Andrew Pearson, Intelligencia Limited: “Most of What We Call AI Marketing Is 20-Year-Old Analytics Wearing a New T-Shirt”
Andrew Pearson has spent over a decade building the data systems behind some of the world’s biggest casino floors. He came from Macao (China) to speak in Cyprus at the AI Expo. After his presentation, he sat down with Marianna Konina from Reputation City (our guest editor) to explain why the AI marketing hype cycle isn’t quite what it looks like - and where the real money actually gets made.
Andrew Pearson doesn’t look like a man trying to sell you AI. He looks more like someone who’s tired of watching people buy the wrong version of it.
For over a decade, Pearson has built business intelligence and customer analytics systems for some of the largest casinos and sportsbooks on the planet - from Macau to Manila to Melbourne. As Managing Director of Intelligencia Limited, he’s spent the last nine years running data integration and AI-powered engagement projects across the Pacific Rim, while also publishing books and speaking internationally on predictive analytics, IoT and mobile marketing. He was presenting on “smart” technology in hospitality and gaming - a talk that ranged from IoT-connected hotel operations to what he calls “emotional AI” in advertising. It's a combination — hospitality, F&B, and casino gaming all in one - that maps closely onto Cyprus itself, an island with a booming hotel and restaurant sector and, with City of Dreams Mediterranean, a casino resort of its own.
But sit down with him for twenty minutes, and the first thing he’ll tell you is that most of what gets marketed today as “AI-powered marketing” isn’t new at all.
MK: Let’s start with the uncomfortable one. You build AI systems for a living, but you’ve said AI is basically a rebrand of tools that have existed for twenty or thirty years. Isn’t that a strange thing for someone in your business to admit?
Not really - it’s just honest. When I started in this industry over a decade ago, we called it analytics or business intelligence, and AI was a small subset of that. Now, AI has become this giant umbrella that swallowed everything underneath it. Logistic regression, market basket analysis, customer segmentation - techniques that are twenty, even forty years old - all get relabeled as “AI” today, even though the math hasn’t changed much. A customer churn model I could have built in 2012 gets called an AI solution in 2026. For sure, the label sells better. That doesn’t make the substance fake, but it does mean businesses need to ask what they’re actually buying, not just what it’s called.
MK: So when a marketing director tells you they want “an AI strategy,” what are they actually asking for, in your experience?
Usually they don’t fully know yet, which is fine - that’s the conversation we need to have. I map it against the customer lifecycle: reach, act, convert, engage. A chatbot that handles basic inquiries is at the reach stage. Once someone wants to join a loyalty program, that’s a different, more complex integration with the CRM system. Then there's the analysis layer — looking at who your best customers actually are. In one project, we found that women working in tech, earning around eighty thousand dollars a year, were disproportionately high-value casino customers, at least on the female side. Nobody's intuition would have guessed that combination. The real payoff shows up the moment she walks through the door: because that profile already lives in the system, the host at the entrance can recognize her on sight and immediately offer something that's actually relevant to her right then — not a generic greeting, a specific, well-timed proposition. That's the part that's genuinely powerful — not the chatbot, the pattern underneath it, and what a human on the floor does with it in real time.
MK: You mentioned emotional AI in your Cyprus talk - systems that read tone, word choice, even micro-expressions to gauge how a customer feels. Where’s the line between smart personalization and something that starts to feel manipulative?
That’s the right question to be asking, and I don’t think the industry has a settled answer yet. Technically, the capability is already here - platforms exist that translate ad creative across fifty-plus languages, adapt the imagery, dub the video, and then pull live performance data to flag when an ad is losing traction, without a human touching it. The line, for me, is whether the system is being used to genuinely serve what the customer needs, or purely to extract more spend from a moment of vulnerability. A casino using data to know a guest is a slot player who responds to competitive ranking is one thing. A system engineered to find someone’s emotional low point and sell into it is another. The technology doesn’t draw that line for you - the business does.
MK: Give us a concrete example of AI doing the “boring but valuable” work, rather than the flashy stuff.
Dynamic pricing. It’s the least glamorous example I have, and it’s one of the most profitable. Casinos adjust minimum table bets the same way Uber adjusts fares or airlines adjust ticket prices - a baccarat table might have a ten-dollar minimum on a Monday morning and a hundred-dollar minimum on a Friday night. The dealer’s salary doesn’t change. The revenue does, substantially. That’s not glamorous, it’s not “generative,” nobody’s going to put it in a keynote highlight reel - but it moves more money than most of the flashy generative content work I see businesses chasing.
MK: You’ve been doing this long enough to have made mistakes. What’s one that actually taught you something?
I’ve had one or two genuine screw-ups, and plenty more successes - but the honest answer is that analytics isn’t a guarantee, it’s a probability game across a lot of moving parts. One example that always surprises people: in loan application analysis, certain phrases that sound positive - things like “God bless you” - actually correlate negatively with repayment, based on historical data. Models pick up patterns humans wouldn’t. And they can also break. COVID disrupted models that had worked reliably for years overnight. If a business treats an AI model as something you deploy once and walk away from, that’s the mistake. It needs to keep learning, and someone needs to keep watching it.
MK: Where does a business start if it doesn’t have a data team and doesn’t want to spend a fortune?
Honestly - data first, ambition second. I’ve seen companies want to jump straight to predictive dashboards without having accurate data at the source, and the output is just confidently wrong information. Start cheap: MailChimp-level tools, open-source options, a basic CRM. And pick one internal champion - usually a CEO or CMO - because AI projects get expensive and political fast if nobody owns them. The businesses I’ve seen succeed are the ones that demand results within one or two weeks on something small, rather than a twelve-month “AI transformation” with no checkpoints.
MK: Who should actually own an AI project inside a company? IT department seems like the obvious answer.
It’s the obvious answer, and I’d argue against it - or at least against giving the IT team the final say. AI implementations touch marketing, customer service, finance and operations at once, so you need someone with authority across those departments, not just technical authority - usually a CEO or a Chief Marketing Officer, senior enough to shepherd the project through the politics of a large organization. IT needs to be involved, especially on security, but I’ve watched projects stall for months because IT was treated as the approver of every small decision rather than a partner in a few big ones. Keep them close on what matters, and at arm’s length on the rest - or you get a technically flawless system that took eighteen months to ship a feature that should have taken six weeks. “Keep your friends close and your IT closer,” to paraphrase another saying.
MK: Final one - is Cyprus actually ready for this, or is it another market chasing the buzzword?
Cyprus has real ingredients - a genuinely active startup scene, a government that’s talking seriously about digital transformation, and sectors like gaming and hospitality that are growing fast enough to need this. What I see here isn’t resistance; it’s more of a generational gap in how comfortable different teams are with the technology, which is a solvable problem, not a structural one. There’s also a telecom angle worth mentioning: with four or five competing phone providers on the island, retention should already be the priority, since acquiring a new customer costs roughly seven times more than keeping one. I’ve seen providers elsewhere solve this by having AI flag customers approaching their data limits and proactively offer a cheaper plan upgrade before an overage charge hits - a small, unglamorous fix that quietly protects the customer relationship and the bill at the same time. My advice to local businesses is the same one I’d give anywhere: don’t start with the biggest, most impressive AI idea in the room. Start with the smallest one you can prove works in two or three weeks, then build from there. 
