Loyalty Circle London asked senior leaders what happens when AI agents do the shopping
Comarch Loyalty brought senior loyalty leaders to KOL, a one-Michelin-star restaurant in London, for a lunch that was built around one question: what does your program look like to a member who isn’t human?
After São Paulo, Singapore and Mexico City, the Loyalty Circle series made its next stop in London. The theme was “Upgrading to Unstoppable AI”. The setting was KOL on Seymour Street. The conversation was candid, and more than one guest left saying “I’m taking that back to my team.”
Who sat around the table
Loyalty Circle is built on one idea: small groups, senior people, no sales pitch. In London, around the table sat leaders from major UK and international brands, joined by seasoned loyalty executive. Retail, banking, telecoms, hospitality, travel and sport, all in one room, all asking the same question: where does AI actually pay off in loyalty?
The conversation was led by Hunter Murray and Philip Shelper of Loyalty & Reward Co, the global loyalty consultancy, together with Łukasz Dubiel, Consulting Director, Loyalty Solutions at Comarch Loyalty. Comarch Loyalty was also represented by Tsvetozar Aleksandrov, Loyalty & Customer Experience Strategy Advisor, and Edward Swain, VP Sales, Central, Eastern & Northern Europe.
Three waves of AI, and one uncomfortable gap
The discussion opened with a simple frame: AI in loyalty has arrived in three waves. Rule-based segmentation. Predictive machine learning. And now generative and agentic AI. None of them replaced the one before. The best programs in 2026 run all three, and the skill is in picking the right one for each problem.
Then the room took on four questions:
- Adoption is almost universal, but still shallow. Why do most programs stop at content creation, and what does it take to go deeper?
- Why do investments in data foundations so often deliver better returns than investments in AI itself?
- What can we learn from brands already using predictive AI for communications, personalization, pricing and next best action?
- Can AI be taught to show empathy, and know when not to send the offer?
The room agreed on one thing: whether to use AI is settled. Where it sits, who owns it and how it is governed are not.
Your next member may not be a person
Then came the question that got the room talking the loudest. What happens when your members stop shopping and start delegating?
New personal AI agents can already browse, compare and buy on a customer’s behalf. Accenture’s January 2026 survey of more than 25,000 consumers found that around three in four are open to the idea, as long as they stay in control. Which means you can lose the sale before the human ever presses “buy”, simply because your program never made the agent’s shortlist.
Together we explored:
- Which traditional loyalty mechanics weaken when an agent makes the choice, from gamification to breakage-led design?
- How can brands stay findable, readable, trusted, actionable and preferred in the eyes of an AI agent?
- How do you design one program for two audiences, the human member and the AI agent?
- Will platforms capture the customer relationship, and how can a consented, first-party loyalty relationship become a brand’s best defence?
The takeaway: keep the human experience rich, and build a clean, machine-readable value layer underneath it. The loyalty account may soon be the best way to tell an authorised agent acting for a known member from an anonymous bot.
