Jack Williams, iTero and the Governance Puzzle of AI Coaching in Esports

Inside a BO5 there is a stretch of time no scoreboard records: the eight to ten minutes between the end of one game and the start of the next. The crowd leaves for water. The casters re-read the gold and damage numbers. And inside the playing room, the coach gets exactly one short window to correct course before everything locks again.

That is where professional esports is actually decided, and it is also where a tool like iTero is quietly stepping in. The conversation with Jack Williams about iTero, about the exclusive deal with GIANTX and about the future of AI coaching made me sit down and write. Not because it revealed some earth-shaking technical secret. But because it touches the exact place this industry has never been willing to name: the line between a legitimate analytical tool and an unfair competitive advantage.

Context: three names and two frames of debate

Jack Williams is tied to iTero, a platform that uses machine-learning models to support coaching. The word support deserves a slow read here. iTero does not play the game for anyone. It turns match data into suggestions: what the opponent should ban, which lane is being abandoned, win rates by phase of the game.

GIANTX is the name that interests me more. The organisation was born from the merger of Excel Esports and Giants Gaming, operating within the League of Legends ecosystem in EMEA. If my memory of that deal is right, GIANTX competes in a franchised league where no relegation slot exists. It sounds dry, but it is the key to the whole story.

The interview splits into two clear halves. The first covers the exclusive partnership with GIANTX and the likelihood of being copied. The second covers the risk of AI-assisted cheating. Those two frames, commercial and integrity, cover almost every public debate about AI in esports today. And both miss the thing sitting right between them.

I have tracked the esports transfer market for a few years now. The transfer market is where people sell the past, but anyone clear-headed buys the future with data. I still use that line when explaining to Vietnamese teams why a young player is sometimes worth more than the first look suggests. But the iTero story is not about valuing players. It is about valuing tools.

The meta actually under discussion

No patch appears anywhere in this story. Dota 2 is mentioned only as a personal memory: Na'Vi lifting the Aegis of Champions at Gamescom, fourteen years ago. Subtract that marker and the article lands around 2026. But Dota 2 here is nostalgic colour, not analytical data. Anyone reading it as a signal about the Dota 2 strategic map is reading the wrong kind of document.

The meta that is shifting is how teams solve a patch. The tools they use to prepare, not the champions they pick, are changing. And right here, the patch cadence of each title determines the value of AI.

The value of an AI coaching tool inverts according to a title's patch cadence: stable patches reward depth of historical modelling, while fast patches reward speed of detecting the meta delta.

From what I understand, Dota 2 under Valve has infrequent but system-upending major patches. A model trained on historical data keeps its validity for longer windows. Conversely, League of Legends, updating roughly every two weeks, shortens the lifespan of any learned pattern very quickly. There, the value of AI is no longer solving the meta but detecting the meta shift faster than opponents, an advantage of tempo rather than of knowledge.

This leads to a question nobody in the interview answers: is iTero title-neutral? If a product is marketed identically for both slow-patch and fast-patch titles, that is a suspicious signal. Those two markets demand two different products.

Exclusivity in a closed league: an advantage that never erodes

The exclusive deal between iTero and GIANTX is the heaviest detail in the whole story, and also the least discussed.

Picture the mechanism. In an open league with relegation, any structural advantage has a short lifespan. Weak teams improve, strong teams rise, slots get swapped, rosters turn over. The advantage is eroded by the market season after season. But in a closed league, every member is a permanent member. Nobody gets relegated. Nobody loses a slot. A structural advantage, say exclusive access to an analytics tool, persists across seasons instead of being competed away.

In a closed league, structural advantages are not eroded season by season the way they are in open leagues, so an exclusive tooling deal carries far greater competitive weight.

Looking at Vietnam, our competitions mostly run on open or semi-open models with promotion and relegation. That unintentionally creates a self-correcting mechanism: a team using a good tool without winning can still lose its slot, and the advantage gets redistributed. I am not saying one model is better. I am saying the two models produce two different governance consequences, and most fans do not realise they are watching two different kinds of arena.

One more detail I noticed on rereading this section: nobody in the interview mentions whether the league operator was informed about the exclusivity arrangement. If the answer is no, then a team holds an advantage that even the league's own governing body has never assessed. That is a governance gap, not a vendor's fault.

Copy risk: where the real advantage sits

The section on being copied is the most predictable part of the interview. Every tool gets copied. The question is not whether copying happens, but how fast.

Here I have to state plainly something the original piece does not disclose. The real advantage of a coaching tool is not the algorithm, but the exclusive data access the algorithm feeds on. Machine-learning algorithms are, at the end of the day, a commodity you can buy. Clean data, correctly labelled, captured inside the right time window, that is what cannot be copied with a download.

And here I remember a principle I keep reminding myself of: data never lies; it just patiently watches you fool yourself.

The interview discloses no figures at all about the tool's effectiveness. No sample size, no evaluation method, no benchmark. A performance claim without a sample size sits in the same position as a promise on a negotiation table. That does not mean the tool fails. It only means there is nothing externally checkable, and in analytics, that is a gap rather than evidence.

If I were advising a Vietnamese team considering paying for a tool like this, the first question I would put on the table is not how accurate it is, but which dataset it was evaluated on, by whom, and whether that evaluation is independent of the vendor. A hidden model is not a model. It is a belief with an interface.

The grey window: where AI cheating actually lives

The second half of the interview deals with the risk of AI-assisted cheating. This is the most easily misread part.

Real-time assistance, meaning a program that reads the match and suggests actions while the player is competing, is already clearly banned in every major title. There is nothing to debate there. The grey zone lies elsewhere: the between-game window of a BO3 or BO5.

This sounds small, but it is not. If league rules only ban assistance during play and say nothing about the break between games, then a team using AI analysis inside the break room is doing something entirely legal. And if the opponent has no equivalent tool, that advantage becomes a visible skill gap in front of an audience that believes it is watching two equally prepared teams.

This is where I think of an old story of my own. In 2026, when the pandemic froze every sports competition, I sat at home building a player valuation model out of matches played without crowds. Covid closed every pitch, but it opened for me a data library I had never dared dream of. This industry always advances inside the gaps regulation has not yet covered. Good or bad, history has never judged. It is simply how every system operates.

The contrarian angle: the missing frame

The two popular frames, commercial copying and integrity cheating, both revolve around the vendor-customer relationship, or the player-rule relationship.

The missing frame sits between them: the internal fairness of the league itself. When a permanent member of a closed league has exclusive access to an analytics tool, the question is no longer whether the product is good, but under what conditions the remaining teams are competing. That is a question for the league operator, not for the tool vendor.

I once assumed leagues would respond the same way, but they do not. Football has VAR, and the argument around it dragged on for more than a decade because the in-stadium explanation mechanism was never designed for the audience. Esports stands before a similar fork with AI, only much faster. An exclusive analytics tool today can become a mandatory standard tomorrow, or be banned outright. Both outcomes are reasonable. What is unreasonable is letting it drift.

And if the league operator chooses silence, it is implicitly answering a question nobody has asked it. Silence in governance is always a decision, merely a decision that never makes the minutes.

Takeaway

The question I want Vietnamese teams to carry is not whether to use AI. The question is: when a tool steps into the between-game break room, who is the player, and who is the decision-maker?

And if that answer changes depending on whether the team pays for the tool, then are we running a fair competition for the fans, or a technology showcase for the sponsors?