As the next iteration of artificial intelligence enters the retail market, retailers must make sure human connection lies at the heart of the technology. Retail Systems news editor Alexandra Leonards explores the opportunities and challenges for retailers as they begin to experiment with agentic AI.

The recent rise of generative AI (GenAI) has demonstrated that the retail industry – spanning grocery to apparel – is eager to embrace advanced technology.

The variety of innovation in the sector over the past six months has been impressive. L’Oréal, for example, is developing a custom AI model that unearths new insights from cosmetic formulation data to drive the use of sustainable raw materials, while Depop has created a listing description feature using image recognition and genAI to expedite its resale process. Meanwhile, Dunelm has introduced a new tool to enhance online product discovery, whilst H&M recently announced plans to create digital twins of fashion models using genAI.

Just as retailers begin to celebrate their progress with AI, an even more advanced iteration is already emerging. Whereas GenAI generates responses based on single interactions through natural language processing, agentic AI navigates multi-step problems using advanced reasoning and interactive planning.

While the latest iteration of AI promises to take retail innovation a step further, with the potential to supercharge what retailers have already achieved, its ability to make decisions autonomously also poses a risk to both the consumer and employee experience. Ultimately, over-automation or a failure to put human needs and wants at the heart of the technology could lead to impersonal interactions and a stifling of creativity.

Human connection

William Hewish, chief information officer (CIO) at Pets at Home, says that there is “significant potential” for agentic AI across the retail market if teams are willing to think creatively whilst ensuring the basics of good technology application are at the centre of every decision.

“This is a special industry because, at its core, retail is about human connection,” he tells Retail Systems. “The best applications of agents will bear this in mind and will help retailers facilitate meaningful human connection, because that’s what resonates best with customers.”

If used in the right way, agentic AI can help to drive more meaningful interactions with customers. For example, the technology could provide store employees with the necessary tools to help customers with a one-off purchase or insights about their past shopping habits for a more personalised experience.

Rather than there being one particularly transformative use case for retail, it will be the retailers that use agentic AI to do the simple things really well that are likely to stand out.

“We’ve all been frustrated by chatbots that take up precious time and don’t answer your questions; we don’t, however, remember the ones that triage requests swiftly and efficiently onto the right expert who knows everything about your case,” he continues. “Processes like that; developing agents and AI applications which seamlessly blend with human expertise is where the largest opportunities are.”

While different forms of AI, such as machine learning, have been a part of the company’s strategy for nearly a decade, Pets at Home is now looking to the latest advancements to help the organisation scale capacity across the multiple business areas it operates in, with the technology playing a key role in how it meets its omnichannel ambitions.

“We’re a retailer, we’re a healthcare provider, we have an app, a website, and we offer expert grooming services,” explains Hewish. “AI works alongside our colleagues to provide them with quicker insight and give them the ability to be more efficient, allowing us to scale and grow the business while improving our overall consumer service.”

The retailer has also been piloting a veterinary digital assistant in wider trials, which it is about to start rolling out.

“This is a revolutionary tool that enhances the efficiency of veterinary practices by automatically transcribing consultations, drafting clinical notes to be reviewed by clinical teams, and scheduling follow-up appointments,” continues the CIO. “The results have been remarkable; our vets trailing the system have been impressed by its speed and accuracy.”

Pets at Home is already using intelligent agents in the contact centre to help manage customer enquiries, with an overall aim of ensuring consumers receive timely, consistent support. However, Hewish points out the need to retain the company’s human agents for more complex needs.

“We have deployed our AI fraud agent, supporting our fraud team in detecting and assessing potential online fraud cases, and we are also currently rolling out a colleague retail assistant which provides our colleagues in store with instant access to knowledge and training, to provide the best possible service to customers,” adds Hewish.

A blueprint for agentic AI

In early 2025, US tech giant Nvidia announced the launch of a new blueprint for the development of agentic AI-powered retail shopping assistants.

The company’s ‘blueprints’ are reference workflows for both agentic and generative AI use cases.

The latest is designed to help developers create AI-powered digital systems that work alongside and for human workers by offering the expertise of a retailer’s sales associates, stylists, or designers to consumers.

The blueprint provides an end-to-end multimodal capability which enables the assistant to understand text and images, while a multi-query capability means that the technology can search for several products at a time.

For instance, a consumer might enquire whether a retailer stocks tops or shoes, whilst simultaneously requesting handbag recommendations to complement a specific skirt.

In the home furnishings market, as well as facilitating furniture recommendations, product details such as fabric care instructions, and a search option to find products that are similar to an image of an item, the shopping assistant can enable a customer to select furniture from a catalogue and see it represented in their own room.

Speaking about Nvidia’s latest agentic AI advancement for the retail space, Cynthia Countouris, the company’s director of product marketing, also highlights the importance of putting the customer at the heart of the technology.

She says that what consumers want is one, unified interface where they can access everything from customer service and information about home delivery to product search.

“Customer service is a natural building block for agentic AI,” Countouris tells Retail Systems, adding that through the technology shopping assistants will become much more natural.

The product marketing director explains that from a retailer’s perspective there are a number of use cases for agentic AI to improve the customer experience, like implementing an AI agent to direct someone around a grocery store; for event, menu or recipe planning; or to support the process of selling and buying over-the-counter medicines.

“And so, what we're doing is basically providing building blocks depending on how companies want to create new user experiences, either for their consumers or to power their employees,” she adds.

The risks

Pets at Home CIO William Hewish says that technological change can be a risk in any sector.

“Ensuring change can be seamlessly integrated and not interrupt anything customer-facing or which doesn’t negatively impact colleagues in store is the biggest challenge,” he continues.

Whilst the technology offers the potential to transform customer interactions and experiences, retailers must make sure that there is a clear balance between personalisation, privacy, and security.

“Staying true to a clear and overarching purpose is key, and at the same time, privacy and security have to come first,” continues Hewish. “A test and learn phase for any significant rollout is crucial.”

Equally, he explains, making sure all data is protected and hosted on platforms that a retailer has the utmost confidence in should always be a priority, especially when it comes to customer data.

Again, when it comes to understanding the risks associated with agentic AI, Hewish highlights the importance of keeping humans at the heart of decision-making.

“Ensuring customers trust us as their retailer of choice and our systems is vital,” he adds.

Nvidia’s Cynthia Countouris points to the risk of not rolling out the technology.

“What’s the risk of not improving my employee productivity, operational efficiency, and customer experience?” she asks. “That’s a bigger risk than deciding to get in and dabble.”

Job losses

Fears about advanced technologies stealing jobs are nothing new. But with agentic AI able to power clever AI agents that make decisions autonomously, it’s likely that this may raise even more concerns for people working in retail about their future, particularly in an industry already facing significant job losses.
But Hewish says that the biggest risk of agentic AI when it comes to jobs is not having the right learning practices in place.

“If a company insists on pursuing the rollout of an agent – or any technology – that doesn’t work for colleagues, that’s when problems will arise,” he continues. “The key is testing and learning; without the learning, any positive change you’re aiming for is much harder to achieve because you’re only really working at the ‘testing’ phase.”

Countouris agrees that when it comes to emerging technologies like agentic AI, it’s important to “crawl, walk, run.”

“You need to start to build some understanding and skills to build the organisation and cultural muscle up and facilitate change management,” she says. “I don’t see this industry implementing AI agents to maintain the status quo.”

She also points to comments made by chief executive of Nvidia Jensen Huang in January in regard to the safety of jobs.

“Suppose the work that you do continues to be important but the effort by which you do it, went from, you know, being a week-long to almost instantaneous,” he says in an episode of Cleo Abram’s YouTube science and technology show. “You know that the effort of drudgery basically goes to zero.”

Countouris says that ultimately, the “human in the loop” is still incredibly important and always will be.

While intelligent automation is an increasingly important tool in the retail space, excessive reliance on AI risks undermining the human interactions essential for fostering consumer loyalty, cultivating a customer-centric culture, enhancing emotional engagement, and delivering a truly personalised shopping experience.

Given these systems’ reliance on data, retailers must strike a careful balance between personalisation, privacy, and security – ultimately safeguarding customer trust. As retailers explore agentic AI, the innovations that prioritise meaningful human connections – whether by enhancing employee support or customer service – will drive transformation and unlock the technology’s full potential.

---