Jack Williams, iTero and GIANTX: The Undefined Boundary of AI Coaching in Esports

In August 2026, at Gamescom in Cologne, Natus Vincere defeated EHOME 3–1 in the first The International final, lifting the Aegis of Champions and taking one million dollars from a total prize pool of 1.6 million dollars. That five-man roster had no analysis room, no data specialist, and no statistical model standing behind any ban or pick decision.

Fourteen years later, in an interview centred on Jack Williams, the iTero product and the organisation GIANTX, that moment is invoked as a personal memory. The focus of the conversation has shifted entirely: analytics tooling has become an asset that can be signed to an exclusive contract, and the rules governing that asset do not yet exist.

I read the piece twice. The first time to absorb its content. The second time to test whether its factual structure could bear load. Of the 13 information points in the original summary, 10 describe the article's author rather than its subject. Only three touch on iTero, GIANTX and AI coaching; two of those are inferred from section headings rather than body text.

A sample that small does not permit conclusions about a product. It permits conclusions about structure. And structure is enough to work with.

Three names, one market with no rulebook

GIANTX is an EMEA-based esports organisation formed through the merger of Excel Esports and Giants Gaming, competing in the League of Legends EMEA Championship — a closed franchise league with ten permanent member teams and no relegation. iTero is an AI-powered coaching tool. Jack Williams is the figure behind the partnership between the two.

Those three facts are everything that can be established with confidence. The rest — revenue, customer count, model architecture, performance metrics — does not appear in the source. I will not fill the gaps myself. But the gaps themselves can be analysed, because a gap in a public document is also a form of data.

Every great spreadsheet begins with an empty cell and a question. The empty cell here is this: what happens when a tool capable of influencing competitive outcomes is granted exclusively to one member of a closed league?

Two section headings disclosed in the interview shape the entire analytical frame. The first covers working with GIANTX exclusively and the likelihood of being copied. The second covers AI-assisted cheating. One commercial frame, one integrity frame. Between them sits a third frame the article never names: league fairness.

Exclusivity inside a league with no relegation

In an open circuit, a structural advantage gets competed away on its own. Weak teams learn from strong ones, copy methods, hire people, buy tools. In a closed franchise league, that self-correcting mechanism disappears. The ten LEC members hold their slots across seasons, so an exclusive advantage is not contested — it accumulates.

This is the point the interview brushes against without developing. The question with real weight is not whether iTero works. The question with real weight is: if it works, who is allowed to use it. And if the answer is a single team, the league has inadvertently legitimised inequality in match preparation.

Based on my experience tracking matches in the K League and international esports competitions, one pattern repeats: every analytics tool that becomes legal passes through four stages — prohibition, suspicion, ubiquity, and finally minimum viable requirement. Video analysis was once a privilege. In-game voice communication was once restricted. Player health tracking was once treated as an intrusion. All three completed that lifecycle.

AI tooling will not take a different road. Only the speed is in question.

The difference is this: video analysis cannot be signed exclusively. AI tooling can. That is the turning point. Before iTero, a preparation advantage always existed in human form — scouts, analysts, coaches. Humans can be bought out of contracts by rival teams, so the advantage had a natural ceiling. Software does not. An exclusive agreement with a software vendor creates a structural barrier, not a personnel barrier.

One detail is worth noting: the LEC operates under Riot Games' regulatory framework, including rules on third-party software and competitive integrity. The operator has never issued a public ruling on whether a tool like iTero falls inside or outside permitted limits. That legal vacuum is precisely the condition that allows an exclusivity deal to exist. When the law is unwritten, the market writes it instead.

Copy risk and the moat of an AI product

The first heading references the likelihood of being copied. This is the existential question for every analytics vendor.

The structure of an AI product in esports has three layers. The raw data layer — replays, match statistics, event logs — is barely proprietary, since most of it can be gathered from public APIs or from broadcast matches themselves. The model layer — machine learning architecture, weights, training pipelines — can be protected as a trade secret, but cannot be patented in most jurisdictions. The interface and workflow layer — how the tool is embedded into strategy sessions, how coaches make decisions based on it — is the hardest layer to copy, because it depends on customer relationships rather than source code.

In other words, iTero's real moat is not the model. It is the exclusivity contract and the operating habits the product creates inside GIANTX. If the tool were removed, the team would lose more than software; it would lose a workflow calibrated over months.

Error does not lie — it only whispers what we are not yet large enough to hear. Here, the error sits in the implicit assumption that copying source code equals copying an advantage. The history of sports analytics shows the opposite. Football xG models have had their methodologies publicly disclosed almost in full since 2026, yet the gap between teams that use them well and teams that use them poorly has persisted for a decade. I once built a manual xG model for FC Seoul in the 2026 K League season at sixteen years old, collecting every shot, position and angle. After round 14, I published that FC Seoul's expected goals were 0.45 per match below their opponents' average while they still sat third thanks to luck. I was mocked. Exactly five rounds later, the club fell to eighth with a four-match losing streak.

The lesson is not that I was right. The lesson is that the data was public, the method was public, and nobody acted on it. Advantage does not come from information. Advantage comes from an organisation that makes decisions based on information.

AI-assisted cheating and the gap between games

The second heading touches on integrity. A clear distinction is needed here that the interview may not have drawn.

Real-time in-game assistance is unambiguously prohibited in every major title. There is no grey zone. Nothing to debate. So if a debate still exists, it must sit somewhere else. That somewhere is the interval between games in a best-of-three or best-of-five series.

During that interval, a coach can open an analytics tool, load data from the game just finished, receive win probabilities by fight sequence, identify the opponent's weak points, and adjust tactics for the next game. The entire process takes minutes, happens outside referee observation, and violates no written clause.

This is the real grey zone. Operators have not defined the boundary between legal analysis and illegal assistance during the break, because that boundary depends on a criterion machines cannot measure: the degree to which the machine intervenes in human decisions.

A shock is only data that history has not yet had time to name. If a team wins a title because an AI tool adjusted its tactics between games, that will be the first shock of its kind. And it will force administrators to write the law they should have written earlier.

The forgotten variable: patch cadence

If I had to pick a single blind spot in this entire story, I would pick patch cadence.

Dota 2 runs on a large, infrequent and highly disruptive patch cycle, interleaved with long stretches of stability. In that environment, machine learning models trained on historical data retain value across wide time windows. The advantage tilts toward deep historical modelling.

League of Legends runs on a two-week cycle. The half-life of any learned pattern is short. In that environment, the value of an AI tool shifts from decoding the meta to detecting meta drift faster than opponents. That is a tempo advantage, not a knowledge advantage.

The commercial consequence is clear. A product marketed identically for both titles is a warning sign. If iTero advertises the same value for Dota 2 and League of Legends, either the product has not been calibrated to patch cadence, or the marketing is concealing the difference. The source does not say whether iTero distinguishes between them. That is the question I leave open.

This is what I always stress when analysing any tool: the mechanics are not in the algorithm. They are in the operating context. A model that predicts well in a stable-patch environment will fail in a fast-patch environment. Confusing raw strength with adaptability is the most common analytical error in the industry — and it repeats here in a new form.

The counterintuitive angle: correlation is not causation

Suppose data showed that teams using iTero win more. The conclusion drawn next is usually immediate: the tool produces wins.

That conclusion fails at one point. Teams that sign exclusivity deals with analytics vendors are teams with financial capacity, internal analytics departments, and a strategic commitment to infrastructure. Those teams win because of their organisational foundation, and they are also the teams most capable of signing a tool contract. The tool is a companion variable, not a causal one.

First alternative hypothesis: successful teams seek out good tools; good tools do not create successful teams.

Second alternative hypothesis: the real effect is not in the quality of analysis but in belief. Players believe they hold an advantage, so they play with more confidence. This psychological effect has been documented across many sports.

Third alternative hypothesis: the effect exists only in the early phase, when few teams use the tool. Once everyone uses it, the advantage disappears and only cost remains.

The transfer market is where emotion is defeated by probability. This is true of players, and equally true of software. In the summer of 2026, I read La Liga data from the 2026/22 season and noticed that Lee Kang-in had an expected assists figure of 0.28 per 90 minutes, second among players under 22, behind only Pedri, while his Mallorca side sat sixteenth. I wrote that if the club kept him another season, his price would triple. In the summer of 2026, Lee Kang-in moved to Paris Saint-Germain for a fee of around 22 million euros. The market had mispriced him for a full year simply because it looked at the team's position instead of the individual's metrics.

Market error is the analyst's opportunity. But for that very reason, the analyst must stay vigilant about their own error.

The limits of what can be said

I have no data on GIANTX's win rate before and after signing with iTero. I have no information on the contract's scope, duration, or termination clauses. I do not know what platform the product runs on, where it processes data, or how it complies with regional data protection rules.

Those three blind spots are enough to lower the confidence of any predictive model about iTero's future to a low level.

What I can assert is structure. An AI tool in esports will pass through four institutional stages: permitted use, controversy over exclusivity, constraint by operator regulation, and finally minimum standard. The second and third stages will host most of the public dispute over the next few years.

When the stands are empty, I hear data speak for the first time. In 2026, when the pandemic forced the K League to play without spectators, I compared data from the 2026 and 2026 seasons across every team in K League 1. The home win rate fell from 46% to 34%. Average goals per match dropped by 0.3. I wrote a 32-page report and sent it to the clubs. Suwon Samsung Bluewings replied and offered me a six-month tactical analysis internship.

The lesson from that experience applies directly to the iTero story. A systemic change always produces data before it produces regulation. The K League took nearly a season to understand that spectators were a tactical variable. Esports may take longer to understand that analytics tooling is also a tactical variable, and needs to be governed as one.

Signal for the next cycle

From the first Excel cell to the summit of Europe, data goes first and people run after. In this story, the data has been there for a long time. The ones running after it are the league administrators.

Three signals I will track over the next twelve months.

First, the moment Riot Games issues any formal statement on AI coaching tools within the LEC framework. If that statement comes, the tooling market will be reshaped within a quarter.

Second, the number of organisations in franchise leagues announcing similar partnerships with analytics vendors. If the number passes three, exclusivity becomes an arms race rather than a unilateral advantage, and the competitive logic changes entirely.

Third, the appearance of any integrity complaint related to the interval between games. This is the earliest indicator that the grey zone has been exploited.

None of those three signals is certain. That is precisely why they are worth watching. Every number is a meditation; every season is an enlightenment. And this season, what needs meditating is not who wins or loses, but who is allowed to use what tool to prepare for the match.

The Jack Williams interview on iTero and GIANTX has not answered that question. It has only confirmed that the question exists. For an industry that runs on spreadsheets, admitting an unanswered question is already progress.