We sat down with Chris Jones, Managing Director at PSE Consulting, a specialist payments consultancy with over 30 years of industry expertise. With a career spanning Accenture and major fintech, retail, and payments clients, Chris is one of the UK and EU's most respected voices on payments innovation. Today, we're talking about a seismic shift unfolding in e-commerce, AI agents that don't just recommend products, they're reshaping the entire discovery journey before a single transaction takes place. And new research suggests the real story isn't who builds the best chatbot, it's who ends up holding the customer's trust.
For our audience who might not be deep in the payments world, how would you explain what's happening with Anthropic and ChatGPT in plain English?
Over the past year, AI tools like ChatGPT have moved well beyond being a novelty for recommendations. People are genuinely using them as the starting point for shopping decisions, asking them to compare products, suggest options and narrow down a shortlist before they ever land on a retailer's website.
We commissioned a large piece of research into exactly this, surveying over 4,000 consumers across the UK, US, France and Germany who already shop this way. What we found is that AI has firmly established itself as the new front door to discovery. Price comparison is the single biggest reason people turn to these tools in the first place.
What's interesting is what AI has not displaced, which is the actual transaction. Consumers are very willing to let AI narrow their choices, but when it comes to completing the purchase, they still want a brand they recognise and a seller they trust. So rather than AI replacing the shopping experience end to end, we're seeing a much more interesting split: AI owns the discovery and comparison phase and existing merchant infrastructure still owns everything that happens after that from payment to logistics and support.
Various AI platforms have experimented with monetising these interactions, charging fees or favouring sponsored options. Do you think AI platforms charging for influence over a purchase feels different to consumers than a standard transaction fee?
It absolutely does, and our research backs that up directly. We asked consumers how they'd feel if an AI assistant was free because it was being paid to promote certain brands or products, and 40% said it would reduce their trust in the recommendations.
But here's the more interesting part: when we then asked people to choose directly between a free AI tool that's commercially influenced versus a paid tool that's fully impartial, 43% still chose the free, advertising-supported option. People are sceptical of the model in the abstract, but in practice they make the same trade-off they've been making with search engines and social media for the last two decades.
So the psychological distinction is real, but it isn't a dealbreaker. What does matter enormously is transparency. Consumers are far more comfortable with commercial influence they can see and understand than with influence that's hidden or unclear.
What does it actually mean for AI to "own discovery" while merchants and marketplaces own everything else, and why does that split matter so much?
Think of it as two genuinely distinct stages in the shopping journey and our research shows consumers experience it that way too. The first stage is discovery: narrowing down a huge range of options based on price, recommendations and comparison. That's where AI is becoming dominant, and very quickly. The second stage is execution: actually completing the purchase, dealing with payment, delivery, returns and customer service if something goes wrong. That's still firmly the domain of established merchants and marketplaces.
What our data shows clearly is that consumers don't want these two roles to merge. Only 10% of people we surveyed said they'd want an AI assistant built directly into a specific retail platform. The overwhelming preference, by a margin of four to one, is for an independent AI tool that can search and compare across multiple sellers rather than one that's tied to a single retailer's interests.
That tells you something important about where value sits. The AI layer is becoming the new front door, but it isn't trying to become the shop itself and frankly consumers don't want it to.
Various studies have pointed to significant conversion uplifts from AI-driven shopping journeys. How confident are you in figures like that, and what's actually driving the improvement?
I'd treat any single conversion statistic with a healthy degree of caution at this stage, simply because the channel is still so new and the methodologies vary enormously between studies. But directionally, it makes complete sense and our data explains why.
By the time someone reaches a retailer through an AI-driven journey, a huge amount of the traditional decision-making process has already happened in the conversation with the assistant. The AI has helped them compare options and narrow down a shortlist. In our findings, price comparison was the single biggest reason people use AI for shopping in the first place, which tells you these are highly motivated shoppers.
What's equally important and this is the bit that often gets missed, is that the assistant hasn't actually closed the sale by that point. Our data shows 89% of consumers still want to recognise the seller's brand and 92% say reviews matter before they'll commit. So the uplift in conversion is real, but it's earned through trust signals that already exist.
Some major platforms have experimented with waiving fees for AI-driven commerce while others have tested charging. Do you think free models are a temporary strategy, or could it stay that way long term?
I'd be very surprised if free models lasted indefinitely across the board. Right now, the priority for most AI platforms is adoption: getting consumers into the habit of using these tools for everyday decisions. Waiving fees, or not charging at all, lowers the barrier and helps build that habit quickly.
But our research actually gives a really useful steer on what happens next, because we asked consumers directly which model they'd prefer. 43% chose a free, advertising-supported assistant over a fully impartial paid one, which only 27% preferred. So there's a clear commercial path here that doesn't require platforms to charge transaction fees at all. They can build a sustainable advertising-funded model, the same one that's underpinned search and social media for years, and consumers have already shown they're willing to accept that trade-off.
The bigger question is whether it arrives as advertising, as a fee on the transaction, or as something else entirely like an affiliate-style model. I suspect we'll see all three coexisting, serving different segments of the market. It's worth remembering we've actually been here before. Google has run a two-tier model for years: free organic search funded by advertising, sitting alongside their enterprise products which are paid for by a standard subscription model. I'd expect agentic commerce to land in a very similar place, a free, advertising-supported tier for the mainstream and a paid, more curated tier for those willing to pay for it.
Who do you think will ultimately control the customer relationship in agentic commerce, the AI platforms, the merchants, or someone else?
This is really the central question that we set out to answer and the data points to something more balanced than a straightforward winner-takes-all outcome.
AI platforms clearly have the advantage at the discovery stage. Consumers are building genuine habits around these tools, and our research shows price comparison and AI recommendations are now meaningfully shaping that first stage of the journey. But merchants and marketplaces are not losing ground at the point of actual purchase. In fact, we found 90% of consumers expect to use retail marketplaces the same or more as AI adoption grows, with only a small minority expecting to use them less.
There's also a hard economic argument for why this split makes sense. Running an AI agent at the very top of the funnel is expensive, and the vast majority of those interactions, comparisons, questions, abandoned shortlists, never convert into anything. It makes far more sense for that cost to sit with the AI providers, who are built to operate at that kind of scale and volume, rather than with merchants. What merchants should care about is the other side of that equation: being willing to pay for the recommendations that do convert, so that a good proportion of those interactions end up completing on their platform. That's a much more sustainable model than every merchant trying to build and fund their own agent.
AI-driven shopping behaviour is growing very quickly. Are merchants actually aware of this shift, or are most still catching up?
Most are still catching up and that's exactly what motivated us to commission this research in the first place. The very largest retailers and digital-native brands are already adapting, because they can see consumer behaviour changing in real time. But for a lot of mid-market merchants, AI-driven shopping still feels like a future trend rather than something happening to their business right now.
The numbers say otherwise. We found that 89% of consumers consider brand recognition essential when acting on an AI recommendation and 92% say the same about reviews. What that tells merchants is that the brand relationship and trust signals they should be building today are exactly the ones that will carry over into the agentic world as it matures. This isn't a new task on top of everything else, it's the same job merchants have always had to do, recognisable branding, strong reviews, consistent reputation. What's changed is the urgency. If you haven't built that trust by the time AI becomes the default starting point for shopping, you're already behind.
How do you see AI shopping adoption playing out differently across the markets you researched, particularly the US versus Europe?
There are some genuinely interesting differences. The US is the most price-driven market we surveyed, with 37% of consumers citing price as the single biggest factor in their AI shopping decisions. The UK, by contrast, is the most brand-reliant market, with the highest proportion of consumers saying they need to recognise the seller before they'll act on a recommendation and the most attached to loyalty schemes specifically.
Germany stood out as the most open to AI-driven recommendations generally, while France had the most evenly distributed set of decision factors of any market we looked at, suggesting French consumers are placing a slightly higher level of trust in the AI's own judgement than elsewhere.
What that tells me is there's no single European or US playbook here. Merchants operating across multiple markets need to understand that the trust signals driving conversion in Frankfurt are not necessarily the same ones driving conversion in London or New York.
Regulation is another area where Europe and the US look quite different, and it's worth flagging because the story has actually flipped. When Strong Customer Authentication came in, a lot of merchants saw it as friction, an extra step that risked losing the sale. In an agentic world, it's starting to look like exactly the opposite. SCA gives EU merchants a genuine advantage, because it shifts a meaningful amount of risk exposure in agentic transactions onto issuers rather than the merchant. What was once viewed as a compliance burden may turn out to be one of the more useful protections European merchants have as AI agents start initiating purchases on a consumer's behalf.
Merchants and platforms are increasingly trying to keep their own checkout infrastructure in the loop even when AI is driving discovery. Is that a viable long-term position, or just buying time?
I think the more interesting answer is that this isn't really a defensive position at all, it's a symbiotic one. Independent AI agents and platform-embedded AI are going to do genuinely different jobs, and both will coexist. An independent agent is good at comparing across the whole market and bringing a highly motivated shopper to your door. A platform's own embedded AI is better placed to refine that choice once the customer has already arrived, helping them pick the right size, the right room, the right flight. Each side gets better at what it's actually good at, without taking on the cost or risk of the other's job. Merchants keeping their checkout in the loop isn't about resisting AI, it's about making sure they're the ones executing the transaction once an agent, independent or embedded, has done its part.
Merchants protecting their checkout and their direct customer relationship is aligned with what the data tells us consumers actually want. The risk isn't that AI takes over checkout against consumers' wishes. It's that merchants who don't make themselves visible and trustworthy at the discovery stage never get the chance to complete that handoff at all.
You've spent 25 years advising merchants and fintechs. What's your honest advice to a mid-size merchant sitting on the fence about all of this right now?
Treat it the way you should have treated mobile commerce or social commerce in their early days: as a genuine behavioural shift. My advice based on what we found is to focus less on the AI hype and more on the fundamentals that are still driving conversion. Make sure your brand is recognisable, your reviews are visible and your product data is accurate and complete, because those are precisely the signals our research shows consumers rely on once an AI assistant has narrowed their options. AI is changing how customers arrive at your business but clearly it hasn't changed what convinces them to buy once they're there.
You don't need a sweeping AI strategy overnight but you do need to start understanding how and where your products are being surfaced in these AI-driven journeys, because that visibility question is only going to become more important. And don't roll this out blind. Put guardrails in place and start small, testing on a limited scale first so you can genuinely understand the downside risks of this new customer channel before you commit to it at any real volume.
What's the one thing about agentic commerce that you think the industry is still seriously underestimating?
That control of the AI layer doesn't equal control of commerce. There's a real temptation in this industry to assume that whoever builds the best AI agent wins the entire customer relationship. Our research strongly suggests otherwise. AI is reshaping where discovery begins, but the systems that actually make commerce work, trust signals, marketplaces, payment infrastructure, remain firmly in place and arguably become more important.
The bigger shift the industry needs to prepare for is that commercial influence over the purchasing decision is moving and redistributing far faster than the infrastructure underneath it, the authentication, the liability frameworks, the dispute resolution processes, can keep up. That's the conversation payment providers need to be having right now.
About the author
Chris manages PSE Consulting’s business and is well known in the UK and EU for his regular insights into payments innovation. He has spent the last 20+ years leading assignments for major clients during his time at PSE and Accenture. His specialisations include customer proposition development, market entry strategies and enterprise value creation. Chris is a highly effective communicator, with very strong analytical skills and able to deliver recommendations to C-level clients and company Boards. Chris regularly supports major enterprises, corporates and global digital merchant’s payments bringing new market perspectives and identifying fresh opportunities for innovation and expansion. He has also delivered payment assignments for Fintechs, banks and processors on topics such as: regulatory impacts, new acceptance methods, open banking, BNPL, gateway/acquirer convergence, and opportunities for M&A and inorganic growth.
About PSE Consulting
PSE Consulting is a leading global provider of payment advisory services to players across the payments landscape. PSE’s expertise has enabled it to deliver actionable market insights and operational optimisation to senior payments leaders for over 30 years.
To learn more, visit: https://pseconsulting.com/
