No Audience in the Analysis Room: Jack Williams, iTero and the Unformed Rhythm of AI Coaching

I remember that sound. In a small room in Hamburg, with no audience and no commentators, there was only the clicking of a mouse against the same frame of footage. An analyst paused at the forty-seventh second of a team fight, rewound, paused, rewound. He did it seventeen times in one evening. On the seventeenth attempt, the screen beside him lit up, and a model proposed a different option, not the one he was looking for. That was the first time I understood that the biggest argument in esports today is not happening on stage. It is happening after the match ends, in the hours nobody streams, when the arena lights are off and only the people who genuinely want to know why their team lost are still awake. The first beat is not made with feet, but with ears. Here, that beat is the click of a mouse belonging to a man trying to understand what a machine is telling him. Jack Williams stands right in the middle of that silence. A recent conversation with esports media centred on iTero, the artificial-intelligence coaching tool he has attached his name to, and the partnership with GIANTX, a European esports organisation. Two sections stood out: one about working exclusively with GIANTX and the likelihood of being copied, another about AI-assisted cheating. Placed side by side, they form a question the industry has not dared to answer plainly. I do not analyse matches; I remember each face when the match ends. What I remember most from those evenings in the analysis room is the feeling of people realising they are being compared to a machine that never tires. That is neither pleasant nor particularly frightening. It is simply new, and new things always make people sign contracts before they understand what they are signing. To understand Jack Williams' story, we need to separate two things esports journalism usually merges: the product and the licence. iTero, as presented, belongs to the category of match-preparation tools: historical data analysis, opponent modelling, draft direction, evaluation of decision quality. This is the kind of tool professional teams have used in various forms for fifteen years, from analyst spreadsheets to proprietary data platforms to machine-learning models trained on tens of thousands of matches. What is new is not the function. What is new is the word AI placed in front of the product name. GIANTX is the other half of the pairing. It is a European esports organisation reported to have emerged from the merger of established brands in the EMEA region, tied to the ecosystem of one of the continent's top franchised leagues. If accurate, the regulatory framework governing the GIANTX and iTero relationship is not the rulebook of an open circuit, but the publisher's rules on third-party software and competitive integrity. This point is usually missed. In a closed franchise league, teams are permanent members with no relegation pressure and no annual qualifiers. A structural advantage, such as exclusive access to a proprietary analytics tool, persists across seasons rather than being competed away. An exclusivity deal in a closed league carries very different weight from the same deal in an open system. That is why the section on copying in Jack Williams' conversation is not merely a business story. It is the story of a licensed competitive advantage. The second section, AI-assisted cheating, is where the story touches a limit. Nearly every major league explicitly bans assistance during a live match. The contested boundary is not there. It sits in the window between games of a multi-game series, when coaches are permitted to enter the room, talk to players and make adjustments. If an AI model analyses game one and proposes adjustments before game two, is that a legitimate tool or organised cheating? No binding text anywhere answers that decisively. That is the real grey zone of esports today, and it is not in the game. It is in the rulebook. If I had to pick one place where the fight over AI coaching will be decided in the next three to five years, I would not pick the server room or the grand final stage. I would pick the seventy seconds in which a coach stands behind a player, looks at a screen, and says something or says nothing. In those seventy seconds, the value of an AI tool is measured by one thing only: whether it helps the coach say the right sentence, at the right moment, to the right person. Not a ten-page report. Not a beautiful dashboard. One sentence. This is where traditional sport and esports diverge. In football, a coach has fifteen minutes at half-time, a tactics board and the ability to draw. In esports, the coach has a much shorter window and, in many leagues, restrictions on what may be said. An AI tool therefore does not exist to replace the coach. It exists to compress the analyst's seventeen rewinds into one. Here, something interesting appears. A tool's usefulness depends heavily on the patch cadence of each title. Imagine two extremes. A title like Dota 2, with infrequent but disruptive major patches. Machine-learning models trained on historical data retain value for longer windows, because the meta shifts slowly and each shift is an analysable shock. At the opposite extreme, a title like League of Legends, with biweekly patches, shortens the half-life of any learned pattern. There, the value of AI shifts from solving the meta to detecting the meta delta faster than opponents. That is a tempo advantage, not a knowledge advantage. I stress this because it follows directly from the word exclusive in the iTero and GIANTX story. If a product is built for a specific patch cadence, signing exclusivity with an organisation only makes sense if that organisation competes in the title the product is optimised for. Conversely, if the product is marketed as title-agnostic, that is a signal worth watching, because such a product is either extremely good at abstraction, which is very hard, or is hiding weaknesses behind marketing language. During my time as a training-ground observer, I once counted how many times a midfielder turned his head to check a teammate's position before receiving the ball. One hundred and twenty-seven times across four practice blocks. That number appeared in no report. It existed only because I sat there long enough. The point is that the data an AI model needs is not always officially recorded. Sometimes it is available only if you are in the room, at the right moment, willing to count. That is why I believe AI coaching tools will not replace human analysts in the short term. They will change what human analysts do. The analyst will no longer spend seventeen rewinds finding the truth. That person will spend time deciding which truth is worth telling a player before the next game. The job shifts from searching to selecting. Selection is a different skill, harder to teach and harder to measure. At this point I want to return to copying. I think it is the wrong question. The right question is not whether rivals can copy. The right question is what remains of the advantage once they do, and they will. In software, features are copied quickly. What is hard to copy is proprietary data, proprietary process and proprietary relationships. If iTero sells features, it will be copied within a season. If iTero sells a process, the way a team turns model output into decisions, it is far harder to copy, because process lives in people, not in source code. That is why a relationship with an organisation like GIANTX can mean more than a purchase contract. It looks more like a laboratory than a customer. But it also raises an ethical question I will address later. When word spreads that an AI tool is supplied exclusively to one team, the first reaction among observers is unfairness. I think that reaction is right but misplaced. Unfairness in esports does not come from one team having a better tool. It comes from one team having a tool another team is not permitted to have, in a system where everyone competes for a single slot. The difference lies in the word permitted. A team building its own analysis system is legitimate, because that is internal capability. A commercial exclusivity bound by contract is different, because it is licensed from outside the arena. This is where publishers and league operators will be forced to speak. In esports history, in-match coach communication has been progressively narrowed for similar reasons. When a new information channel appears, regulators adjust it after it has already been used. AI coaching is at exactly that stage: used first, regulated later. That is what draws my attention to the cheating section. Not because I believe teams are cheating. Because cheating always defines the boundary of what is legitimate. When someone asks whether using AI this way is cheating, the real question being asked is what we want esports to look like in five years. That is a question of identity, not of technique. There is something I learned at St. Pauli: a training session has its own heartbeat. And that heartbeat changes when a stranger enters the room. The AI tool is the newest stranger in the analysis room. The issue is not how powerful it is. The issue is whether teams still feel the room belongs to them. Seen from the German and European market where I live and work, there is another layer. Esports organisations here operate within a legal system that values contractual transparency and worker rights. An AI tool participating in a team's decision-making raises a question of responsibility: when a wrong decision follows a model's suggestion, who is accountable? The coach, the analyst, or the tool vendor? Nobody has answered. I once sat in a team meeting in Hamburg where an assistant coach presented a twenty-page report and was asked exactly one question: so what are we changing. He went quiet. The whole room went quiet. The meeting ended without anyone changing anything. An AI tool, doing its job properly, would answer that question in one sentence. But it would also make that assistant feel partly replaced. And that feeling, no model can measure. When the ground is empty, I understand who I am keeping the rhythm for. In the story of Jack Williams, iTero and GIANTX, there is something very similar to the empty-stadium sessions I once followed: a new environment, no clear rules yet, and people inside trying to find their own rhythm before the outside imposes one. We watch matches, but we live in the silences between them. If there is one thing I want to take from this story, it is a question for the rule-makers, not the technologists: does esports want to keep that silence as a place where humans still decide, or will it let it become a data pipeline with nobody at the other end? I do not have the answer. I only know that on that evening, in the small room in Hamburg, the analyst stopped at the seventeenth rewind, looked at the screen beside him, and turned it off. He decided on his own. And that is the final detail I want to keep from this story.