Conversational AI robots start talking to shoppers in stores

Physical stores are testing social robots and 'digital humans' that can hold a real conversation with shoppers — not another screen or chatbot, but digital colleagues that, thanks to a language model, know in real time what is in that specific store's assortment, stock and services. That is the finding of an analysis published by Dutch trade outlet Emerce, based on applied research projects run with retailers and technology vendors. According to the analysis, the question is no longer whether conversational AI will play a role on the shop floor, but how it is already changing customer contact there.

From customer profile to conversation on the floor

Retailers have spent years investing in customer data: purchase history, loyalty programs, detailed profiles. But on the shop floor, a need often only surfaces in the moment — a shopper is looking for a gift, torn between two products, or wants to know what fits a specific situation. A store does not always need to know who that customer is first; it mostly needs to understand what they need right now. A social robot or digital human can start a conversation, probe further and sharpen that need step by step — essentially what a good sales assistant has always done, now supported by a language model and organized at scale.

That dialogue only becomes genuinely useful once it draws on local store knowledge. Generic product information — specs, prices, reviews — is now available everywhere, from webshops to comparison sites. What sets a store apart is its own assortment, layout, stock and staff: carefully sourced products, local or artisanal items, temporary promotions, combinations staff often recommend. Fed with that store-specific knowledge, a robot or digital human can answer questions that only matter in that particular branch: where a product sits, whether it is in stock here, what alternatives exist, which staff member can help with a specialist question.

What every conversation teaches the retailer

Till and loyalty data show what customers ultimately buy, but rarely which questions came before that purchase, or why a shopper walked out empty-handed. By actively asking questions, a social robot or digital human surfaces a new source of information: interaction data on what customers hesitate over, what product information they are missing, and what alternatives they are looking for. Properly captured, that data helps retailers not just with direct customer service, but also with assortment decisions, store layout and service improvements.

The same technology can also capture practical expertise that often disappears once an experienced staff member leaves — a real risk given staff shortages, turnover and widespread part-time work. That does not happen automatically: retailers first need to make that knowledge explicit — product stories, advice rules, frequently asked questions — before a language model can use it. Once linked to a robot or digital human, that expertise becomes available more broadly, including during busy periods, for new colleagues, or when the specialist is not around — scalable expertise rather than knowledge trapped in one person's head.

A global push, and pushback

What Dutch stores are testing fits a much broader international trend. A survey of U.S. retail and technology executives, published by Retail Customer Experience, finds that 92% of U.S. retailers are increasing their AI investment in 2025, with the global market on track to reach $52.45 billion by 2030. The same survey finds that 70% of shoppers have already used an AI tool during their shopping journey, and expects consumers in 2026 to lean further on AI agents to plan, compare and complete purchases — an extension into commerce of the broader adoption of agentic AI in the enterprise that is pushing retailers to connect their own data more effectively.

The same sources show it is not friction-free. A Kinsta survey cited in that same report finds that 93% of consumers still prefer talking to a human, 84% say humans are more accurate, and 80% believe AI is mainly used to cut costs rather than improve service. That tension lines up directly with the sixth and final shift identified in the Dutch research: simply placing a robot or digital human in a store changes nothing on its own. The technology only becomes a genuine digital colleague once staff are brought along and their role in the service process is rethought with them.

Elevated in-store experiences will remain a differentiator for brands and retailers — and one of the most impactful ways to connect with customers even in the age of AI.

• 92% of U.S. retailers are raising AI investment in 2025, toward a $52.45 billion global market by 2030 (Retail Customer Experience).
• 70% of shoppers have already used an AI tool while shopping, ahead of the expected rise of AI shopping agents in 2026.
• 93% of consumers still prefer a human, 84% call humans more accurate (Kinsta survey, cited by Retail Customer Experience).
• Six shifts identified in stores by Emerce: from customer profile to dialogue, from generic to local knowledge, from till data to stated need, from vanishing expertise to shared knowledge, from interface to service personality, and from standalone tech to digital colleague.

What this means for American retailers

For an American retailer, this means homework before installation, not after: store-specific knowledge — stock, assortment, the advice rules of top-performing sales staff — has to be documented first, or the robot ends up sounding as generic as any webshop chatbot. It also means bringing staff into the new division of labor rather than sidelining them, especially since the same U.S. data shows a majority of shoppers still distrust an AI-only experience. Done properly, that groundwork produces a sales assistant that never misses a busy Saturday and shares the best colleague's knowledge with the whole team; done carelessly, it produces just one more expensive screen.