The AI opportunities in iGaming are huge. Platforms can already recommend which promotion to run, flag a risky session before it becomes a problem, shape a customer journey in real time, and run a dozen third-party systems under one roof. Adding a smart assistant on top of that seems like a logical step in the AI transformation process, one that should help with day-to-day work immensely. Right?
I want to back away from the hype for a moment and ask a bigger-picture question that tends to get skipped when the conversation turns to features: are you actually ready to take advantage of any of this, or are you waiting for a miracle?
What I mean is simple. AI is only as smart as the people working with it, and it can’t know the platform you operate on better than you do. It works with what you give it and answers what you ask it. If your team doesn’t know the platform, no model on top will make up for that.
As Chief Client Officer, I sit between our teams and our partners’ teams, and I watch how decisions actually get made on both sides. The pattern I keep running into has very little to do with how smart the technology is and everything to do with how well the people using it understand what they’re working with.
AI only answers the question you ask
Here’s the part that gets lost in the conversation: AI doesn’t decide anything on its own. Someone has to ask it something first. Whether that’s a prompt typed into an assistant, a rule configured in a risk engine, or a goal set for a recommendation model, there is always a human request at the start of the chain. And the quality of the output is capped by the quality of that request.
Before AI enters the picture, are you comfortable with the dashboards and reports the platform already gives you? Do you know where to look when a cohort drops off, or when you need to see which promotion performed best last month? AI’s biggest value is in how much faster and more conveniently it gets you to an answer. It doesn’t pull answers out of thin air. It works on the data you point it at, and if you don’t know where that data lives, what it measures, and how deep it goes, you can’t feed it the right inputs, and you can’t tell whether what comes back is right.
This is the gap I see most often. Operators get excited about what AI could do, add it to a platform they’ve only half explored, and then get frustrated when the results are underwhelming. AI doesn’t have a really big impact on any numbers; they conclude it isn’t that useful, and the tool gets blamed for a problem that started with the question.
Any AI opportunity gets you exactly as far as you understand what to request of it. And you cannot request what you’ve never dug into.
Why the hood stays closed
I don’t think the issue stems from laziness. Many operator teams are stretched thin, running multiple markets, and dealing with a stack that has grown by acquisition and necessity rather than design. Learning the platform properly feels like a cost with no immediate return, especially when there’s a vendor on the other end who can just answer the question for you.
And that’s the habit I spend a lot of my time trying to break, on both sides. Remote, global teams make it very easy to say “I’ll check with the provider” or “let me get back to you” about things that could be answered in five minutes by someone who actually knows the product. Every one of those deferrals is a small signal that the team is operating a system it doesn’t fully own. The knowledge sits in one person’s head, or at the vendor’s, instead of in a shared system that the whole team, and eventually the AI, can draw on. Multiply that by a year, and you get an operator who has paid for a sophisticated platform and uses a fraction of it.
Now put AI on top of that. The team that defers the simple questions will defer the AI questions too. They’ll accept the first recommendation because they can’t evaluate it, or ignore it because they don’t trust it, and neither outcome has anything to do with how good the model is. The cost of not knowing your platform used to be inefficiency. With AI in the loop, it becomes a ceiling on everything the technology could have done for you.
Who owes what
It would be convenient for me to stop here and say the problem sits with operators, but it doesn’t. If a partner can’t ask good questions of our platform, that’s partly on us, and I treat it that way.
The provider’s half of the job is to make the platform knowable. Documentation people actually read, not a wiki nobody opens. Honesty about where the limits are, so a partner isn’t guessing which levers exist. Onboarding that teaches the logic behind the system, not just the buttons.
And answering the question in the room. When a partner asks something, the right response is the answer, or “I’ll find out and tell you by three o’clock,” not a ticket that resurfaces in a week. We try to hold our own teams to that, because every fast, specific answer is a lesson in what the platform can do, and those lessons add up to a partner who eventually doesn’t need to ask.
The operator’s half is simpler to state and harder to do: put in the time. Make sure someone on your team can explain what the platform does today, the real version, with the configuration you’re actually running and the features you’ve paid for and never switched on. Bring that person into the room when decisions get made. Mutual ownership is something I talk about a lot, because I think it’s the only model that works in a partnership this technical. The provider can’t own the operator’s outcomes, and the operator can’t own the provider’s roadmap. But each side can own its part of the knowledge, and when both sides do, the technology stops being a miracle you hope for and starts being a tool you use.
Take ownership of the platform powering your operations
Intelligence in the platform doesn’t substitute for intelligence about the platform. The operators who get the most out of AI over the next few years, I think, will be the ones who know their stack well enough to ask it the right questions.
That’s the readiness I’m asking about: whether your team has done the unglamorous work of understanding what they’re running before they ask AI to think for them.
Luck doesn’t build a winning operation, and neither does an algorithm on its own. Champions know their equipment. Everyone else is waiting for a miracle.
The AI opportunities in iGaming are huge. Platforms can already recommend which promotion to run, flag a risky session before it becomes a problem, shape a customer journey in real time, and run a dozen third-party systems under one roof. Adding a smart assistant on top of that seems like a logical step in the