iTero, GIANTX, and the Gap Between Two Headings on AI Coaching

In August 2026, at Gamescom in Cologne, Natus Vincere lifted the Aegis of Champions — the shield awarded to the winners of the very first season of The International. Fourteen years later, that memory resurfaced in an interview about coaching with artificial intelligence. The person recalling it was Jack Williams, attached to iTero. The phrase "fourteen years ago" anchors the piece to roughly 2026, and it also reveals the most notable thing about it: a commercial technology subject being told through the memory of a closed era.

In the recorded content of that interview, two section headings were preserved. The first dealt with an exclusive partnership with GIANTX and the likelihood of being copied. The second dealt with cheating assisted by AI. Between those two headings sits a gap. That gap does not belong to the writer. It belongs to an entire industry.

CONTEXT: AN AGREEMENT WRITTEN AS A REGULATION

Jack Williams, according to what has been recorded, is behind iTero — a developer of coaching tools built on artificial intelligence for esports teams. iTero holds an exclusive partnership agreement with GIANTX. GIANTX, according to the background knowledge of those who follow the EMEA region, is an organisation born from the merger of Excel Esports and Giants Gaming, competing in Riot Games' LEC system — a closed league in which membership slots are fixed and no relegation mechanism exists.

The difference between an open circuit and a closed league is something I learned not from esports, but from the running track. In athletics, final slots are reallocated from scratch every season. Structural advantage cannot accumulate over years, because it is eliminated before it can crystallise. In a closed league, that cleansing mechanism disappears. An advantage established this season sits untouched next season, and the season after that.

Years of dissecting electronic timing data taught me to read agreements of this kind differently from how they are usually read in the press. It does not stop at the commercial story. It is a competition regulation written in the form of a contract, signed by two parties while a third party affected by it is absent from the negotiating table.

THE CORE: VALUE INVERTS WITH THE PATCH CADENCE

To assess an AI coaching tool, the first task is to determine which platform it runs on. This is the point the esports industry rarely states plainly: the value of the same machine-learning model inverts between titles, depending on the publisher's patch cadence.

Valve runs Dota 2 on a sparse cadence. Large patches bring systemic change, followed by long stretches of stability. In that environment, a model trained on historical match data retains its validity for months. The advantage tilts toward whichever team models more deeply, not whichever reacts more quickly.

Riot Games runs League of Legends on a two-week patch cycle. The half-life of any learned pattern is shortened to a brutal degree. Here, the value of AI shifts from "solving the meta" to "detecting the meta delta several days faster than the opponent." What is being sold is no longer knowledge. What is being sold is tempo.

The same product, two markets, two promises that invert each other. If iTero markets a single message across both ecosystems, that is a signal that the entire pricing model and the entire method of measuring effectiveness need to be re-examined.

And this is where I want to pause longer, because it touches my former work.

In 2026, at the 29th SEA Games in Kuala Lumpur, I was assigned to cover the men's 800m final. A nineteen-year-old athlete finished fifth. Electronic timing data showed his cadence had reached 198 steps per minute, far beyond the optimal threshold of around 180. I wrote a piece proposing he drop to 185, lengthen his stride to conserve energy, and predicted he could run under 1 minute 49 seconds. His coach called to complain that I had left the young man rattled. The analysis was mechanically correct. It was wrong in that it failed to account for a nineteen-year-old reading that article the day before his race.

That is why I read the iTero section on AI coaching carefully. A tool that analyses movement, opponent data and outcome probabilities for a coach can be accurate to the decimal place and still fail, because what it touches is a human being preparing to step into the arena.

Raw data does not lie; it merely conceals a very deep systemic fault. The systemic fault here is the assumption that better inputs always produce better decisions.

Under that frame, the between-game window in a BO3 or BO5 series is the most discussion-worthy grey zone. Real-time AI assistance is already explicitly prohibited in every major title, so nothing remains to argue about there. The break between game two and game three, however, has never been cleanly defined. A model reads the data from game one, identifies a mid-lane weakness, and proposes a draft change within four minutes — is that post-match analysis, or is that in-match coaching? That boundary is not drawn by technology. It is drawn by rules, and the rules are lagging.

Back to the exclusivity story. If GIANTX holds a tool that the other ten LEC teams do not have, then in a closed league that gap does not close on its own. It accumulates. Season one is a small edge. Season three is a structural edge. Season five is the default condition of the league.

THE CONTRARIAN ANGLE: THE ARGUMENT IS POINTED AT THE WRONG PLACE

The industry is arguing about cheating. That argument is real, and necessary. But it is a chosen headline, not a chosen problem. It appeals because it has a good side and a bad side, suspicion and provability.

The problem lies elsewhere. When a tool vendor signs an exclusive deal with one member of a closed league, the league operator faces a choice it has never confronted at this level: force equal access for all members, restrict the tool, or accept that its league operates on an uneven playing field. The governance history of esports shows publishers have walked this exact path with in-match coach communication — from permitted, to restricted, to fully banned.

The technical difficulty is that there is no common denominator for measurement. In athletics, disciplines have absolute units: seconds, metres, grams. In esports, the edge of an analytics tool cannot be reduced to a single measure. It is visible only through results, and results are muddied by far too many variables. Every transfer deal is a model waiting for its error term to surface — and an exclusive tooling agreement is the same, except its error term takes more seasons to appear.

There is another notable detail. All the claims about the tool's effectiveness, as recorded, come with no sample size and no evaluation methodology. When the stadium stands empty, I hear the ticking of history clearly. In 2026, when every competition was suspended, I compiled the records of 120 Vietnamese athletes across the 2026 to 2026 period, covering peak age, number of coach changes, and training locations. I checked every figure so obsessively that the study ran a month late, and the results showed that 78 percent of athletes achieved their best performances within two years of settling under a coach with fewer than five years of experience. Those forty pages of data made no promise at all. They merely showed a pattern.

That contrast is worth holding onto when reading any technology claim issued by a party that is selling the product.

And there is one more variable that analysts in Vietnam rarely bring to the table. While the LEC debates access to proprietary tools, most teams in the VCS system still cannot budget for a full-time data analyst. The gap between the two sides is not in model quality. It is in the ability to pay for a model. When a debate about competitive fairness unfolds without representation from the region that has no voice, its conclusions will faithfully reflect who was seated at the table.

WHAT REMAINS

From Tokyo in 2026, I drew a principle I still use today: when the data is insufficient to assert, speak in the language of probability. With the iTero and GIANTX story, the data is insufficient along nearly every dimension — no sample size, no patch cadence, no evaluation methodology, no contract terms. There are only two section headings and one name.

If I had to bet on one thing that will shape the game over the next three years, I would not bet on what AI can do. I would bet on whether leagues compel disclosure of the tools each team uses — in the way sports compel disclosure of medication and recovery methods. That is the only thing that can turn a structural advantage into a governance question.

What shapes the game will not come from the next patch. It will come from the next contract, and from whether anyone is willing to write the third heading.