When AI agents choose, what's left of loyalty?
The hold music is changing tune.

Picture this. Your phone rings. It's your AI agent: “Hi, I have the energy company on hold for you. I renegotiated your contract and you'll save $500 a year. Shall I put you through so you can confirm?” You say yes. Thirty seconds later, it's done.
This actually happened to a relative of Steven Van Belleghem in the US, using Meta's new AI agent, Muse. For the customer, it's a small miracle. For the energy company, it's a wake-up call (pun intended). After decades of “your call is important to us”, it's now the one waiting on hold.
Suddenly, customers are no longer the ones choosing. Their agents are. What's left of loyalty then?
The end of customer inertia
Let's be honest about what “loyalty” often means: customers who can't be bothered to switch. Telcos, energy suppliers and insurers have quietly built their business on it.

People stay not because they love their provider, but because changing takes time and effort. (If you're still on the mobile plan you signed up for years ago, you know exactly what we mean.) Steven Van Belleghem calls what Muse does "customer inertia being released": the friction that kept customers in place simply disappears.
The new generation of AI agents does exactly that. Muse launched with a simple promise: upload your telco contract, your energy contract and your insurance policies, and it will renegotiate them for you. It can even make phone calls on your behalf. According to Steven, the app was downloaded faster in its first month than ChatGPT, and the number one use case is shopping.
In Silicon Valley, the startup Instinct offers something similar via WhatsApp. It books restaurant tables by checking a website a hundred times a day until a spot opens up, and it drafts dozens of emails in minutes.
The consequences are already visible:
- Commodity markets are under attack. Where customers don't really care which brand they use, an agent simply picks the best deal. Investors noticed: share prices of US insurers and telcos dipped as Muse took off.
- Platforms are picking sides. Amazon blocked the agents, fearing it would become a mere back-end and lose its advertising business, because agents don't look at ads. Shopify went the other way and declared itself radically agent-friendly, hoping to win business Amazon turns away.
- Banks have reason to worry. An agent instructed to keep moving your money to wherever the interest rate is highest could, at scale, trigger something close to a bank run.
More than ten million people set up Muse in its first six weeks. Now imagine it reaching the scale of ChatGPT. Meta certainly does: it's already building capacity to serve over a billion people.
The last moat is your brand
A billion agents, each comparing every offer in seconds. Is there anything left that a machine can't commoditise? Steven recently returned from our learning tour through Seattle and San Francisco, and one conclusion came up with almost every speaker there: the last moat is your brand.
Why? In San Francisco today, a technological advantage over your competitors lasts about 48 hours. Build something that feels groundbreaking, and two days later others have it too. In other words, your competitive edge now has roughly the shelf life of a croissant. Of course, you still need to be in that race, but you can't win on technology alone anymore.
And then there's a less obvious reason. We tend to think of AI recommendations as cold and rational, but they're not. When a language model recommends a brand, it draws on what people write online: reviews, feedback, Reddit posts. And people write those emotionally. So, ironically, a machine's choice ends up being far more emotional than rational.
That turns a lot of marketing logic on its head. Many companies still lead with products, features and specs. But what drives an agent's recommendation is how customers feel about you, and how they put that into words.
When Steven discussed his tour with fellow nexxworks partners Peter Hinssen and Pascal Coppens on Radar (our podcast), it was one of his key conclusions: in a world where machines make the comparisons, emotion becomes the deciding factor.
Justice for the cheese sandwich
So where do those emotions come from? Often from something surprisingly small. Take the Dutch cheese sandwich riot. (Yes, you're in for a treat.)
KLM was one of the few European airlines that still served economy passengers a free snack: a simple sandwich with Dutch cheese and a bit of butter. Recently, KLM decided to start charging six euros for it. The Netherlands erupted. A hashtag, #justiceforbroodjekaas, took off. KLM responded with a mock funeral video for the sandwich, which only made people angrier.
Why would anyone care this much about a cheese sandwich? Because it was more than that: it was a ritual.
In his new book Deep Loyalty, Steven Van Belleghem highlights the power of rituals in building lasting customer relationships. Think of the rituals in your own family: Christmas Eve, a yearly trip, a Sunday breakfast. You look forward to them, you enjoy the moment, and they become stories that last a generation.

Some brands manage to create the same thing. Disney has its fireworks at the end of the day. IKEA has its meatballs. KLM has its collectible Delft Blue houses in intercontinental business class, which people proudly display in their offices. And, until recently, it had its cheese sandwich.
KLM will have its reasons, and cost savings is likely one of them. But killing a ritual chips away at precisely the kind of loyalty a price comparison can't touch. Rituals leave an emotional residue that ends up in reviews, in conversations, and ultimately in the data an AI agent uses to decide which brand to recommend.
What this means for your company
AI agents won't kill loyalty. They will expose which kind of loyalty you actually have. Loyalty built on inertia will evaporate. Loyalty built on emotion will be amplified, because emotion is exactly what agents read. Five questions you can't afford to postpone:
- How much of your customer base stays out of habit? Be honest about it. If an agent offered your customers a better deal tomorrow and handled the switch for them, how many would stay?
- What do customers write about you? Your carefully crafted product pages and spec sheets matter less when recommendations are built on reviews and conversations. The emotional story customers tell is your new shelf placement.
- Which rituals do you have, and which are you about to cut? Before you remove that small free extra to save on costs, ask whether it's a line item or a ritual. The savings will show up in a spreadsheet. The damage will show up on Reddit.
- Will you block agents or welcome them? Amazon and Shopify made opposite bets. There's no universal right answer, but there is a wrong one: not choosing at all.
- How will you earn the trust of agents and their owners? On Radar, Pascal Coppens pointed to Ant Group in China, which promised to refund customers if its agent placed a wrong order. As Peter Hinssen added, credit card companies did the same decades ago. Amazon did it in its early days with lost packages. New behaviour always needs a safety net before it takes off.
Loyalty in the age of agents

You have to admit, there's a certain poetic justice in that phone call from the agent. For years, customers waited on hold for telcos and energy companies. Now the roles are reversed. And for many companies, the hold music is starting to sound like the Jaws theme. As Steven put it: it's payback time for customers.
The companies that come out on top won't be the ones with the cleverest technology, because that advantage lasts about 48 hours. They'll be the ones customers feel something for, strongly enough that it shows up in what they share online, and ultimately in what their agents choose.
Want to stay ahead of the Day After Tomorrow? Tune in to Radar, the nexxworks podcast in which Steven Van Belleghem, Peter Hinssen and Pascal Coppens share what they see on their travels between Silicon Valley, China and Europe. This article is based on the episode featured below.
Check out more episodes (and don't forget to subscribe 🙏) on Apple Podcasts or Spotify.

