iTero, GIANTX and the Red Line of AI Coaching: When Exclusive Advantage Enters the Professional Arena

iTero, GIANTX and the Red Line of AI Coaching: When Exclusive Advantage Enters the Professional Arena

The five minutes nobody sees

There is a stretch of time no spectator watches, yet it decides the outcome of an entire best-of-five: the five minutes between game two and game three. No highlights, no close-ups, no caster screaming. Just a head coach in front of a screen, a few pages of notes, and one decision — keep the composition you built across two games, or tear the whole thing down.

For most of professional esports history, those five minutes lived almost entirely inside the head of the coach and the coaching staff: a blend of experience, intuition, memory of past losses, and sometimes nothing more than a feeling. Watching international tournaments over the past few years, I noticed something else quietly sliding into that window. On the coaching desk, next to the notebook, a second screen appeared — one where a tool returns probabilities, trends, and behavioural patterns the human eye struggles to catch.

The story of Jack Williams and the tool iTero, together with its relationship to the organisation GIANTX, drags that something into the light. But the light does not make it clearer. It only confirms one thing: esports still has no governance framework for a tool that can tilt match outcomes while almost nobody controls it.

Context: when did AI coaching leave the lab

When people hear "AI coaching" in esports, most think of a virtual assistant calling out every move mid-game. That is not where the real debate happens, because real-time in-game assistance has been explicitly banned in every major title for years. The grey zone sits elsewhere: the pre-match window, the between-game window, and the post-match window.

Three tiers of tools need separating. The first tier is traditional statistical tooling — spreadsheets, win rates, head-to-head history. That has existed for a decade and nobody argues about it. The second tier is model-driven analysis — probability simulations built on thousands of past matches to suggest draft directions. That is newer, but still permitted. The third tier — and this is where iTero sits — is machine learning capable of extracting new behavioural patterns that humans never encoded, then turning them into actionable recommendations within seconds.

The difference between tier two and tier three is not technology. It is learning speed. A statistics dashboard only answers questions someone already knew to ask. A machine-learning model can answer a question nobody thought to pose — and that is precisely the boundary that forces league administrators to sit down at the table.

According to a recent interview with Jack Williams, the man behind iTero, the content revolves around two axes: first, the exclusive partnership with GIANTX and the risk of being copied; second, the possibility that the tool could be abused for cheating. Those two axes look separate, but they are two faces of the same governance problem.

One thing must be said plainly about the limits of the source: the interview discloses no patch, no version, no sample size, and no methodology for evaluating iTero's performance. Which means any claim of the form "iTero is X percent stronger than rivals" is unverifiable from the available material. I will not invent a single number to fill that hole. Instead, I analyse the structure of the problem — the part that can be reasoned about without internal data.

Who GIANTX is, and why the name matters

GIANTX, by general industry understanding, is an esports organisation based in the EMEA region, formed through the merger of Excel Esports and Giants Gaming, and competing within Riot Games' league system — specifically the LEC. Details of that merger need further verification against official sources, but the direction is clear: GIANTX operates inside a closed league.

"Closed league" — franchised league — is the key concept for understanding why the iTero story is not merely a business story. In a closed league there is no relegation. Member teams persist season after season, regardless of results. That completely inverts the usual logic of competition.

In an open system, any advantage that fails to produce wins gets competed away over time — weak teams drop, strong teams rise, and structural advantages get flattened by result pressure. In a closed system, that pressure disappears. A structural advantage — such as exclusive access to an analytics tool — does not get competed away. It accumulates.

This is the point very few pieces about AI in esports ever reach: the franchised nature of the league turns any exclusive tooling deal from a commercial decision into a competitive-fairness decision. If GIANTX has the right to use iTero and other LEC teams do not, then the capability gap between them and the rest is decided not on the server, but at the negotiating table.

Core analysis: when an advantage cannot be competed away

I want to stay here a little longer, because this is the heart of the whole story.

When everything is too stable, I start looking for the crack. And the crack in closed leagues is not on the standings. It is in the things that never appear on the standings: the practice room, the analysis room, and the vendor contract.

Picture the mechanism in three steps. Step one: iTero collects data from GIANTX matches and their opponents'. Step two: the model learns from that data and produces increasingly accurate recommendations. Step three: GIANTX uses those recommendations to win more, generating more high-quality data, closing the loop.

The critical point is step three. In an open system, this loop gets blocked — either because other teams do the same thing, or because weak teams get removed from the game. In a closed system, nothing blocks it. Whichever team enters the loop first keeps the advantage, and every passing season only deepens it.

Data does not create revolutions, it only exposes who is running on gut feeling. But data plus exclusivity creates something else: a barrier to entry that no current rulebook defines. Riot Games has rules on third-party software, rules on competitive integrity. Whether an analytics tool that never touches the game client falls inside those rules — that question has no answer yet.

The interview devotes a section to the likelihood of being copied. That shows Jack Williams himself is aware that iTero's exclusive advantage is not durable if the technology is copied quickly. But here lies a paradox: if the technology is easy enough to copy that copying is a worry, then its competitive edge is not large enough to create serious inequality. Conversely, if it is hard enough to copy that real inequality emerges, then the barrier to entry it creates becomes a fairness problem for the entire league.

Glory is only the canopy; the root is who dares take responsibility. Here, the root is a question nobody wants to answer: when an exclusive tool helps produce a championship, who is accountable for the fairness of that championship?

Patch cadence: the quiet variable that decides every model's value

There is a variable almost always skipped in pieces about AI in esports, yet it decides the entire commercial viability of a tool: patch cadence.

Dota 2 and League of Legends run on two completely different philosophies of change speed. Valve — Dota 2's developer — pursues large, infrequent patches with high systemic disruption, interspersed with long stable stretches. Riot Games — League of Legends' developer — pursues a biweekly cadence, small but continuous.

That difference decides the value of a machine-learning model in ways few notice. For Dota 2, a model trained on historical data retains accuracy across long windows, because the meta is stable. For League of Legends, every learned pattern has a short half-life — meaning the tool's value lies not in "solving" the meta, but in detecting the meta shifting faster than opponents can.

Those are two different products, sold to two different customer groups, with two different core values. A tool advertised identically for both titles is a red flag. And if iTero is positioned inside the League of Legends ecosystem through GIANTX, then its value depends on update speed, not on the depth of its data lake.

This is why I say any judgment of iTero requires one fact absent from the interview: the model's refresh cycle. Without it, every claim of "durable advantage" lacks foundation.

The boundary between assistance and cheating

The interview's second section — the risk of AI-assisted cheating — lands on the most sensitive spot.

Three categories of behaviour must be separated. Category one: post-match analysis to improve tactics. That is fully legitimate, like reviewing footage. Category two: pre-match and between-game analysis to adjust plans. That is the grey zone — legitimate if the tool only aggregates public data, potentially unfair if it uses proprietary data. Category three: in-game assistance while the match is live. That is unambiguous cheating, banned in every major title.

iTero, as described in the interview, sits in category two. And precisely because it sits in category two it is controversial — because no clear legal definition exists for this grey zone.

This is where I believe league governance is lagging. For years, administrators spent their energy limiting coach communication during matches — rules on who may speak, for how long, and where they stand. But they have not built an equivalent framework for the between-game window, where a tool can deliver more information than any coach could speak aloud.

When a tool reads behavioural patterns the human eye misses, the question is no longer "is this player good." The question is: is this match being decided by player skill, or by the quality of the software that team signed a contract for?

And here the exclusivity story returns. If every team had the right to an equivalent tool, the debate would only be about technical limits. But when the tool is exclusive, the debate becomes about league structure. Their defeat did not come from bad luck, it came from faulty design. And "faulty design" here may be the design of the league format itself, not of the losing team.

The contrarian angle: where I might be wrong

I have laid out a fairly aggressive argument: that an exclusive deal inside a closed league creates structural inequality. Now it is time to argue against myself.

First, I may be overstating the power of analytics. If iTero's effect on match outcomes sits at only a few percentage points of probability, then the structural inequality I describe is real but negligible — and esports has more urgent problems. No public data lets me rule this out.

Second, the barrier to entry I worry about may dissolve on its own. Technology history shows exclusive analytics tools are often overtaken quickly, by rivals copying them or by open-source toolkits. If that happens in esports analytics, iTero's exclusive edge is temporary, and my structural concern will not survive time.

Third — and this is the possibility I take most seriously — perhaps league administrators have already accounted for this, with internal rules the public does not know about. If Riot Games has had a framework for third-party analytics tools for a long time, then my whole analysis is describing a gap that does not exist.

I raise these three possibilities not to dilute the argument, but to place it on an honest scale. An argument that cannot state the conditions under which it fails is an incomplete argument. The conditions under which mine fails are: the tool's effect is small, or the tool is easily replaced, or a governance framework already exists. If any one of the three holds, my conclusion collapses.

Takeaway: what the next 18 months will verify

I do not want to end with a summary. I want to set out falsifiable predictions, so that time itself can judge.

Prediction one: within 18 months, at least one major league will issue formal rules governing AI-assisted analytics tools in the between-game window. The hole is too obvious to stay unfilled.

Prediction two: if an iTero–GIANTX style exclusive deal exists and shows measurable effect, other organisations will respond in one of two ways: build their own internal tools, or demand the league guarantee equal tool access for all members.

Prediction three: public performance data will become the dividing line between trustworthy tool vendors and vendors selling only a story. Any company refusing to disclose its evaluation methodology will gradually lose credibility in the eyes of professional organisations.

Empires do not fall in a single night; they fall from the moment they believe they are empires. And in esports, the "empire" may be the belief that technology is always neutral. Technology is not neutral. Technology only mirrors the structure of the market it is sold into. When that structure is a closed league with no relegation, technology does not level the playing field — it builds another wall.

The question I leave with readers, and with those actually running leagues: if a tool can decide a championship, but only one team is allowed to use it, whose championship is it — the team that plays better, or the team that signs better contracts?

That is not a technical question. It is a question about what kind of sport esports wants to become as it grows up.