Tell us about yourself
I’m Safouen Selmi, Founder and CEO of Embedia.io, a digital transformation consultancy focused on the automotive and mechatronic industry. I have spent fifteen years working at the intersection of complex engineering systems and digital change, supporting automotive OEMs and suppliers on programs where the margin for error is measured in milliseconds.
Alongside Embedia, I am the founder of HoopX, an AI basketball coaching platform currently in development. Two separate businesses, one consistent conviction: rigorous engineering thinking applied to the right problem changes things that actually matter.
What have you learnt from working on large scale industrial systems?
The most important lesson is simple: complexity breaks at interfaces, not inside components. In large-scale programs, you are never dealing with a single technical problem. You are dealing with a system of systems, where human behaviour, legacy processes, and technical constraints interact in ways no single person can fully model.
The second lesson is the principle every well-run industrial program runs on: what you cannot measure, you cannot monitor; what you cannot monitor, you cannot control. Every system is instrumented. Feedback must be real-time: a signal that arrives the following day reports a problem, it does not prevent it. The value is in closing the loop before the damage compounds. That maps directly onto coaching: a player who gets feedback after the session has already repeated the error fifty times.
The third lesson is about cost. The most fragile systems are those only a few can afford to run. Fragility hides behind exclusivity. The best-engineered solutions work reliably at scale, for every user. That conviction shaped HoopX from day one: hardware coaches already own, at a cost any grassroots club can sustain.
Finally, the hardest problems in industrial programs are never purely technical. They are cross-functional: safety engineers, software architects, and program managers all optimising for different things. The organisations that deliver are the ones that make those perspectives work together. I carry that discipline directly into HoopX.
How has your background shaped the way you approach complexity, decision-making, and execution across everything you build?
I apply the same engineering rigour to business decisions as to system design: verify before you claim, trace every decision back to a requirement, and flag uncertainty rather than filling gaps with comfortable assumptions.
This makes me more deliberate to start and faster to finish. Projects collapse when people skip the specification phase. We do not build until we have a clear, approved plan. That discipline is what allows us to move with confidence at every step.
Execution is about reducing the distance between a decision and its consequence. The shorter that distance, the faster you learn.
How does your background in automotive and digital transformation apply to HoopX?
Working on computer vision programs in automotive gave me a specific set of instincts: build systems that are robust, not just accurate, and design for the edge cases that never appear in your training data. Tracking a basketball player in a community gym is the same class of real-time perception problem. We approached it with the same standards.
Beyond the technology, fifteen years of digital transformation shapes how we engage institutional clients: federations, academies, and professional clubs. They require a clear conversation about data governance, privacy, and scalable deployment before adopting any new platform. That is a language I have spoken for a long time.
What is AI-driven coaching and how does it work?
Traditional coaching operates on a delay. A coach watches, forms an impression, and delivers feedback after the session. At elite level, performance scientists compress that delay with video analysts and specialised tools. At grassroots level, the coach is working alone with fifteen players and forty-five minutes.
HoopX closes that gap. Using computer vision on standard camera hardware, HoopX tracks player movement, posture, and technique in real time, and delivers immediate, structured feedback to the coach’s device. No post-processing delay. No expensive equipment.
Most grassroots players are training blind: putting in the hours with no objective picture of what is actually working. HoopX changes that.
What is so unique about HoopX? What are the key features that make HoopX unique?
Three things make HoopX different.
First, it is designed for the real world, not the lab. Most AI sports tools are built for controlled environments and retrofitted to the field. HoopX was designed for the field from day one: standard hardware, inconsistent lighting, players of all levels. It must work in a community centre in Tunis as reliably as it works in a professional facility in Brussels.
Second, HoopX brings systems engineering discipline to sports performance. Most sports AI tools are point solutions: they measure one thing and report it. HoopX models the player as a system, where technique, positioning, decision-making, and physical condition interact. That produces actionable coaching insights, not just data.
Third, the feedback loop is built for the coach, not the analyst. Output is contextual and delivered in time to change what happens next in the session, not in a report reviewed the following day.
HoopX is also supported through our membership in the NVIDIA Inception program.
What is your mission?
Think about Giannis Antetokounmpo. Born into hardship. Discovered late. Trained in obscurity for years. And yet he became a two-time NBA MVP. Now ask yourself: how many players with that same potential are training right now in Lagos, Tunis, or Brussels, and being completely missed, simply because no one had the tools to see them?
That question is the foundation of HoopX.
The technology that enables real-time performance analysis already exists. But access to it has remained deeply unequal. An elite academy in Paris has performance analysts and individualised development programs. A player in a community gym in Tunisia has a coach with twenty kids and no tools. The gap is not in talent. It is in access.
HoopX is built on the belief that AI should be a democratising force: not a privilege reserved for clubs with budgets, but a standard tool available to every coach, at every level, in every country. Democratising technology does not just mean making it cheaper. It means making it genuinely usable and impactful in the hands of a coach in Nigeria as much as one in Belgium.
Talent is everywhere. But opportunity needs structure.
We are not promising the next Giannis. We are promising to stop missing him.
How are you transforming the industry?
HoopX is transforming the economics of coaching. Elite-level performance analysis previously required staff and budget only professional clubs could sustain. HoopX makes it available as a platform at a price point any grassroots club can afford.
The deeper transformation is cultural. Most coaches at community level have never had access to objective, real-time data. Once they do, the way they coach changes permanently: they stop relying purely on intuition and start combining experience with evidence.
HoopX also creates a shared data layer where the technical coach, physical trainer, and any support staff see the same picture of the player in real time. That coordinated, multi-discipline approach was previously only viable at elite level.
What has been your biggest success since launching?
Being accepted into the NVIDIA Inception program has been a strong signal that we are moving in the right direction. NVIDIA Inception supports early-stage AI startups building applied technology, and the access it gives us to expertise and infrastructure is something we could not have reached alone.
The 40+ coaches and clubs on our waitlist across Italy, Belgium, and Tunisia, before we have even entered our pilot phase, tells us the market pull is real. When coaches are actively asking when they can get access, you know you are building something worth building.
How are you building HoopX?
HoopX is built around systems thinking applied to sports performance. This is also deeply personal: I am an active player and a father of young players. I have been on courts where a coach is trying to manage fifteen players alone, with no tools, no data, and thirty seconds per kid. That is where this idea became real for me. I know what coaches are missing because I have stood next to them and watched.
The product is in its MVP phase, with early pilot discussions underway in Italy, Belgium, and Tunisia. The team is deliberately cross-functional: AI engineers, sports scientists, and coaches shaping the product together. This is fundamentally a systems thinking problem applied to sport, and the people closest to the court inform every key decision.
We are building toward our first structured pilot deployments and entering our growth and investment phase.
Why are you more focused on bringing a real world AI solution with HoopX?
Because AI that does not work in the real world is not AI. It is a demo.
The most common failure mode in applied AI is the gap between benchmark performance and field performance. A model that performs well on curated data performs poorly when the gym lighting is inconsistent, the camera angle is off, or the player is wearing a loose jersey.
Every system I have built had to work in the field, not the lab. HoopX is held to the same standard. We test in real gyms, with real coaches, across different countries and conditions. We design for the worst case, not the best-case demo.
What is currently driving the expansion of professional basketball in the GCC?
The Middle East is rapidly emerging as a new centre of gravity for basketball and sports innovation. With the FIBA Basketball World Cup 2027 in Qatar, the expansion of NBA activities in the UAE, and the entry of Dubai-based teams into elite competitions like the EuroLeague, the region is investing heavily in infrastructure and long-term talent development.
This creates strong alignment with a platform like HoopX. The ambition is being built top-down, through investment, infrastructure, and institutional commitment. HoopX works bottom-up, developing the talent pipeline that makes all of it sustainable. Both directions are necessary. We are building from the ground.
What is your long-term vision for HoopX? What do you see as the future of the industry?
The long-term vision is a global AI coaching platform operating at every level of the game: from a ten-year-old in a community gym in Tunis to a semi-professional club preparing for regional competition. A platform where every player has access to the same quality of analysis previously reserved for the elite.
AI will become infrastructure in sport the way video became infrastructure in the 1990s. The question is not whether that shift will happen. It will. The question is whether the tools driving it are designed for the elite and trickle down, or designed for the many from the beginning.
HoopX is building from the bottom up. That is where the scale is, where the impact is, and where the most interesting talent is waiting to be found.
Where can readers find out more?
You can learn more about HoopX at www.hoop-x.ai. For those curious about the engineering methodology behind it, Embedia is where that thinking lives in full: systems engineering, digital transformation, and applied AI for the automotive and industrial world, at www.embedia.io. Connect with me on LinkedIn. I welcome conversations with those building the future of coaching and player development.
https://www.linkedin.com/in/safouen-selmi/
