AI in the Esports Coaching Room: iTero, GiantX and the Grey Zone of Exclusive Advantage

Opening: Two in the Morning and a Single Line of Notes

"The old television set still remembers the summer we watched football together."

On the night of 14 June 2026, I was fifteen, a tenth-grader in Saigon. Russia beat Saudi Arabia 5-0 in the World Cup opener, and what I remember is not the five goals but the sound of an entire neighbourhood cracking open at once. That night I started a blog called "Dau Truong Meta", wrote 1,200 words, called overlapping runs down the flank "pushing the side lane" and quick counters "reading the cheese". Forty-seven shares overnight. I thought I had just learned how to tell a sports story.

Seven years later, at two in the morning, I sat in a small studio in District 1 with a second monitor showing a match from the EMEA league. No stands, no roar, only the hum of cooling fans and someone speaking very quietly into a headset. What kept me awake was not a pentakill. It was one line in the notes I was reading: a company called iTero sells AI-powered coaching tools to professional esports teams, and an organisation called GiantX has signed an exclusive agreement with them.

The match was over. The story had only just begun.

Context: An Interview With No Data Attached

The conversation centres on Jack Williams, iTero, GiantX and the future of AI coaching in esports. Two content areas are explicitly signposted: one deals with the exclusive partnership with GiantX and the likelihood of being copied; the other deals with AI-assisted cheating.

To understand why those two areas matter more than they appear to, it helps to state plainly what an AI coaching tool actually does. In the pre-match layer, it builds opponent models: ban-pick frequency by meta phase, the probability of opening through the top lane, the habit of contesting major objectives at the eight-minute mark. In the post-match layer, it automatically cuts scrims into repeating situations, tags errors and ranks their severity. In the between-game layer — the ten to fifteen minutes of a BO3 or BO5 — it can surface pick-ban suggestions and warnings about tendencies the opponent has just revealed.

At the first layer and during live play, the legal boundary is clear: every major title prohibits real-time assistance during a match. The grey zone sits exactly where nobody wants to draw a map: the between-game window and the entire preparation cycle behind the curtain.

One detail about timing is worth flagging. The interview references Natus Vincere lifting the Aegis of Champions at Gamescom "fourteen years ago". Na'Vi first lifted the Aegis at The International 2026 at Gamescom, which places the interview around 2026. That is arithmetic drawn from the article's own wording, not a disclosed fact.

And here is the point I have to state plainly before analysing anything: the interview discloses no numbers about product effectiveness. No sample size. No evaluation methodology. No win rate before and after adoption. For someone who builds his own statistical tables for every tournament, that gap is louder than any answer.

Why Patch Cadence Is the First Commercial Variable

In esports, the value of a machine-learning tool does not live in its algorithm. It lives in the lifespan of the patterns that algorithm learns. That lifespan is decided by the publisher, through patch cadence.

Dota 2 runs on Valve's rhythm: major patches arrive infrequently, but when they arrive they break structures. Between those milestones lie long stretches of stability. Inside stability, coaching models trained on historical data retain their validity longer. This is favourable terrain for deep statistical and machine-learning tooling.

League of Legends runs on Riot Games' rhythm: a patch every two weeks. Every cycle shortens the half-life of a learned pattern a little more. In that environment, the tool's value shifts from "solving the meta" to "detecting the meta delta faster than your opponent". That is a tempo advantage, not a knowledge advantage.

The distinction matters more than it looks. A single product marketed identically across both titles is a red flag. If value comes from the depth of historical modelling, the product must be re-architected for the slow-patch title. If value comes from speed of change detection, it must be re-architected for the fast-patch title. Those two architectures cannot be the same thing.

I have tested this principle against football data, where I have watched longer. When the 2026-20 Champions League restarted in empty stadiums, I logged the whole competition and found home win rate falling to 32 percent, down from 45 percent the season before. One variable was pulled out of the system and the entire meta shifted with it. AI coaching tools do the opposite: they add a variable back into the system — but not for everyone.

Exclusive Agreements Inside a Closed League

This is where I believe the most important angle of the interview goes unexamined.

GiantX is widely known as an EMEA-rooted organisation operating in the League of Legends ecosystem, formed through the merger of an English and a Spanish organisation. If that is accurate, the governing framework for the iTero arrangement is the publisher's third-party software and competitive integrity rules.

In a closed, franchised league, every participant is a permanent member. There is no relegation slot. That means a structural advantage is not competed away season by season, the way it is in open-circuit systems where weak teams leave and strong teams rise. A tool that only one member may use does not neutralise itself over time. It accumulates.

The consequence does not stop at competition. It reaches the league operator. If a tool materially affects competitive outcomes, the operator will soon face pressure to choose one of two paths: mandate equal access for all members, or restrict the tool itself. History has already walked this road with the regulation of in-game coach communication.

I call this the league-fairness frame. It sits precisely between the two content areas the interview covers, and neither touches it. The exclusivity-and-copying area is about commerce. The AI-cheating area is about integrity. The fairness frame is about whether a league is quietly selling a piece of its own competitive balance.

The Between-Game Window: The Real Grey Zone

In every major title, live in-game assistance is unambiguously banned. So where does the remaining work of an AI tool live? In the break.

A BO5 has four breaks. Each lasts around ten to fifteen minutes. In those fifteen minutes, a team has a head coach, an analyst, and possibly a model that has just finished crunching the two games already played. That model can say: the opponent won both opening games by contesting the major objective at eight minutes; or: when trailing, they always swing to the bottom lane within the first three minutes.

The question no rulebook answers cleanly: is a machine-generated pick-ban suggestion a form of coaching? Does a tendency warning extracted by a machine from fifteen minutes ago violate the spirit of the ban on live assistance?

Current rulebooks were written for human speed. Humans need time to review footage, take notes, cross-reference, argue. Machines do not. When a machine compresses that window to nearly zero, the rule still stands on its old assumption. That is the real tension, and it does not live in the letters "AI" in a headline.

Lessons From Empty Stadiums and From the Bench

I found three analogies from football that clarify this story.

The first analogy is the empty stadium. As noted, removing the crowd from the equation collapsed home advantage from 45 percent to 32 percent in the 2026-20 Champions League. "The no-crowd meta taught me: the loudest applause is the applause of belief." The lesson is not that crowds matter. The lesson is that a single changed variable can restructure an entire table. An exclusive coaching tool is exactly such a variable — except it is added, not removed.

The second analogy is the five-substitution rule. Modern football allows five changes, which makes squad depth more valuable while turning the final twenty minutes into a war of attrition as both sides keep feeding in fresh legs. In esports, an AI tool deepens one team's preparation capacity. But if only one team holds that tool, the final twenty minutes of a series stop being a contest between two tactics. They become a contest between a team with machine memory and a team with human memory.

The third analogy is the transfer valuation model. Models that price young players routinely overvalue youth potential and undervalue dressing-room chemistry. That is the structural blind spot of every quantitative model: it can price patterns, not nerves, not the silence in the room at half-time. AI coaching tools carry the same blind spot. It can tell you the opponent invades the jungle at 1:20 in 68 percent of their losses. It cannot tell you your mid laner's hands are shaking.

No Data, No Verdict

I have to be explicit about this, because I know which trap I fall into.

Since 2026 I have built the habit of keeping my own records for every tournament. At the 2026 World Cup I logged 214 decisive plays across 52 matches and predicted Japan to beat Germany 2-1 after re-watching seven of Japan's qualifiers. I say that not to boast but to state a principle I trust: do not conclude before the data is in.

The interview about iTero and GiantX contains no data. That makes every claim about product effectiveness unverifiable. I am not denying the tool might work well. I am declining to grade a product I have no way to measure.

And I am grateful, because it means I still have work to do. If everything were already clear, nobody would need to sit reading notes at two in the morning.

The Contrarian Angle: The Debate Is Asking the Wrong Question

Here is where I want to push back against the most comfortable way of telling this story.

The most comfortable framing is: "AI will not replace the human coach." It sounds warm, it sounds humane, and it is eminently shareable. It is also a straw man. Nobody serious in the industry thinks a model will sit in the head coach's chair and call the draft. The real debate is not whether a machine can replace a person. It is who gets to rent the advantage, and for how long.

The second comfortable framing is the AI-cheating headline. It is appealing because it has a villain. But it obscures the larger part of the story: the commercial part. In a closed league, an exclusive agreement does not need anyone to cheat in order to produce inequality. It only needs a signature.

As for the question "will we be copied", I think it is asked backwards. If the model is easy to copy, the competitive moat evaporates and GiantX has paid for something rivals will have free within months. If the model is hard to copy, the moat is real and the league has a permanent inequality problem. Both branches reach the same conclusion: this is a governance question packaged as a product question.

There is one further layer that I consider the most dangerous. The edge of an AI tool is non-stationary. Once every team has an AI tool, the edge returns to zero, and the differentiating factor becomes data quality once more. Data quality depends on access. The loop closes exactly where it began: exclusivity.

Empty pitch, empty stands, but the hearts of the fans have never been muted. And in a room with no spectators, an exclusive contract can still decide who wins the title.

What I Am Waiting For

Across seven years of watching esports, I have learned that turning points rarely arrive through patch notes. They arrive through contracts.

A patch is announced, discussed, analysed for forty-eight hours and then forgotten. A contract is not. An exclusive agreement over a coaching tool signed today will quietly shape the meta of the next two or three seasons, and by the time anyone notices, nobody will remember where it started.

What I am waiting for is not a statement from a publisher. What I am waiting for is a rulebook on third-party analytics tools, written before a season is decided by one. The league that writes the rule early will shape its own meta. The league that waits will discover its competitive balance was sold in an appendix.

Some summers we do not need to rewind, because they are still playing in our hearts. But the meta always needs rewinding, note-taking and cross-checking — even when what changes is not a champion's stat line, but a signature at the bottom of a page.

When advantage is rented rather than built, who truly owns the title?