AI transforms customer support into a revenue driver: Interview with Harpo Lilja
Customer support in the global iGaming sector is evolving rapidly. Artificial Intelligence agents might not be able to replace human agents entirely, but they can certainly enhance their roles. By becoming trainers for AI, human agents are able to turn support into a strategic revenue generator. In this 2-part interview, Harpo Lilja, CEO of Tugi Tark, explains how AI reshapes support, creates new income streams, and improves customer experience worldwide.
New revenue streams with AI customer support agents
Customer support using AI opens novel opportunities for monetisation in iGaming. Harpo explains that AI’s speed, availability, and consistency allow operators to handle up to 80% of support tickets autonomously. Companies can also embed promotional offers within AI interactions. For example, when a player asks about shipping, the AI bot can present premium delivery options. This integration turns support into a revenue channel rather than simply a cost.
SiGMA News: How is AI transforming customer support from a cost to a value centre in iGaming?
Harpo Lilja: AI excels in accuracy, consistency, and availability. Unlike humans, AI does not suffer from mood swings. The AI agents are never sick, they never go on holiday, and they’re available 24/7.
“Human agents typically take 30 to 60 seconds to respond initially, but an AI agent responds in just a few seconds. AI can handle around 60% of tickets independently. With deeper integration, AI can manage over 80% of cases, such as automatically granting bonuses. Human agents solving tickets cost between USD$5 and $15, depending on location, while AI ticket handling costs range from USD$0.20 to $3.”
This significant cost reduction enables new revenue opportunities by embedding promotional offers within AI conversations. This way, customer support evolves from a cost centre to a monetisation channel.
Personalisation as a competitive edge in AI support
Personalised interactions give operators a market advantage. Multilingual AI agents can be configurable by region and culture, and can tailor conversations using local expressions and tones for markets as diverse as Latin America. Harpo points out that AI can track user behaviour in real-time, such as time spent on a deposit page, and proactively offer relevant assistance or cross-selling suggestions aligned to the customer’s needs.
SiGMA News: How does AI personalise support across diverse markets, especially in Latin America, with regional language differences?
Harpo Lilja: “Our AI agents can be set per market, supporting over 250 languages and can be customised per market, even by region. For Latin American markets, AI adopts local expressions familiar to players from Colombia, Mexico, or Argentina. Human trainers define the tone of voice and continuously train AI based on feedback. The system tracks player IP, account language, and location to tailor communication accurately. This helps worldwide operators avoid hiring dozens of native agents. For example, a Philippine-based support centre staffed with English-speaking agents can communicate with players worldwide, with AI handling real-time translation. This setup increases service efficiency and reduces costs.”
SiGMA News: How does AI determine regional language nuances?
Harpo Lilja: “When a chat begins, AI uses the player’s IP address, language settings, and authentication data to adjust responses. This granular approach allows operators to serve multiple markets efficiently. Instead of managing 80 agents for 10 languages, AI handles most queries, drastically lowering operational complexity and employee turnover challenges.“
Proactive automation to reduce churn
While the AI offers rapid, accurate responses 24/7, operators must monitor whether the interaction enhances or disrupts the player journey. Harpo emphasises that customer feedback loops are essential to refine AI behaviour.
SiGMA News: What role does proactive automation (like error detection in deposits) play in reducing player churn compared to conventional human-led support teams?
Harpo Lilja: “It’s a good example, you take a failed deposit, because I think that’s one of the most essential support tickets that exists. Someone tries to pay but can’t. The payment system gets an error message that something went wrong. Usually, the player might try again, try again, and then leave the site or go to customer support. Then, customer support needs to open the case, investigate, and find the issue. If not, tell the player to come back later, which equals a high risk of churn.
If Tugi is integrated into your payment system, we receive an API call from the payment system about this error. We open chat, get notified via API that this has happened for the player. The AI agent opens chat and informs the player: ‘You tried to make a deposit. It failed because of X. To solve it, you need to do this.’ This happens instantly and reduces churn significantly. You can implement it for anything. When surfing the website, players don’t know where to go. AI can see patterns and open a chat with suggestions. For example, if a player lingers three minutes on the deposit page, the AI can ask if they need help. This is more effective than conventional support.”
Compliance and traceability in AI solutions
In the current iGaming landscape, regulatory scrutiny is intensifying globally. Therefore, balancing operational efficiency with strict compliance and transparent traceability is essential for AI-driven customer support solutions.
SiGMA News: Given increasing regulatory scrutiny, how do AI-driven solutions balance efficiency with compliance and traceability in customer interactions?
Harpo Lilja: “I think computers are better overall to handle this, but not everything is AI. We use something called event sourcing, a data design pattern where every event in the system is locked. You don’t have any loss. For example, on Amazon, if you add three products to your basket, then remove one, then buy, legacy systems often see you bought two. They don’t know you removed one. This info is very interesting for player satisfaction. From a regulatory standpoint, you know all events happened with no loss. What to do with information is up to each operator, but good synergies exist for exchange between regulatory and operators when required.”
Next-generation AI agents could become key to VIP player management and retention by responding with unmatched speed and precision. Time is crucial; what’s a relevant offer today may be obsolete tomorrow. AI’s real-time player behaviour analysis and rapid adaptation capabilities make quick, tailored interaction crucial for increasing lifetime value and loyalty for the users. Using AI swiftly in a dynamic sector guarantees that operators take advantage of every opportunity before it passes them by. In the second instalment of this article, we will explore in more detail how next-generation AI agents could become central to VIP player management and retention strategies in global iGaming.
This article was first published in Spanish on 1 October 2025.
The rhythms of Cyprus change with the seasons, but the stakes remain high. Join SiGMA iGathering: Dine Drink Dance – Autumn Edition in Limassol on 17 October. From investment whispers to CIS partnerships, SiGMA iGathering is where future deals get their first handshake.
