Consumers favour independent AI shopping assistants over AI tools embedded within retail, travel and delivery platforms for product discovery, but continue to rely on marketplaces to complete transactions, according to new international research from PSE Consulting.
A survey of 4,250 consumers across the UK, US, France and Germany who use AI for online shopping found that 74% prefer an independent AI shopping assistant, while just 10% favour AI built directly into retail, travel or delivery platforms.
The research is based on a structured online survey conducted by market research firm OnePoll on behalf of PSE Consulting. Respondents were adults who already use AI for online shopping and were recruited across four major Western markets. The study was fielded between 5–18 March 2026, with participants completing a paid questionnaire under double opt-in conditions in line with the Market Research Society’s code of conduct.
The findings suggest consumers increasingly want AI to search, compare and navigate choice across the internet, rather than relying on AI operating within a single retailer or marketplace ecosystem.
The results align with broader signals from industry research on the rapid emergence of AI in commerce. It coincides with another industry research we earlier covered. Research from Merkle, developed exclusively for the Shoptalk conference, found that 37% of US consumers were using AI tools at least sometimes or often regularly for shopping in November 2025. By February 2026, that number had already risen to 47%.
Surveys of retailers indicate that AI agents are already being piloted or partially deployed across the industry, with most applications currently concentrated in customer service and marketing rather than fully consumer-facing shopping assistants. At the same time, consumer-facing platforms in retail, travel and delivery are increasingly experimenting with embedded AI features, but adoption remains uneven and still in early stages across many sectors.
Some examples aggregated over the recent few years are:
- Mercari marketplace platform for secondhand items has launched its ChatGPT-powered shopping assistant Merchat AI in 2023.
- The built-in Klarna plugin for ChatGPT offered users the opportunity to request advice on making purchases and receive recommendations related to specific products the same year.
- Amazon began testing a shopping assistant and a health-focused chatbot with a subset of users in 2025.
By contrast, just 10% of consumers say they want their AI shopping assistant embedded directly within a retail, travel or delivery platform. This preference for independence reflects how consumers are using AI in the shopping journey. Independent AI assistants are well suited to comparison-based discovery, enabling users to search across multiple merchants rather than being confined to a single commercial ecosystem.
An important note is that this independent AI assistant should also be free to use. Another part of PSE Consulting research revealed earlier showed that, when facing the options to pay for impartial AI selection and use a free AI shopping assistant even if its recommendations were shaped by advertising, 43% would choose the latter.
However, customers’ strong preference towards independent AI shopping assistants (whether ad-influenced or not) doesn’t mean retailers should not work on embedded AI helpers at all. After all, AI assistance technology is still emerging and is not universal, so shoppers may simply opt for the tool they already know and use instead of checking out whether a particular platform has an embedded one. The research also challenges predictions that AI will displace marketplaces altogether.
Despite growing adoption of AI shopping tools, 90% of consumers expect their use of retail marketplaces and travel platforms to stay the same or increase, suggesting consumers increasingly see discovery and transaction execution as separate stages of the shopping journey. The same pattern holds in adjacent categories, with 88% expecting continued or increased use of food delivery platforms.
This separation is central to the study’s findings. Consumers appear to be outsourcing product discovery and comparison to AI tools while continuing to rely on established platforms for payment, fulfilment and post-purchase support. In practice, this positions independent AI assistants as a discovery layer, while marketplaces retain their role as the execution and trust layer of commerce.
Chris Jones, Managing Director at PSE Consulting, said the research challenges assumptions that agentic AI will disintermediate the role of marketplaces:
“There has been a narrative that agentic AI would make marketplaces less relevant by allowing consumers to curate the internet in their terms, rather than relying on marketplaces. What the research suggests instead is that consumers increasingly see discovery and execution as distinct stages of the shopping journey. Consumers increasingly want AI to help them navigate choice across the internet, but they still rely on established brands such as marketplaces when it comes to fulfilment, payments, logistics, customer service and operational trust.”
He added:
“The implication for merchants and platforms is significant. The competitive battleground is moving upstream into the AI discovery layer. Success will depend on whether products and services are surfaced by AI assistants before consumers ever enter a marketplace environment.”
At the same time, the findings reveal a certain gap between experimentation on the supply side and preference for independence on the demand side, which may be critical for strategic investment decisions. Rather than requiring retailers to aggressively build proprietary AI shopping assistants, the research suggests the more important battleground may be visibility within external AI ecosystems rather than ownership of the interface itself.
The study concludes that platform-embedded AI is more likely to complement independent assistants than replace them. While independent tools dominate early-stage discovery and comparison, embedded AI within marketplaces may play a stronger role in later stages of the journey, helping users refine choices, personalise options, and complete transactions.
On the second thought, embedded AI assistants in marketplaces might also serve as a bridge between sellers and customers’ own AI agents if (or when) the agentic commerce vision tech leaders share will come true.
“The time will come where agents will be buying for us from other agents. That opens completely new business models. Who the new winners of this commerce age will be, we still don’t know. They might be walking around here today.”
Joao Silveira, Global Black Belt Business Applications EMEA at Microsoft
PSE Consulting research is part of a broader study into the commercial and structural implications of agentic commerce across four major Western ecommerce markets, with the company hosting a webinar on 28 July to explore the findings in more detail.
