Above all, both the scouts and the kids being analyzed needed to understand where the data comes from, with no ambiguity[7:08]

Designing predefined drills

They worked with the clubs to develop predefined drills, including 10-meter sprints, countermovement jumps, dribbling through cones, passing, and shooting[7:22]
He notes that these drills are not new but mirror what scouts normally look at when assessing players[7:28]

Translating scout intuition into algorithms

The team sat down with scouts and reviewed a large amount of video, repeatedly asking whether they preferred player A or player B[7:46]
Richard says the art of scouting is incredibly complex and much of what experienced scouts do is intuitive[7:40]
Scouts often knew they preferred one player but could not easily articulate why[7:46]
The AI Scout team had to sit with scouts, ask detailed questions, and convert their insights into quantifiable criteria usable for scoring[7:55]
They created an algorithm reflecting those preferences and then ran thousands of videos through it[7:58]

Establishing benchmarks across age and gender

By processing thousands of videos with the algorithm, they began to create benchmarks and standards across age and gender[8:07]
Richard emphasizes that you cannot fairly compare a 13-year-old to a 22-year-old, since the older player is almost always bigger, stronger, and faster[8:13]
Instead, 13-year-olds must be compared to other 13-year-olds to reveal who truly stands out[8:17]

Early validation: discovering a local, overlooked talent

Testing with UK college players

During early app development, they recruited 50 college kids from the UK to test the system[8:27]
Richard describes these players as generally average footballers, except for one player named Ben[8:31]
All participants performed the drills, and Ben was clearly head and shoulders above the rest[8:37]
Ben was 17 years old, and the team was surprised that he had not been scouted before[8:40]

Proximity to elite infrastructure but still invisible

Richard notes that Ben lived just minutes away from Chelsea FC's training ground, not in a remote village[8:47]
Despite being near one of the world's best academies, the existing scouting system had not identified him[8:53]
Richard contrasts this with AI Scout, which did see and flag Ben's talent[8:40]

Outcome of Ben's discovery

Richard briefly summarizes Ben's trajectory as evidence that the system worked[8:53]
• Ben received a trial at Chelsea Football Club
• He scored in his under-18 debut for Chelsea
• He later signed for another Premier League club and represented his country
This case gave the team confidence that what they were building was effective[8:59]

Extending access: remote and underserved regions

Ensuring the technology works for everyone

Richard says their goal was not just to help nearby players like Ben; the system should work for everyone, including those in remote places[9:11]
He explains that because the heavy processing-video analysis, scoring, and modeling-occurs in the cloud, location becomes less of a barrier[9:19]
If someone has access to a smartphone, whether in London or Mumbai, the system can work for them[9:25]

Partnership with Reliance Foundation in India

AI Scout partnered with Reliance Foundation in India[9:28]
Richard describes Reliance Foundation's program that sends scouts each year to find the best 11-year-old talent[9:35]
• These selected 11-year-olds receive five-year scholarships to play sport and receive free education
• The best of these scholarship athletes often go on to play professional sport in India
Reliance faced the same problem as Chelsea: a few scouts could only see a few thousand people, while potentially millions were eligible[9:49]
They turned to AI Scout to help find hard-to-reach kids in difficult locations[9:55]

WhatsApp-based outreach and remote trials

Reliance put out a call on WhatsApp to their audience, a new approach for AI Scout[9:57]
Parents and students were asked to download the app, complete the drills, and trial for Reliance Foundation directly through AI Scout[10:03]
Richard reports that tens of thousands of kids now do this every year[10:11]
The top performers, based on their data, are invited to an in-person talent identification day where scouts decide who receives scholarships[10:13]
He presents this as an example of augmenting the existing scouting process rather than replacing it[10:23]

Success story: community phone user in India

Richard highlights one notable success: a player who downloaded the app from a shared community phone[10:30]
This player had never played organized sport before using the app[10:33]
Despite that, he earned a five-year scholarship through the Reliance Foundation pathway[10:37]

Supporting Youth Olympics preparation in Senegal

IOC and Intel partnership context

Richard explains that the International Olympic Committee (IOC) and their partner at the time, Intel, reached out to AI Scout[10:41]
They were focused on the upcoming Youth Olympics in Senegal[10:43]
There was concern that Senegalese national teams did not have enough talent to fill all the teams for the Youth Olympics[10:47]

Using AI Scout to match athletes to sports

Richard notes a key capability of the technology: athletes can try out for a specific sport, like football, through the app[10:52]
Conversely, the system can analyze an athlete's strengths and suggest sports they might be good at[11:03]
• If someone has great acceleration and reactive strength, suggested sports could include rugby sevens or futsal
• If someone has great upper-body power and hand-eye coordination, suggested sports could include baseball or softball

Implementation in Senegal

For Senegal, they loaded the app onto tablets and gave them to military leaders and school teachers[11:19]
These adults recorded the kids in their classes performing the drills[11:23]
Within a few days and after thousands of participating kids, 40 athletes were selected to be trained ahead of the Youth Olympics[11:28]
• Selected athletes were placed into sports such as wrestling, athletics, and football

Current deployment, scaling, and data transparency

Partnership-driven usage model

Richard explains that the app currently operates through partnership programs and talent ID initiatives with clubs and federations[11:37]
Sometimes there are brand partners on the app[11:38]
Through these programs, hundreds of thousands of kids complete the drills and trial via the app[11:42]
Hundreds of these kids have had successful outcomes and are now playing professional sport[11:48]

Global scaling: language and cloud infrastructure

Richard says the next step is to go global with the system[11:52]
He notes that the app is now multi-language[11:55]
Multi-cloud capability is coming, making the system cloud-agnostic[11:58]
Cloud-agnostic infrastructure will allow them to place country-specific components into the app for any region they work with[12:01]

MLS Next program in the United States

Richard announces that Major League Soccer (MLS) has rolled out the app to their MLS Next program in the US[12:12]
He states that 45,000 kids are currently using the app three times per year: pre-season, mid-season, and post-season[12:18]
This schedule allows them to track and monitor players' changes over time[12:21]

Data access and transparency for coaches and scouts

Scouts and coaches receive all the data in real time via AI Scout's control center[12:27]
Richard emphasizes that the system is not a black box[12:30]
Instead, it is something that scouts and coaches can trust, learn from, and use to see where the data comes from[12:32]

Future directions: healthcare and multi-sport movement libraries

Potential expansion into healthcare and medical

Richard says it is not a big leap to imagine moving into at-home healthcare and medical applications with the same technology[12:42]

Building movement libraries across multiple sports

They are starting to create movement libraries for sports like American football, basketball, baseball, and cricket[12:46]
He explains that underlying movement primitives-such as cutting, decelerating, jumping, throwing, and striking-translate well across many sports[12:56]

Closing message: universality of talent and role of technology

Universal talent and hidden brilliance

Richard asserts that in sport, one thing is always true: talent is universal[13:01]
He adds that brilliance exists in every corner of the globe[13:03]

Role of smartphones and technology in leveling the field

With technology and a smartphone, hidden talent can be made visible[13:07]
He concludes that this visibility can help level the playing field in sports[13:12]
Richard ends by thanking the audience[13:14]

Post-talk information and credits

TED curation guidelines mention

After the talk, listeners are directed to TED.com slash curation guidelines to learn more about TED's curation[13:32]

Production and fact-checking credits

The episode was fact-checked by the TED Sports Research Team[13:38]
Production and editing credits include Martha Estefanos, Oliver Friedman, Brian Green, Lucy Little, and Tanzika Sangmarnivong[13:46]
The episode was mixed by Christopher Faisy-Bogan, with additional support from Emma Taubner and Daniela Balarezo[13:54]

Closing by host

Host Elise Hu says she will be back tomorrow with another idea and thanks listeners for listening[14:01]

Lessons Learned

Actionable insights and wisdom you can apply to your business, career, and personal life.

1

Standardizing and digitizing evaluation criteria can dramatically expand access to opportunities by reducing reliance on subjective, one-off judgments.

Reflection Questions:

  • • Where in my work or life do important decisions still depend on a single person's opinion rather than clear, shared criteria?
  • • How could I introduce simple, standardized assessments or checklists to make a process I care about fairer and more transparent?
  • • What is one decision process this month that I can redesign to rely less on intuition alone and more on clearly defined factors?
2

Leveraging widely available technology, like smartphones and cloud-based tools, allows you to reach people far beyond traditional physical or geographic limits.

Reflection Questions:

  • • Which groups or regions that I care about might I be overlooking because I assume they are too hard to reach?
  • • How could existing tools I already use-such as mobile apps, video, or cloud services-be repurposed to include more people in my projects or initiatives?
  • • What concrete step can I take this quarter to make one product, service, or opportunity accessible to someone who currently can't reach it?
3

To build effective AI or data tools, you must deeply encode expert intuition into measurable signals rather than expecting experts to be replaced.

Reflection Questions:

  • • In my domain, whose tacit knowledge or intuition is critical but not yet captured in any structured way?
  • • How might I sit with those experts, ask detailed questions, and translate their instincts into observable, measurable indicators?
  • • What is one process where combining expert judgment with simple data analysis could immediately improve the outcomes?
4

Creating age- and context-appropriate benchmarks ensures that comparisons are fair and that genuine outliers can be identified within their peer groups.

Reflection Questions:

  • • Where am I currently comparing people, projects, or results without accounting for differences in starting point or context?
  • • How could I define more appropriate peer groups or baseline expectations so that I recognize true standouts accurately?
  • • What is one evaluation system I use (for people, teams, or ideas) that I can refine this week to be more tailored and equitable?
5

Augmenting existing systems with technology, rather than trying to replace them outright, can make adoption easier and outcomes more trusted.

Reflection Questions:

  • • Which current processes in my organization could be strengthened by adding data and tools instead of trying to rebuild them from scratch?
  • • How can I involve the people who currently run a process so that any new technology reflects their needs and earns their trust?
  • • What is one pilot project I could launch that layers a simple digital tool on top of how things are already done to show clear improvement?

Episode Summary - Notes by Jamie

How AI is discovering athletes that human scouts miss | Richard Felton-Thomas
0:00 0:00