What if AI makes your customer experience worse, not better?

CX Matters is the Hello Customer podcast about what makes customers stay. In this episode, keynote speaker Steven Van Belleghem and Michel Stevens of CXM Academy join Bram De Vos to explore what AI does to customer experience and to trust.

The conversation covers the AI tools the guests use themselves, why customer service still lags three and a half years after ChatGPT, why new technology takes a decade longer than expected, agents talking to agents, and why the answer to all of it is the oldest question in marketing.

Steven Van BelleghemCX author & keynote speaker

International keynote speaker and author of bestselling books on customer experience, including When Digital Becomes Human and Customers the Day After Tomorrow.

LinkedIn

Michel StevensCX expert & course director

Course director at CXM Academy, partner at goCX and host of the Table 7 podcast. Works with organisations across Europe on customer-centric transformation.

LinkedIn

Bram De VosCEO, Hello Customer

Your host of CX Matters, the webinar and podcast series powered by Hello Customer, where every episode looks at one idea: the way you treat your customers shapes the way your business grows.

LinkedIn

Does AI make customer experience easier, or more fragile?

As AI becomes central to customer experience, does that experience become easier to build, or more fragile as decisions get automated?

AI is changing how organisations process information and make decisions, and it is changing how customers do business with companies. It cuts both ways. This episode looks at what that means for the connection between companies and their customers, and for the trust that holds it together.

How CX leaders use AI themselves

Which AI tools do people working in customer experience actually use, and for what?

The answers are practical and personal. ChatGPT works as a sous chef at home, a travel planner, a translator and a research tool. A Perplexity agent sends a briefing at eight every morning with the latest CX news on chosen topics. Gemini serves as an "evil twin" that argues the other side of an idea, because people are so caught up in their bubbles and echo chambers that they don't see what they don't see. Manus fetches information and prepares morning briefings, more like an agent than a tool. And Claude is the newest favourite, impressive enough to make one long-time ChatGPT user unfaithful.

One choice stands out. Writing a book is a way to learn, so one author writes it himself and uses AI only as a research assistant and editor: does this chapter make sense, is evidence missing, is there research that could help? Meanwhile the switching between tools never stops, because the race keeps changing who is ahead.

Still amazed, and still disappointed

What still surprises people about AI, and what has been a disappointment?

The most delightful uses are often human-driven. A mother who owns a Thermomix photographs an ordinary recipe, sends it to ChatGPT and asks for a Thermomix version. The coolest uses come from people trying things out.

The disappointment is customer service. Three and a half years after ChatGPT arrived, you would expect most companies to have rebuilt their service around chatbots of that quality, yet no large incumbent seems to have done so. The problem sits in the back office of large organisations: data that is not ready, slow processes, and fear.

Part of that fear traces back to the Air Canada case, in which the airline's chatbot gave a customer wrong information about a bereavement fare refund and a Canadian tribunal held the airline responsible for what its chatbot said. In the episode's account, that case is where the "human in the loop" reflex comes from.

Why big companies struggle to adopt AI

Why are large companies so slow to adopt AI in customer service?

Large organisations are trying to turn a racehorse into a Formula 1 car. They carry legacy data systems, history and old processes, and they try to bolt something new on top. They know what is, and they have not yet figured out what can be.

In the European Union, regulation adds to the hesitation. The rules come with good intentions, and they also make companies worry about data protection before they try anything, which risks Europe missing the boat. The United States is not much faster, and China is probably further ahead because companies there implement faster and more pragmatically. The technology is there. The adoption isn't.

Why adoption takes a decade longer

How fast will AI really change customer experience?

New technology usually takes about a decade longer than anyone expects. Google's driverless cars were already driving around Silicon Valley in 2013, and everyone assumed they would be everywhere a year later; today they operate in certain US cities. Google Glass is only starting to see real successors 14 years on. Even after the iPhone launched in 2007, it took seven or eight years before most of the apps on people's home screens existed. AI in large organisations, customer service included, will follow a similar curve.

Things need to become boring before they become interesting. Chess computers that beat grandmasters once looked like superintelligence; today they are well-understood algorithms. Once a technology is boring, its boundaries are known: what it can do, where it fails and where it fits. Regulation, rules and ethical debates are part of getting there.

Some things still move very fast. Between September and January, the talk in San Francisco went from barely mentioning vibe coding to developers saying they no longer write code by hand and work ten times faster, after a new Claude model arrived. Early adopters changed the way they work within weeks, and they are now teaching everyone else.

Will AI widen the distance to the customer?

What will more AI and automation do to the connection between companies and customers?

The distance between customers and brands will grow. Adoption of AI tools will be massive, and a Journal of Marketing study suggests one reason: people who know less about AI are more receptive to it, partly because it feels almost magical to them. Customers will use those tools to find cheaper alternatives more easily than ever.

At the same time, great customer service is becoming a commodity, and many companies did not invest enough in differentiation over the past two decades because they were hooked on short-term results. Together, that is an enormous threat to the customer relationship. If companies are not careful, their first AI investments will push customers further away.

Agents talking to agents, and the case for deep loyalty

How does AI increase that distance?

Customers will stop talking to companies directly and talk to their agents instead, and soon the customer's own agent will talk to the company's agent. That is a long way from a customer walking in and buying directly. Most companies are putting all their eggs in the basket of transactional perfection. They need it, yet it will not be what sets them apart.

The relationship with the customer will be the differentiator. Deep loyalty is about belonging: customers who want to be part of the group and want to show the world they belong to a brand. That makes the winning strategy more emotional. The companies that focus on emotional connection will win over the ones that focus only on operational excellence. Deep Loyalty is also the title of Steven Van Belleghem's new book.

The emotional and the social layer

What protects a brand when AI makes every competitor equally efficient?

Most companies now understand what customers want on a functional level. The emotional and social layers are where they can still stand apart. A community around a brand, with an emotional connection, is a strong defensive line that is very hard to copy. If a technological wave makes a rival's product better, that loyalty buys time.

AI also raises the bar for human service. ChatGPT is friendly and never gets frustrated, even when asked the same thing seven times. Ten years ago customers rejected "fake" friendliness; today most would choose programmed friendliness over none at all. So in a physical store, people have to be better than the friendly machine.

That requires knowing how the brand sounds. Many chatbots return the requested information and nothing more. One that answers "do you have this shoe in size nine?" with "yes, and the black ones are going fast" turns a query into a conversation that builds on what the customer said.

Customer centricity must also be seen to be done

Where does customer feedback fit into this?

Companies have learned that feedback contains functional lessons about what to fix, and increasingly they see its emotional layer too. The third layer, the social one, is the one most of them forget.

Bram De Vos, who has a legal background, borrows a line from law here. In 1923 the English Lord Chief Justice ruled that justice should not only be done, but should be seen to be done. Applied to customers, the principle means showing people what changed because of their feedback, and AI can help companies make that visible. Companies that don't will blend into the mediocre mass.

When nobody believes it is real

What happens to trust when nobody can tell what is real anymore?

Intent matters as much as efficiency. In one case discussed in the episode, a British Member of Parliament who answered constituents' emails with ChatGPT faced a huge backlash. Data protection was secondary. People were upset about the lack of effort in a human dialogue. In another, a business leader running automated outreach on social networks refused to answer people who replied to his bot, so as not to interfere with the automated flow. That is a dark place to end up.

The opposite problem is growing too. Something impressive used to make people say "wow"; now they ask "is this real?" Someone who answers every email within 24 hours gets replies addressed to their chatbot. Videos of acrobats are dismissed as AI. If you are exceptional, people assume it is fake. Expectations get inflated as well: tourists who came to Amsterdam for the Christmas lights they had seen online complained that the city looked nothing like the pictures.

Curation and visible effort are part of the answer. Readers gravitate towards work that feels genuine and lose interest as soon as a text sounds machine-written. Human creativity is under threat and it still matters, because AI only knows what it has learned from human work. It stands on the shoulders of giants.

AI will not solve your problems

What is the biggest mistake companies will make with AI in customer experience?

Expecting AI to solve their problems. Achieving anything still takes hard work and dedication, and the old rule applies: garbage in, garbage out. AI amplifies whatever is already there, including bad data and bad processes. It can make things faster or cheaper, but it will not change the fundamentals of an organisation.

The biggest danger is mediocrity: average service, average communication and average attempts to surprise people, because everyone follows the same AI towards efficiency. That leaves a company exposed to the price players. The solution is to know your real strengths and differentiators, and to go all in on them.

It is also an old solution. In a time obsessed with innovation, the things that do not change hold a lot of the answers. What sets a company apart is the oldest question in marketing, and it is worth revisiting before getting lost in every new tool, however fascinating.

This article is an edited summary of the conversation. For every word, watch the full episode in the video above, or click a section title to jump to that moment.