GIANTX's AI coaching app iTero passes one million downloads

iTero, the artificial intelligence coaching app owned by the League of Legends organization GIANTX, has passed one million downloads, its founder Jack Williams said in an interview published by Esports Insider on September 11. The app is built for solo queue, the ranked mode where players queue alone and are matched with strangers, and GIANTX also runs its models for its own professional teams.

The milestone puts a number on a question the scene keeps returning to: how much of the coaching once done by people at a desk, with replays and notebooks, can be handed to software, and where the rules of the game publisher stop it. Williams' answers, as Esports Insider reported them, cover both the growth of the app and the limits he sees for tools of this kind.

Figures given by iTero founder Jack Williams in an interview published by Esports Insider on September 11, 2026; the shares are his approximations.

What the app does

iTero works on the two moments where a League of Legends player makes the decisions that weigh most: the draft, when the ten players pick their champions before the match, and the game itself, reviewed afterwards. In League of Legends, each team of five picks its champions in turn, and the combination of picks shapes how a match is likely to unfold before it starts.

For the draft, the app recommends champions, according to Esports Insider. Its suggestions are weighted by two factors: how often a champion is picked in the current meta, the set of strategies and champions that dominate a given patch, and how much experience the individual player has on that champion, which the game records as champion mastery. A strong pick on paper is worth less to a player who has rarely played it, and the weighting is meant to account for that.

After the match, a feature the app calls a macro coach reviews the player's recent games to find mistakes that keep coming back. Macro, in League of Legends, is the side of the game that sits above individual fights: where to be on the map, when to take an objective, when to group with the team. The app also offers simple comparisons, such as how much farm, the gold earned by killing minions and monsters, a player collects against the average player of the same elo, the rating that places players on the ranked ladder.

Williams said the focus of the product is solo queue data rather than professional matches.

From a team tool to a solo queue app

The app did not start as a product for amateurs. Williams, who had worked as a data scientist and artificial intelligence engineer, first at HSBC and then at Deloitte, and who holds a master's degree in artificial intelligence, built models that predicted match outcomes from the draft, Esports Insider reported. He pitched them to Excel Esports, the organization that is now GIANTX, and resigned from Deloitte to build tools for teams.

Professional teams did not buy. After about three months of sales attempts that went nowhere, around 2020, Williams won a pitch competition at an Esports Insider conference, and Overwolf, the company that runs a platform for in-game apps, suggested he turn the work into an app for ordinary players. The app went live three months later, according to the interview.

It reached about 10,000 downloads shortly after launch, and the retention, the share of users who keep coming back after the first try, was unusually high, Williams said, and he summed up his priority at the time in one line to Esports Insider: "The only thing I cared about was retention." Overwolf also gave him a non-dilutive grant, money that takes no stake in the company in return, which paid for the hire of a developer. After the first two years on Overwolf's platform, downloads stood at between 100,000 and 200,000, he said.

Growth since GIANTX took over

GIANTX acquired iTero in 2024. The organization, which says on its own site that it has been in esports since 2008, fields a team in the LEC, the top League of Legends league in Europe, and a team in the Spanish league that competes under the name GIANTX Itero. Its site also lists Valorant teams and other games.

Williams described growth that accelerated after the acquisition. The chart below shows the download totals he gave, counted from the date of the interview.

Williams put the growth down to three choices: building the product first, partnerships with content creators that grew organically rather than through paid campaigns, and the educational channels GIANTX already ran for its audience. The last of these shows in who uses the app. Spanish-speaking players made up about 20 percent of users before the acquisition and about 70 percent now, he said.

The download totals are the founder's own figures, given in an interview, and a download is not the same thing as an active user: the count says how many people installed the app, not how many still open it.

Inside GIANTX, and nowhere else

The professional side of iTero stays in house. The models run entirely within GIANTX, including for its LEC team and its Spanish league team, Esports Insider reported, and Williams put it plainly: "We're internal-only." He said the tools are not offered to other professional teams, for reasons of competitive integrity.

The arrangement splits the product in two. Any amateur can install the app, but the work done for professional rosters is not for sale: the same organization runs a public product for solo queue players and a set of internal tools that its own staff can use and its rivals cannot.

Williams set a limit on what prediction can do at that level: even with perfect data, he said, a model predicting professional matches tops out at an accuracy of about 80 to 85 percent.

For individual players, he made a narrower claim: users who consistently follow the app's top recommendations see an average increase in win rate of about 6 percent. The figure is his, and the interview does not say how it was measured or over what period.

Where Riot draws the line

Every app of this kind lives under the rules of Riot Games, which publishes League of Legends and decides which third-party products may use its data. Riot's general policies for developers, last updated on May 29, 2025, say that products cannot create an unfair advantage for players, and that they should increase rather than decrease the variety of game decisions, such as builds, team compositions and champion choices. Products may highlight important decisions and offer several options to help players choose well, the policies say.

The same policies set the basics of how such a product may make money. A product must be registered in Riot's developer portal with an approved or acknowledged status before it can be monetized, it must keep a free tier of access for players, which may carry advertising, and paid content must be transformative, adding something of its own to Riot's data. Monetization cannot gouge players, and advertising may not appear inside Riot's own properties, such as the game itself, its loading screens or the Riot Client.

Williams called Riot's competitive advantage standard fuzzy, and said iTero takes a conservative line, staying away from features that sit in a gray area. Timers for summoner spells, the two extra abilities each player chooses before a match, which come back only after a long cooldown, were common across tools of this kind. When Riot decided it no longer wanted them, iTero removed its own, according to the interview.

The limits Williams sees

The founder described a risk that comes from outside the official channels. Tools that run artificial intelligence models on what appears on the player's monitor, reading the screen rather than using the publisher's data interface, are much harder to control, he said. He saw that as a worrying direction for competitive integrity, more so in first-person shooters than in League of Legends, and gave the example of a model that identifies where an opponent's head is on the screen.

He also named features that, in his view, cross the line into AI-assisted cheating even inside League of Legends: phantom markers placed on the map, and voice prompts that call out when to take an objective during a live game. In both cases the software would act during the live game itself, rather than helping a player prepare before it or learn after it.

On the competitive map, Williams did not expect AI coaching to close the gap between Western teams and the teams of Asia. His reasoning was economic: artificial intelligence helps every team roughly in proportion to what it invests, so teams with larger budgets gain at least as much as the others, and the relative gap may not shrink. He went as far as saying that Western teams would probably still be losing to Korea in ten years.

The .esports registry publishes Esports Observer, so the whole scene can read its news for free.