AI in the Coach's Chair: iTero, GIANTX and the Line Between Advantage and Cheating

On August 21, 2026, at the Gamescom trade fair in Cologne, Natus Vincere beat EHOME 3-1 in the first The International grand final, taking home 1 million USD and the Aegis of Champions. Fourteen years later, an interview about artificial-intelligence coaching still opens on that exact moment, with the writer admitting he still dreams of recreating that feeling.

Set the nostalgia aside and what remains is far more worth reading. Jack Williams, the figure behind iTero, talks about an exclusivity arrangement with GIANTX, about the likelihood of the product being copied, and about a question the whole industry keeps dodging: at what point does using AI for coaching become cheating? Both topics sound technical. In substance they are governance and money.

Who GIANTX is, and why that detail cannot be skipped

GIANTX was formed in 2026 from the merger of Excel Esports and Giants Gaming, and competes within the League of Legends EMEA Championship system. That detail shapes the entire analysis that follows. The LEC runs on a closed-league model: teams are permanent members, and there is no relegation slot.

In an open circuit, structural advantages get flattened by the market. Weak teams drop out, strong teams move up, and every temporary edge is competed away over time. In a closed league, structural advantages persist across seasons. If a permanent member gains exclusive access to an analytics tool with real value, that gap does not disappear on its own. It accumulates.

This is why the most important angle of the interview is the one the interview itself leaves blank. The two topics raised are exclusivity and copying, alongside AI-assisted cheating. The missing frame sits between them: competitive fairness inside the league.

Who actually holds power in this story

The tail of this story is not iTero or GIANTX. It is Valve and Riot Games — two publishers with fundamentally different patch cadences and governance philosophies.

Dota 2 ships major updates on a sparse cadence, with systemic changes and long stretches of stability in between. A machine-learning model trained on historical data retains its value across a wider time window. In Dota 2, the edge tilts toward depth of modelling.

League of Legends patches every two weeks. The half-life of any tactical pattern is shortened. The value of an AI tool shifts from "solving the meta" to "detecting the meta delta faster than the opponent". That is a speed advantage, not a knowledge advantage.

A single product marketed identically across both ecosystems is a warning sign. Correct positioning has to differ. When data speaks, the whole world suddenly listens — but data only speaks when it is placed inside the right patch cycle.

Three time windows, and only one is contested

The three windows need separating. Pre-match, AI raises no controversy: opponent analysis, drafting scenarios, probability estimates. Post-match, the same. The contested window sits in between, inside the break between games of a BO3 or BO5. That is when a model can read the data from the game just finished and propose adjustments before the next game begins. Real-time in-game assistance is already explicitly banned in every major competition, so the genuine grey zone is that break — and it is wide enough for every party to read it in its own favour.

What is exclusivity worth?

Software has a marginal cost close to zero. An exclusivity licence does not make a product more expensive to produce, but it makes the sale price far higher, because the buyer is paying for distance from rivals rather than for features. This is classic price discrimination wearing a technology coat.

Pricing models for this kind of product usually go one of two ways. The first is a per-season subscription, easy to scale but easy to replace. The second is a per-organisation exclusivity deal, with far higher margins but dependence on a single customer and on the league rulebook. The second is what iTero is testing with GIANTX.

What is that gap worth? Not prize money. In regional leagues, prize money is only a small slice of an organisation's revenue structure. Most comes from sponsorship, league revenue sharing, commerce and transfers. A Worlds slot, or a season that beats expectations, can push sponsorship contract value sharply higher the following season.

At The International, the scale was once on another level: the TI10 prize pool in 2026 reached 40,018,195 USD. That figure makes any investment in match-preparation tooling look tiny in proportional terms. Numbers do not lie; only the people reading them get it wrong. What the 40 million figure says is not "AI is valuable", but "a small error term in win probability is enough to repay the entire analytics budget".

But this is where the simple arithmetic misleads the reader. An exclusivity edge is only worth something if it is not neutralised. And there are three routes to neutralisation.

First, copying. A rival cannot copy the algorithm within a quarter, but it can copy the conclusions. In esports, conclusions travel far faster than source code. Second, data. If every team has access to public and semi-public match data stores, the edge sits in the process rather than the tool. Third, and this is the biggest risk of all, the publisher.

The twist: the biggest risk is not being copied

Every management team fears being copied. That is a comfortable fear, because it sits within their control and can be measured as market share erodes year by year.

The real risk is elsewhere: one line in a competitive rulebook. Publishers have repeatedly tightened coach communication during matches, from direct talk with players, down to listening only, and then to tightly limited windows. Each time, the value of an assistance tool is re-priced from scratch without a single line of code changing.

An AI coaching product can be legalised on a Tuesday and restricted on a Wednesday. For iTero, the exclusivity deal with GIANTX is a time-limited asset, and the term is not set by iTero.

This is where short-term enthusiasm separates from long-term value. The story of "AI changes esports" generates headlines, conference panels and valuations. The story of "who is allowed to use that tool, at what hour, in which match" is what decides how long the cash flow survives. A contract exclusivity is not a moat. The real moat is proprietary data and the depth of integration into a team's workflow — two things far harder to copy than a contract with an expiry date.

The view from Vietnam and Korea

In 2026, the Asia-Pacific regional league system was restructured, and Vietnamese teams entered a shared partnership structure with teams from Japan, Taiwan, Hong Kong, Macau and Oceania. That structure sits closer to permanent membership than to open qualification.

Based on my experience following regional and international matches across many seasons, I think this is an easily underestimated point. A region with a smaller analytics budget cannot win by spending more on tools. It has to win by asking better questions. But if the tooling layer is locked behind exclusivity deals, the thinking-based route narrows too, because good thinking still needs clean data and models strong enough to falsify it.

Korean teams are ahead at this layer. Many organisations in Seoul already run internal analytics units, hire data engineers, and build their own pipelines instead of buying packaged tools. That is an expensive strategy early and a cheap one later, because the asset sits with the team rather than with the vendor.

The world looks at the star; I look at the value sheet. For a player, value lives in the stats. For an organisation, value lives in control of its own data layer.

The landing point

The AI coaching debate will not be settled by an impressive demo or an interview. It will be settled by a document nobody broadcasts: the competitive rulebook. Who may use the tool, in which time window, and who owns match data. Those three lines determine the entire value of the analytics industry growing up around the leagues.

While waiting for the answer, teams can pick one of two roads. Buy the tool and depend on someone else's schedule. Or build internal data capability and set your own tempo. Stop arguing about the love of the game; argue about value — and in esports, value starts with which layer of the chain you own.