AI Coaching Enters the Boundary Line: iTero, GIANTX and the Unanswered Fairness Question

Between Game 2 and Game 3 of a BO5 series, there is a silence the audience never sees. No broadcast camera points into that room, no scoreboard reflects it. Only five players, one coach, and — for the past few seasons — a sixth screen, running. That screen does not play the game. It reads it.

Jack Williams' conversation circles around AI coaching, iTero and GIANTX. In the recorded material, two section headings surface: an exclusive partnership with GIANTX and the likelihood of being copied, plus a separate topic on AI-assisted cheating. Those two names alone raise a question far larger than the product itself. But before reaching that question, I need to state plainly one thing about source quality.

Of the thirteen information points I hold from the first collection stage, ten describe the original article's author rather than its subject. Only three carry substantive content about Jack Williams, iTero, GIANTX and AI coaching — and two of those are recorded only at heading level, not body text. In other words, this is the kind of source every multi-layer analyst must face at least once: enough data to ask the question, not enough to close the conclusion.

The empty stadiums of 2026 taught me that data never lies. But they also taught me that missing data does not lie on your behalf — it simply stays silent. And silence, in this profession, is itself a signal.

Context: a product standing between two governance frames

AI coaching in esports is not a new topic. It is a frequently mentioned but rarely defined one. An AI coaching tool can be anything from a simple statistical model reconstructing win probability by game phase, to a machine-learning system predicting the next move from millions of historical matches. The boundary between "analysis" and "in-game assistance" is precisely the line every major ruleset draws.

What stands out is that this line has been drawn very clearly for the live competitive arena. Real-time in-game assistance is near-absolutely banned across major tournaments. There is nothing left to argue there. The genuine grey zone — and most likely the zone the iTero conversation touches — sits in the between-game window. Those three to five minutes when a team steps out of the playing room, breathes, drinks water, and decides how Game 3 will go.

In that window, everything is legal unless the rulebook says otherwise. And rulebooks, as usual, tend to stay silent longer than the pace at which tools evolve.

I once wrote about a similar mechanism in football: how the international federation gradually tightened rules on wearable performance-tracking devices. Tools arrive first, rules arrive later, and in the gap between the two, a few teams hold an advantage no one names.

Core analysis: the meta actually shifting is not inside the game

There is no patch anywhere in my source. No win-rate data, no pick-ban rates, no game version, no map, no items. If I wrote about the in-game meta here, I would be fabricating. So the only meta-layer I can analyse honestly is the preparation layer — how teams solve the meta, not the meta itself.

And at this layer, one structural variable decides the value of any AI tool: patch cadence.

Consider two opposite poles. At one end is Dota 2, where Valve ships major updates infrequently and disruptively — systemic changes, then long stretches of stability in between. In that environment, a machine-learning model trained on historical data retains validity across a longer window. AI's value here is depth of modelling.

At the other end is League of Legends, where Riot ships patches on a two-week cadence. The half-life of any learned pattern shortens. AI's value shifts from "solving the meta" to "detecting the meta delta faster than opponents". That is a tempo advantage, not a knowledge advantage.

Two environments, two different product logics, two different data dependencies. If a single tool is marketed identically to both, that is a red flag worth questioning. Not because it is necessarily wrong, but because it suggests the vendor is selling a shared story rather than a specific solution.

That is my first hypothesis. The second matters more, and it belongs to tournament structure.

GIANTX, in industry context, is known as an organisation based across Europe, Turkey and the Middle East, competing in a franchised league system. If that is accurate, then the exclusivity deal iTero signed with GIANTX is not merely a commercial agreement. It is a competitive variable.

In a closed league, every participant is a permanent member with no relegation pressure. A structural advantage one member holds — such as exclusive access to a proprietary analytics tool — persists across seasons rather than being competed away. In an open system, that advantage can at least be neutralised by team turnover. In a closed system, it accumulates.

This is the point I believe the original article missed. The two disclosed headings cover the commercial frame (exclusivity and copying) and the integrity frame (AI-assisted cheating). But the frame between them — the league-fairness frame — does not appear. And that is precisely the frame tournament operators will face soonest.

The winner on stage had already won beforehand — in the analysis room. But if Team A's analysis room is better equipped than Team B's because of a commercial clause Team B never had a chance to sign, then winning in the analysis room is no longer a victory of intelligence. It is a victory of contracts.

Contrarian angle: what gets copied is not the tool, but the discipline of using it

Williams speaks about the likelihood of being copied. Most readers will take that as concern over intellectual property — rivals copying the algorithm, the model, the interface. I think that misreads the centre of gravity.

What actually creates advantage is not the model. Models can be rebuilt, data can be recollected, and in an industry where personnel move between teams at high frequency, knowledge of how to build a model spreads faster than any legal barrier. What is harder to copy is operational discipline: who reads the output, when they read it, and how an organisation is structured to decide based on it rather than on instinct.

I have seen this in track and field. National teams all have access to the same type of physiological data from wearables. But performance gaps between nations did not narrow at the pace wearables spread. They narrowed only when a generation of coaches emerged who knew how to ask data the right questions.

Turning to Korean football, which I follow directly, the story is the same. K League clubs all have data analysis rooms, but the effectiveness of those rooms depends on whether the head coach is willing to sit down with the numbers the next morning. Same tool, different outcomes. That is why I do not believe the argument that "tools create advantage". People create advantage; tools only amplify it.

With iTero, I have no way to verify effectiveness claims. The source provides no sample size, no evaluation methodology, no before-and-after figures. If I were asked whether this product creates a durable edge, the honest answer is I do not know. And in this profession, saying "I do not know" is sometimes worth more than a tidy conclusion.

Numbers ask the question; psychology gives the final answer. But here, neither the numbers nor the psychology has been exposed enough.

A pretext for the future

There is one small detail in the original piece I do not want to skip. The writer mentions hoping to one day replicate Natus Vincere lifting the Aegis of Champions at Gamescom fourteen years ago, and notes that Natus Vincere lifted the Aegis of Champions at Gamescom fourteen years ago.

Natus Vincere won the first The International in 2026. If the fourteen-year gap is accurate, this article emerged around 2026. That is simple arithmetic, and it gives me a timestamp — an important one, because it places the AI-coaching debate in the context of an industry whose commercial clauses are far more complex than a decade ago.

In 2026, esports' biggest question was how to run tournaments decently. In 2026, the biggest question is who owns the advantage and whether that advantage is being distributed fairly.

Whether on grass or in an electronic arena, strategy is the common language of every game. But a language only has value when all players are allowed to learn it. If AI coaching becomes a language only a small group is permitted to study, it stops being a language. It becomes a cipher.

That is what I think franchised-league operators need to start writing into their rulebooks — not because of any specific product, but because of the structural incentive that product opens. The question is not whether AI coaching is legal. The question is by what mechanism access to it is distributed. Once an exclusive vendor becomes an official partner of a handful of teams in a closed league, the system has already chosen its answer — even if no one names that choice.