The award

AINext recognized Sebastian for his work in artificial intelligence, autonomous systems, robotics and collaborative robotics research, built over more than 23 years. The award sits in the Excellence in AI & Technology Industry category.

His story starts with Oxsy, the RoboCup team he founded, which has competed since 2003 and earned multiple podium finishes at the robot soccer world championship, including first place in the Cooperative Challenge. Since 2017 he has also contributed to the MultiversX blockchain protocol as a Core Team member, working on scalable blockchain infrastructure.

With Foxsy AI he is building a community-powered data and evolution engine for robotics: competitions and simulations that generate multi-agent data, progress validated on real robots and Web3 for open, transparent participation. The goal is to let anyone contribute to robotics research without owning specialized hardware.

In Sebastian’s words: “This award belongs to everyone who has supported me, believed in my dreams and shared my belief in RoboCup’s goals, from my first participation in 2003 to what we have achieved together with Foxsy AI today.”

Receiving the award on stage and with the trophy afterwards. AINext’s announcement is on ainextconference.com.

Dubai’s global stage for AI leaders

The conference brought together AI leaders, builders and decision makers at the Crowne Plaza Dubai Deira for a day of keynotes, panels and the AINext Awards.

“Dubai’s Global Stage for AI Leaders, Builders & Bold Ideas”: the speakers and awardees of AINext Dubai 2026.

AI and the Future of Work: Upskilling Billions

Moderated by Michelle Raymond, Founder & CEO of The People’s Partner, the panel brought together Michal Shatz (Founder & CEO, OpsAgents AI), Sebastian Marian (CEO & Founder, Foxsy AI), Aman Merchant (Managing Director MENA, 10Pearls) and Harel Wilner (Artificial Intelligence Engineer, Cywareness).

The panel on stage, from left: Michelle Raymond, Sebastian Marian, Aman Merchant, Harel Wilner and Michal Shatz.

After a lightning round, each theme had one panelist leading and two reacting. Sebastian was the only panelist whose work brings together AI, robotics and Web3. He shared a story drawn from more than twenty three years of firsthand experience: from competing in robot soccer simulations to building communities that learn by playing and contribute to the robots of tomorrow, generating the scarce multi-agent behavior data that machines need to learn to work together.

Lightning round: one skill for the next five years

Asked for the one skill that matters most for workers in the next five years, in one sentence, Sebastian answered:

“Coaching machines: giving AI the goal and the limits, then coming back to assess the results and decide what to change, when to push further and when to stop. Above all, one of the most important skills for the next generation will be the ability to learn faster than the tools are changing. That means continuously experimenting, adapting and learning to use the next tool.”

Augment or replace?

Michelle asked whether we are upskilling people or teaching them to supervise their replacement. She also asked whether AI can genuinely enhance human creativity. Sebastian answered from the robotics side:

“We have already played this game. When I started, the hard problem was making a robot walk. Today, walking is largely solved and the hard problem has moved up: what should eleven robots do together and how should they collaborate to achieve a common goal? Every wave moves people one step up: from doing to supervising to directing.

“So yes, tasks will be replaced and we should say that honestly. But supervising your replacement and coaching a team are different jobs. Supervision is watching. Coaching is deciding.

“And creativity? Our AI players never get tired and can execute the same strategy over and over. But the tactics that surprise everyone still come from people. The machine scales the idea. Humans give it direction.”

Can simulation and community democratize upskilling?

The question put to Sebastian directly: if anyone can train robots without specialist hardware, can simulation and community genuinely democratize upskilling? His answer:

Sebastian on the question of access: simulation, community and learning by playing.

“When I started in 2003, I was a young student in Romania and I couldn’t afford a robot. I had just a dream and a computer. RoboCup has a simulation league, so with nothing more than that, I could compete against the best universities in the world. Twenty world championships later, I can tell you: simulation is how so many of us learned.

“So yes, simulation and community can democratize upskilling if we get three things right.

“First, remove the barrier. A humanoid robot costs tens of thousands of dollars. A simulation runs on a phone.

“Second, give people a reason to come back. In FoxLeague, you build a team of AI players, set the tactics and come back the next day to see what worked. No code needed. That’s strategy, data and AI, learned by playing.

“Third, turn learners into contributors. Language models had the whole internet to learn from. Robots have nothing comparable. Every match generates the kind of behavior data robots need to learn how to cooperate, adapt and achieve a common goal. So people aren’t only learning about AI. They’re helping to train it.

“RoboCup’s dream is to build a team of humanoid robots capable of defeating the human World Cup champions by the middle of this century. I believe the people who will coach that team are playing on their phones today.”

Sebastian’s view on the other themes

The panel also covered who pays for upskilling, where humans must stay in charge and how to tell real capability from dependency. In short, Sebastian’s view:

Who pays: “A country can buy computing power, but it cannot buy a generation that knows how to direct AI. That takes practice. RoboCup offers a model: compete, learn from mistakes and share what you built so everyone starts from a higher level next year. Organizations can do the same through small teams and real problems. When learning generates useful strategies and data, it also creates value worth supporting and contributors worth rewarding. And when learning itself creates value, the question of who pays starts to answer itself.”

Where humans stay in charge: “The machine can play, but it cannot take responsibility. Someone must set the goal, judge the result and own the decision. Autonomy is a dial, not a switch: some actions happen independently, some require notification and others need human approval. Limits belong in the code, not in the AI model’s good intentions. And juniors build judgment the way pilots do: in a simulator, learning from consequences before the risks become real.”

Capable or dependent: “Can people judge when an AI response is right and when it is confidently wrong? That takes enough knowledge of their field to know what good looks like. Then take the AI away and change the rules. If they can still explain their reasoning, spot the mistake and deliver, that is capability. If they can only move forward when the AI gives them an answer, that is dependency. Simulation gives people a place to practice that judgment, learn from mistakes and adapt when the rules change. Learn the tools, yes. But go deeper in your own craft.”

Moments from the panel: with moderator Michelle Raymond and the discussion with Aman Merchant, Harel Wilner and Michal Shatz.

About Foxsy AI

Foxsy AI is building the data and evolution engine for robotics. Its flagship product, FoxLeague, has been live on MultiversX since 28 September 2026: a soccer league where anyone can coach teams of autonomous AI players and where every match generates the multi-agent data that the robots of tomorrow will learn from.

Enter FoxLeague at foxleague.ai.