When AI Takes the Coaching Seat: The iTero–GIANTX Case and the Boundary the LEC Hasn't Drawn

Hook

I was watching the LEC Spring 2026 match with my laptop open and a stat sheet loaded on the side. GIANTX drafted Sett jungle, Cassiopeia mid, banned Kalista in the first rotation — the same template three games in a row. A caster on stream said: "They're playing like they can read the future." Chat erupted in laughter. I didn't laugh, because three weeks earlier I had read an interview with Jack Williams — CEO of iTero — on what looked like a B2B industry trade outlet. That article mentioned neither Sett nor Cassiopeia. It talked about something the esports industry has been deliberately ignoring: when an AI analytics tool is signed on an exclusive contract with a team inside a closed league, the line between "coaching" and "structural cheating" has been quietly erased, and nobody noticed when it happened.

Context

Jack Williams is not a familiar name to Vietnamese fans. He is the figure behind iTero — an AI tool reportedly providing tactical analysis to LEC teams. The original interview appeared around mid-2026, when the global esports industry entered a phase of sponsorship saturation paired with an explosion of LLM- and machine-learning-based support tools. Around the same period, GIANTX — the organisation formed through the merger of Excel Esports and Giants Gaming in 2026–2026 — was looking to compete with G2, Fnatic, and Team BDS in EMEA.

What makes the interview worth analysing is not the technology itself, but the two sub-headings that were disclosed: one on working exclusively with GIANTX and "the likelihood of being copied", another on "AI-assisted cheating". The gap between those two headings — a gap neither Williams nor the editor chose to address — is where the real story sits. I spent three days re-reading every public LEC document on third-party software, cross-referencing Riot's competitive integrity rules from 2026 to date, and reached a conclusion that English-language press has been sidestepping.

Core

First, the commercial structure of the deal must be understood. iTero does not license to every team — if it did, the product would become a commodity and margins would collapse within a season. They sign exclusivity region by region or team by team. Williams has openly admitted this: he fears competitors will copy his model. That fear is grounded. A successful esports AI coaching tool is not valuable because of the algorithm — similar algorithms have existed in financial analytics for two decades. The real value sits in proprietary data: logs of thousands of internal scrim blocks, secret draft scenarios, even ping data from training servers. Once a team has shared tactical data with the vendor, switching providers means losing years of training data. This is the classic switching-cost trap — and it is precisely what is creating a new form of competitive imbalance that no clause of the current LEC regulations resolves.

Second, the patch-cadence issue. In Dota 2, Valve releases major patches months apart, so AI models trained on older patches have a long half-life. In League of Legends, Riot patches every two weeks, so a machine-learned model's half-life shrinks to a few versions. If iTero sells the same product to both ecosystems, the product's value inverts depending on the title. In Dota 2 it sells "modelling depth". In League of Legends it must sell "speed of meta-delta detection" — a tempo advantage, not a knowledge advantage. Same product name, two completely different value mechanisms. Any vendor marketing identically across both should be treated as a red flag by investors, because real-world performance will not replicate across titles with fundamentally different meta rhythms.

Third, and most important: the grey zone of the between-game window. Real-time in-game AI assistance is explicitly banned in every major league — nothing to debate. But the interval between games of a BO3 or BO5, or between match days — 30 to 90 minutes — has no precedent. That is exactly when GIANTX's head coach can receive a pick/ban suggestion from iTero, cross-reference it with the opponent's mistakes exposed in game one, and make a decision. No LEC rule prohibits a third-party vendor from providing on-demand analysis inside that window, because when the rulebook was written in 2026, this scenario did not exist. This is a real-time legal loophole — not in-game, but between games — and it is being exploited by exactly the teams paying the most.

The final consequence is not GIANTX's — it is Riot Games'. Once an exclusive AI tool produces enough wins, Riot faces two options: (a) mandate equal access for every team — i.e., break iTero's business model, or (b) ban this class of tool entirely — i.e., admit the LEC ecosystem was no longer a level playing field before AI arrived. Both options carry internal political costs, and both create legal exposure with sponsors who have poured capital into the ecosystem. That is why this story will not appear in any official press release over the next 12 months — every party is waiting for a specific scandal to force a reaction.

There is an angle Western esports media is missing: AI is not a new problem for esports, it is an accelerant of a process traditional sport has lived through for two decades. In football, teams have used proprietary data analytics from Opta, StatsBomb, Catapult for years — but the difference is that those tools exist in an open ecosystem where any team with money can buy access. LEC is a closed league, every team is a permanent member, there is no relegation pressure to naturally erode advantages. In that environment, exclusive advantages compound exponentially across seasons rather than being competed away.

Look across to VCS — Vietnam's League of Legends league. VCS is currently transitioning from a closed model to a hybrid structure. If Riot applies new rules to LEC first and only later propagates them to VCS, Vietnamese teams will have at least one year to prepare. That window is golden for VCS to draft its own internal rules before being imposed from above — a lesson almost nobody in the industry is paying attention to, because everyone is busy looking West. I am not a prophet. I just read probability faster than you read emotion.

Contrarian

I may be wrong on three counts. First, my sample size is too small — three LEC Spring 2026 matches are anecdotal, not statistically meaningful. Second, Williams may be overselling iTero's capability; in B2B AI, the "revenue cosmetics" move is common and vendors routinely overstate efficacy to land the next contract. Third, GIANTX's internal policy may explicitly forbid using AI suggestions for pick/ban — meaning iTero serves only post-match analysis, and my entire thesis collapses without direct evidence. Legends don't die from mistakes. Legends die when the data learns to count. Until someone publishes an independent audit of the data flow between vendor and coaching desk, my suspicion stands on structural logic alone — but I am ready to overturn myself publicly if contradicting evidence appears.

Takeaway

The question is no longer "should AI be allowed in esports". The question is: who owns a team's tactical data, and who has the right to sell it to potential opponents? If you are a team considering a contract with iTero or any competitor, read the data-ownership clause carefully before you sign. If you are a viewer, look more closely at the second pick/ban rotation of top teams — not to suspect, but to understand that you may be watching a match in which an algorithm has already participated from behind the glass. And if you run a Southeast Asian regional league, this is the moment to draft your own rules before the crisis hits — not after.