Artificial Intelligence Takes a Seat at the Coaching Table: What Boundary Remains for Esports?

During the break between game two and game three of a major match, the coaching room was so quiet you could hear the computer fans. On a secondary screen, a line of suggestion quietly appeared: if the team kept this roster into the next game, its estimated win rate would drop by eleven percent. No one spoke for three seconds. Then the head coach tapped the table twice, a familiar signal, and the room erupted into debate. Fifteen minutes later, the team walked out with a choice almost no fan could have predicted.

To me, that was the moment esports crossed a boundary. Over years of following the scene, I grew used to coaches holding sticky notes, to analysis sessions dragged out until dawn, to notebooks filled with symbols only the writer could read. A tool that can offer a recommendation within seconds is something else entirely. It is no longer standing in a supporting role. It sits level with the human being.

The story of Jack Williams, iTero and Giant X made me pause longer than usual, because it touches precisely the boundary very few people in the industry want to name.

From notebook to algorithm

Giant X was formed from the merger of Excel Esports and Giants Gaming, becoming a competing member within the EMEA league system. This is a closed league model: member teams face no relegation threat, their slots remain stable across seasons, and a structural advantage held by one organisation can persist year after year rather than being competed away.

iTero, as it appeared in the conversation with Jack Williams, is a coaching platform built on artificial intelligence. It reads through a vast corpus of played matches and returns to the coaching staff patterns, predictive models and recommendations that the human eye cannot assemble in time within the few short minutes of a match night.

What stands out lies in the two subject areas the conversation reportedly covered. The first is the question of exclusivity with Giant X and the likelihood of being copied by rivals. The second is AI-assisted cheating. Just by looking at these two areas, the nature of the story becomes clear: this is not a purely technical story, but a governance story.

When a tool steps into the coaching room, the first question a tournament organiser must answer is not how accurate it is. The first question is who is allowed to use it, and under what conditions.

Which tools are allowed to sit in the coaching room?

Start from a fact most rulebooks agree on: any form of live assistance during the match itself is banned. No one may whisper into a player's ear while a game is running. That line was drawn long ago and there is little left to argue about.

The real grey zone sits in the time between games. In a best-of-three or best-of-five series, there are short breaks, a few minutes to fifteen, for the team to adjust. During those breaks, analysing data, cross-checking models and weighing composition choices is entirely legitimate. An AI tool only needs to compress that process from hours into seconds to create a decisive advantage, without violating any clause on in-match assistance.

This is precisely where iTero touches the nerve. If their product operates at the between-games layer, they are selling something that is legal on paper yet carries enormous competitive weight in practice. The fifteen-minute break was designed for coaches to adjust tactics through experience and instinct. Now it becomes a window that an algorithm can fill with models no human can fully read in time.

Another variable determines the value of such a tool: the patch cadence of each title. For titles that update infrequently, systematically and with large changes, such as Dota 2, machine-learning models trained on historical data retain validity over long windows. There, the edge belongs to depth of historical modelling. By contrast, for titles that patch biweekly, such as League of Legends, the lifespan of any learned pattern shrinks considerably. There, the value shifts from decoding the meta to detecting the meta shifting faster than opponents, an edge of tempo rather than of knowledge.

A single product marketed identically across both groups of titles is a suspicious signal. The two environments demand two entirely different kinds of value. In infrequent-patch titles, buyers purchase depth of data. In frequent-patch titles, buyers purchase speed of reaction. These are two business models, two pricing logics and two kinds of customer.

What is worth noting is that both models rest on the same assumption: that past data predicts the near future. That assumption holds in most cases, yet collapses precisely in the most important moments, when a team unveils a choice that has never appeared in any database.

Faith in the number and the shadow behind it

What I always remind myself when sitting in front of an analytics report is this: every number has a dark zone behind it. No exceptions.

In this case, no verifiable data has been published. No sample size. No evaluation methodology. No independent evidence that iTero's tool genuinely produces a win-rate advantage, rather than merely a feeling of reassurance for the coaching staff. A tool that makes coaches more confident can improve results. But it can also make them confident without cause.

I have witnessed something similar in another field. In the summer of 2026, when European football returned after the pandemic, matches were played in empty stadiums. That summer the pitch had no people, yet every corner of the stands echoed with longing. With no crowd, you could clearly hear boots striking the ball, breathing, the goalkeeper's shout. And in that silence, something unexpected emerged: statistics could no longer conceal the feel of the match. When the roar no longer filled the gap, viewers were forced to face what was actually happening on the pitch.

Esports stands at a similar moment. The numbers AI produces are beautiful, clean, convincing. But what lies behind them? A model validated across thousands of matches, or an algorithm tuned to look intelligent? The difference between those two things is not in the interface, but in the capacity to withstand failure over time.

Esports has its own stoppage time, when the screen goes dark but the heart stays lit. And in that stoppage time, what decides is not the prediction model, but a human instinct facing a situation that has never appeared in the data. That is why I believe no tool, however strong, can replace the moment a player must decide alone in the flickering light of the screen.

The blind spot of a closed system

This is the part I believe the conversation with Jack Williams has not yet reached, though it sits right between the two areas that were covered.

Those two areas are exclusive partnership and AI-assisted cheating. The first carries a commercial frame. The second carries an integrity frame. But between them lies a third, unnamed frame: fairness within the league structure.

Consider the specific context. Giant X competes in a closed league, where every member is a permanent member. In such a system, structural advantage is not competed away. If one member holds exclusive access to an analytics tool of genuine competitive weight, that advantage persists across seasons, accumulates, and becomes part of the league structure, rather than a temporary commercial deal.

Here, exclusive advantage means something very different from an open-circuit system, where teams rise and fall and advantages dilute over time. In an open system, gaps in tooling are eroded by competitive pressure and roster turnover. In a closed system, that gap only grows.

For the league operator, this question will surface sooner or later. If a tool genuinely affects competitive outcomes, the operator will face two options: mandate equal access for all members under equal conditions, or restrict the tool itself. History shows how publishers handle similar problems: gradually tightening rules on communication between coach and player, until the boundary becomes clear and enforceable.

A contract struck at the right tempo is like a poem, not a word too many. But when that poem is written for a single member within a closed system, it ceases to be a poem and becomes a rule not yet written down. And rules not yet written down are usually written after someone has already paid the price.

I once watched a match where the whole stadium held its breath for a single moment. Three seconds in Kazan stretched longer than a fan's lifetime. Those three seconds were not in any prediction model. They were created by a human deciding to accelerate when every number advised otherwise. Esports needs to preserve such moments. And to preserve them, it needs a fair system, where the human remains the final decider.

What remains on the screen

People change players, change tactics, but no one can change memory. And in esports, memory is made of moments that cannot be calculated in advance.

The conversation with Jack Williams about iTero and Giant X does not end with an answer, and I think that is the right way for it to end. Artificial intelligence will keep stepping into the coaching room, carrying numbers more beautiful, faster, more convincing. But the central question is not how strong the tool is. The question is this: what does a sport want to keep, when it lets an algorithm sit level with a human being?

When a tool recommends a composition and a player decides to go the other way, that may be a mistake, or it may be the finest moment of the match. The boundary of esports in the years ahead does not lie on the pitch. It lies in the decision of someone, in a silent room, wondering whether to trust the number or to trust themselves. That is a question no algorithm can answer on their behalf. And it is the question this sport must answer for itself before it is too late.