An update on iTero, fundraising and the 'AI Coach'

Market Conditions

If you live a peaceful live, purposefully depriving yourself of the news, then you may not be aware that the market isn’t in the healthiest of shapes currently. It doesn’t quite feel like it yet, as in most Western countries (where a majority of you reside) the unemployment rate remains at historic lows. The only real sign of trouble brewing is seen in inflation, which although seems to have passed the worst, remains at eye-watering levels. There’s also wakes in the waters of the financial sector, a common starting-point for market contagion. Several banks have gone under, and there looks like there’s more to come (another announced yesterday).

Now, I won’t pretend to know what will happen next. It could be catastrophically bad, or a return to the normal, and anywhere in between. This is also the opinion of most sensible folk and therefore the general consensus is “We don’t know, so let’s play it safe”. This then results in tightening of purse-strings and a preference for risk-aversion. Two areas that get cut-back on in times like these are investment and marketing, as focus moves to consolidation over growth.

The iTero Latest

I provide all the above as to give context for when we talk about iTero Gaming, as these conditions are interwoven into our story. As a brief reminder (although I strongly suggest reading the full introductory article), we’re building an AI Coaching tool for competitive gamers. We’re starting with League of Legends, due to the availability of data and my own personal knowledge of the game.

In September, we launched a Drafting Tool and last month we hit 100,000 downloads. Every single day around 8,000-10,000 people will use the app. Our primary revenue source is from advertising, although last week we launched “iTero+”; a monthly subscription of $2.99 that removes the adverts and provides additional functionality from the AI. We expect to make around 5-10% of our revenue from this new stream.

Initially, we had plans to fund-raise around the region of £500,000 in order to expediate our growth plans. However, since then two major factors have changed:

  • The market conditions, as explained, meant it suddenly became a lot harder to fund raise. That meant it was going to take a lot more brute force effort to close the round, effort that could instead be spent on the product itself.
  • The AI Coach, the next step in our plan, is moving far faster than I anticipated without the need for extra resources. I’ll come back to this.

The AI Coach

This is ultimately what we have been aiming for since our inception. The ability for AI to perform something which feels almost like a real-life 1-on-1 coaching session. Although, as a caveat, I do not believe we can ever replace a real coach. Luckily, we don’t have to as we have the major advantage of being able to it at a much, much lower cost and scale to millions of students. The hardcore players will keep paying for coaches and will get the appropriate advantage that it offers.

So, here’s how it works in the simplest form:

  • We show the AI millions of games and ask it to learn what is helping players win, or more importantly what causes a loss. We focus on macro decision making, such as whether they are grouping together with their team, playing for objectives, positioning correctly, etc…
  • Once the AI has the ability to accurately predict the outcome of a game based on these decisions, we show it games from a single account. We ask it to look for regular patterns of behaviour that negatively impact the probability of them winning their games.
  • We then provide a prioritised list of improvement areas for each player. Are they smashing the laning phase but then failing to capitalise on their leads? Are they a smart, macro player who just needs to play with more restraint in the early game? Do they play safe when behind and aggressively when ahead?

To this end, we’re very close to finished. The data, which is at least 80% of the work, is basically complete apart from building a few more granular features. The AI itself works and the only required improvement is to pass more data through it. The last major hurdle is getting all this into the Cloud. Which, on paper is easy - but doing it in a way that will scale to potentially hundreds of thousands of concurrent users is not so.

The other side of this is building a frontend that looks and feels as professional as our model deserves. We want it to be intuitive to use, even for our most mathphobic users, yet still provide sophisticated features for our power users. How do people like to receive feedback? Do we want to replicate a real coach as best we can, or just provide the raw data and output for the user to interpret? There’s a lot of questions and frankly, we won’t know with a lot of them until its out there being used.

We’re about 90% through the full design of the app, including a marked improvement on the drafting side. This means it’s almost time to start building. That is a 2 month process at least, so we expect by the end of June we’ll have something to share with some trusted friends of iTero. Then it’s on to a closed Beta and finally, around September, we’ll unleash it out to the world.

How I know it’s worked

As with most products, there will be some teething problems after launch. However, once we’ve dealt with those - how will we know it’s been a success? There are countless metrics we could pick from, yet for me the only one I truly care about is the R-value. You may be with familiar with this term, since it was often quoted during the peak of covid. In our case, it refers to our natural growth rate. If it was 1.0, that means week-on-week, with no marketing efforts, we maintain the same amount of users. Anywhere greater than this shows natural growth. That is my singular focus. To build a product so valuable that it sells itself. Partially this is because we make so little money per user that a grand majority of advertising methods would cost more to acquire users than they could ever return.

But, to me it’s more than a financial calculation. It satisfies my core purpose to build something of genuine use. You can throw dollar after bad dollar on synthetically growing a companies user base. However, to achieve self-sustained growth is the pinnacle of a technological product. It’s the evidence to say that there is a market for what you are building, and that market likes the way you’re doing it.

To clarify, that doesn’t mean we won’t market the product. Of course we will. However, we’ll do that only once we prove that there is a natural growth level. Only then will we start reinvesting resources into growth, as opposed to today where it will all go into product development.

Essentially, we’ve made good progress and we have the financial freedom to continue making good progress without having to fight against the market currents in the world of fundraising.

If you aren’t already on the iTero Discord, it’s the first place you’ll hear about any news on the AI Coach. It’s also where I’ll look for beta testers:

Finally, on a personal note: this Friday I travel to London to watch my first ever in-person esports event - the League of Legends Mid-Season Invitational. I am unbelievably excited!