As consumers increasingly turn to ChatGPT, Gemini and other AI assistants to research products, compare options and discover brands, retailers face a new challenge: how do you optimise for AI without losing the emotional connections that drive purchasing decisions?
That question took centre stage during a London Tech Week panel moderated by Oliver Pickup, Human-evolution Storyteller at Pickup Media, featuring Piero Sierra, Chief AI Officer at Skyscanner, Daniel Hulme, Chief AI Officer at WPP, Kipp Bodnar, Chief Marketing Officer at HubSpot, and Denise Fender, SVP of Colleague, Customer & Channel Technology at Pandora.
While discussions around AI often focus on automation and efficiency, the panel repeatedly returned to the theme of trust.
The consensus was clear, AI will transform how customers discover products and services, but businesses that focus solely on machines risk losing the human experiences that make brands memorable.
Marketing to humans is no longer enough
According to Hulme, “We now, as a marketing community, need to figure out not only how we capture the attention of human beings, but how we capture the attention of the billions of AIs that are going to make decisions for us.”
Drawing on Aristotle’s principles of persuasion, Hulme argued that brands must continue balancing credibility, emotion and logic.
While AI is highly effective at evaluating evidence and factual claims, he suggested businesses need to become far better at articulating both the value of their products and the values of their organisation.
Using Patagonia as an example, he explained that consumers increasingly choose brands because they align with their beliefs, not simply because their products perform a particular function.
In a world where AI agents may eventually compare or even purchase products on behalf of consumers, those values could become an increasingly important differentiator.
Search is changing faster than many brands realise
For Bodnar, one of the biggest changes is happening in search.
“The average search term in Google was four words,” he said. “The average search term in AI search engines is around 40 words.”
Rather than typing a few keywords, consumers are increasingly asking detailed, conversational questions that include budgets, preferences and personal requirements.
A traveller may no longer search for “hotel Covent Garden”. Instead, they might ask for a hotel in Covent Garden with a satisfaction score above 4.5 stars, fast Wi-Fi, breakfast included and easy access to public transport.
That shift means AI systems are looking for far more specific information when deciding which brands, products or services to recommend.
“It’s not enough to simply be a hotel in Covent Garden,” Bodnar explained. “AI is going to look for very specific and differentiated information.”
Historically, much of that information has remained locked away in internal systems. But in an AI-driven world, businesses increasingly need to make product details, customer reviews, satisfaction scores, service information and other differentiators publicly accessible and structured.
“If that information isn’t available, you’re just going to be left out of the search.”
Hulme added that brands should also consider how they appear inside large language models, where sources such as websites, reviews, Reddit discussions and other publicly available content increasingly influence recommendations.
The result is a move beyond traditional SEO towards AI discoverability, where success depends not only on being visible online, but on providing the depth of information AI systems need to confidently recommend a brand.
Trust remains Skyscanner’s biggest advantage
While some businesses fear AI could disrupt traditional discovery channels, Skyscanner sees opportunity in the shift.
Sierra described the evolution as a move from search to answers and now from answers to actions.
However, he urged businesses not to lose sight of the fundamentals.
“As the world makes this transition, it’s important to double down on the thing that you’re really good at,” he said.
For Skyscanner, that means continuing to provide trusted, real-time travel information.
The company scans more than 100 billion travel prices every day, helping travellers find the best options while building long-term trust with customers.
Sierra also argued that traditional SEO remains critical. AI systems still rely heavily on search engines and structured information when they need accurate or real-time data.
At the same time, businesses should prepare for customers arriving through new AI-driven channels and expecting more natural, conversational experiences when they visit websites and apps.
“Customers are going to come to your site, but they’re going to come with an expectation of natural language and memory and all the things they can do in their chat box.”
Don’t lose the brand magic
Perhaps the strongest warning came from Pandora. Fender acknowledged the opportunities AI creates to streamline shopping journeys, from product discovery through to purchase and customer support.
However, she warned that brands must be careful not to optimise exclusively for machines. “Think about the consumer,” she said. “A teenager buys a bracelet and then builds that bracelet charm by charm over a lifetime.”
For Pandora, jewellery is connected to memories, milestones and personal identity. While AI can help customers discover products more efficiently, it cannot replace the emotional significance behind many purchases.
“If you over-engineer for AI agents, you’re going to lose that brand magic.”
Fender argued that the brands that win will be those that use AI to remove friction without removing emotion. Discovery, service and transactions may become increasingly automated, but customers will still form lasting relationships with brands through experiences, memories and personal moments.
The challenge for retailers, she argued, is finding the balance between making information accessible to AI systems while preserving the experiences that create emotional connections and long-term loyalty.
As shopping journeys become increasingly automated, emotional connection may become one of the few advantages AI cannot easily replicate.
Why consumers still don’t trust AI agents
Despite growing excitement around agentic commerce, the panel agreed that widespread adoption will depend on trust.
Consumers are increasingly comfortable using AI to research products, compare prices and build recommendations, but handing over purchasing authority remains another matter.
For example in travel, AI can already help build itineraries, compare flights and suggest hotels, but allowing an AI agent to spend money independently requires a much higher level of confidence.
“There is a really big gap between the demo and what you would actually trust an agent to do,” said Sierra.
“No one is going to hand over their credit card unless they trust the brand, trust the price and trust the process behind it.”
He pointed to work being carried out by payment providers including Visa and Mastercard, alongside major AI labs, to develop safeguards around consent and spending controls.
The vision is to allow consumers to define exactly what an AI agent can purchase and how much it is allowed to spend, creating a framework that enables autonomous transactions while maintaining accountability.
Even so, the panellists cautioned against assuming the technology is already mature.
Hulme drew a comparison with the early days of big data, referencing a famous observation from behavioural economist Dan Ariely.
“Big data is like teenage sex,” he said. “Everybody talks about it. Everybody thinks everybody else is doing it. But nobody really knows what they’re doing.”
According to Hulme, AI agents risk following a similar trajectory.
“Everybody says they’re doing agents, but many people are doing it badly.”
He warned that organisations are rushing to deploy AI agents before the technology is ready.
“At the moment, an agent is a little bit like an intoxicated graduate.”
Businesses, he suggested, risk unleashing large numbers of poorly governed agents across their operations before they are capable of handling complex tasks reliably.
“It’s going to be a shit show.”
Despite the warning, Hulme remains optimistic about the long-term trajectory of the technology, arguing that today’s agents may resemble graduates, but eventually businesses will have something closer to a professor in their pocket.
Start with practical use cases
Rather than chasing futuristic visions of autonomous shopping, the panellists encouraged businesses to focus on practical applications delivering value today.
Customer support, sales prospecting, content creation and internal productivity were repeatedly highlighted as areas where AI is already proving its worth.
Bodnar argued that AI is democratising capabilities that were previously available only to large enterprises, allowing smaller businesses to automate repetitive processes and improve customer experiences without major investment.
Meanwhile, Sierra suggested businesses may see greater returns by focusing on internal productivity and operational improvements before chasing third-party AI traffic.
The panel’s final message was straightforward. Focus on trust, improve your data, make information accessible and experiment carefully.
Because even if AI increasingly influences purchasing decisions, customers will still choose brands they trust.
And those brands will still need a little bit of magic.
