How LLMs Are Redefining Customer Loyalty

For years, AI in loyalty programs worked behind the scenes. Machine learning calculated customer lifetime value, predicted churn, and flagged fraud. Members rarely saw those models at work, but the brands running the programs saw the results every day, in retention, in fraud caught early, and in offers that landed. That work still matters. It's just not the whole job anymore.

Large language models change how people work with the platform itself. Jerzy Biernacki, Comarch Loyalty's Chief AI Officer, puts it simply: the payoff isn't faster math. It's a different relationship between brands, marketers, and the people actually redeeming points.

What changes for marketers and members

Marketers used to need a dedicated program manager and a few weeks to get a multi-tiered campaign live. Someone had to touch the rule engine every time the terms changed. Now, a marketing team can describe the promotion they want, in plain language, and adjust it themselves the same day.

Members get something too. Instead of digging through terms and conditions or waiting on hold, they can just ask: what am I earning, what am I missing, how close am I to the next tier. And get an answer on the spot.

Personalization that's actually personal

Mass segments and one-size-fits-all messaging don't hold up against that. Real personalization in loyalty takes both kinds of AI. Machine learning works out which offer and which reward fit each member, through next best offer and reward recommendations. LLMs handle the words, writing each message in a tone that fits the person reading it. Together, they aim at the person actually redeeming the points, not a segment of ten thousand people who happen to share a demographic.

Get that right and the program stops feeling like a rewards program. It starts feeling like the brand actually knows who's on the other end.

Watch Jerzy Biernacki explain it in full:

Transcription

AI is often used as a buzzword, but actually, AI is nothing new. We had machine learning models running for years inside the platform — like customer lifetime value, churn prediction, product recommendations, and also fraud detection.

Now we're approaching a phase where AI, especially large language models, really embeds itself into how both marketers and members interact with loyalty platforms.

For marketers, it means an easier way to configure and manage even the most complex loyalty programs. And for the customers, for the members, it means a way to easily navigate the loyalty program, on their side, to get personalized paths to rewards, personalized recommendations, and assistance.