Not long ago, online shopping still required quite a lot of shopping. You searched Google, opened half a dozen tabs, compared prices, skimmed reviews, checked delivery times and eventually made a decision. It was more convenient than driving to a store, but the work of choosing was still yours.
That is starting to change in a much more fundamental way. AI systems are moving beyond recommending products and towards acting on our behalf. They can compare options, assemble a basket, apply preferences, initiate payment and increasingly coordinate what happens after the purchase. McKinsey & Company estimates that by 2030, agentic commerce could orchestrate between US$3 trillion and US$5 trillion in global retail revenue.
It is tempting to interpret this as the next chapter of e-commerce: shopping gets faster, stores get less important and another layer of friction disappears. I suspect that misses the more interesting shift.
AI may not simply change how we shop. It could split shopping in two. At one end, more buying becomes almost invisible as machines take care of the routine decisions we would happily avoid. At the other, the shopping we still choose to do ourselves may need to become more human, more social and more worthwhile.
The future of retail, in other words, may be simultaneously more automated and more experiential.
AI is crossing the line from adviser to buyer
The first wave of AI shopping tools mostly helped us decide. Ask ChatGPT for a good pair of noise-cancelling headphones or Gemini for a gift idea and it could narrow the field. Useful, certainly, but not radically different from the role search engines, review sites and recommendation algorithms have played for years.
The important development is what happens next. AI is beginning to cross the line between advice and action.
PayPal, for instance, announced that it would adopt OpenAI’s Agentic Commerce Protocol and allow shoppers to use PayPal within ChatGPT’s checkout experience. The idea is that a customer can move from conversation to transaction without returning to the old sequence of search results, retailer websites and checkout pages. Similar moves are underway across Google, Microsoft, Visa, Mastercard, Shopify and other parts of the commerce ecosystem.
McKinsey describes agentic commerce as shopping powered by AI agents that can anticipate needs, navigate options, negotiate and execute transactions in line with a person’s intent. That last word matters. The interface shifts from telling a website exactly what to do towards telling an agent what outcome you want.
Imagine saying: ‘I need a pair of black sneakers for a work trip next week. Under $200, comfortable enough to walk 15 kilometres a day, and nothing that looks too much like a running shoe.’ The agent could know your size, check stock and delivery dates, compare reviews, assess your past preferences and present the strongest options. With the right permissions, it could eventually make the purchase too.
At that point, buying has not disappeared. But a surprising amount of shopping has.
The customer may no longer be the person browsing your website
For retailers, this creates a shift that is easy to underestimate. For three decades, digital commerce has been built around attracting a human being to a digital destination. Get them to your website. Keep them there. Make navigation intuitive. Reduce abandoned carts. Optimise every click between discovery and payment.
An AI agent does not need to shop that way. It can move across platforms, ingest structured product information, compare prices and policies, evaluate reviews and make millions of small trade-offs without ever experiencing the beautifully designed homepage a retailer spent months perfecting. McKinsey argues that designing the ‘agent experience’ could therefore become as important as designing the customer experience.
This changes the meaning of visibility. In the search era, the commercial question was whether a customer could find you. In the agentic era, the question increasingly becomes whether their AI can understand you, trust your information and transact with you.
That makes some fairly unglamorous things strategically important: accurate product descriptions, live availability, transparent pricing, credible reviews, clear returns policies and data that machines can reliably interpret. The Guardian has reported retailers already experimenting with ways to make their products more legible to AI systems as shoppers shift from keyword searches towards conversational recommendations.
This is a continuation of the shift from SEO to answer engine optimisation that I have written about before, but agentic commerce takes it one step further. Being cited in an answer is valuable. Being selected by an agent that can actually buy is something else entirely.
Retailers have spent years trying to understand the customer journey. They may now need to understand the journey their customer’s agent takes as well.
The Shopping Split
This is where I think the next phase of retail becomes particularly interesting. Not every purchase is equally suited to automation.
There is a vast category of things most of us buy without deriving much joy from the process: dishwasher tablets, printer ink, pet food, toothpaste, pantry staples and the same household products we have purchased dozens of times before. These are predictable, repeatable and generally low-regret decisions. If a trusted AI can keep an eye on price, stock levels and our preferences, many consumers will happily let it do more of the work.
Call this Invisible Commerce. The purchase still happens, but the shopping becomes less visible. We set the parameters and the system handles the repetition.
Then there is Intentional Commerce: purchases where the process itself matters. A birthday gift. A suit. A new couch. A special bottle of wine. A hobby purchase. A family holiday. In these categories, we may want to browse, touch, compare, ask questions, change our minds and discover something we did not know we wanted.
The distinction is not simply between cheap and expensive products, nor online and offline retail. The same item can move between the two categories depending on the occasion. Buying another white business shirt may be almost entirely delegable. Choosing what to wear to your daughter’s wedding probably isn’t. What separates the two is the value we place on participating in the decision.
That gives retailers a more useful question than ‘How much of shopping will AI automate?’ The better question is: Which parts of our customer’s journey will they gladly outsource, and which parts will they still value doing themselves?
As buying gets easier, stores need a better reason to exist
The Shopping Split also changes how we should think about physical retail.
The familiar prediction is that every improvement in digital shopping makes stores less relevant. Certainly, some categories will continue moving online and some routine store visits will disappear. But recent research suggests the picture is more complicated. McKinsey and the International Council of Shopping Centers surveyed more than 3,000 US consumers in 2026 and found physical stores remain important even as AI takes a larger role in discovery and routine purchasing. Their research points towards two increasingly distinct reasons to visit a store: convenience and discovery.
Some physical retailers will become exceptionally good at speed. You know what you need, you know where it is, you can confirm it is in stock, collect it quickly and leave. Smart carts, checkout-free systems, digital shelf labels and tighter links between apps and stores all fit this model. Other retailers will need to head in almost the opposite direction. They will give people a reason to linger.
We can already see this in destination stores, specialist retailers and brands that treat physical space as somewhere to play, learn, get advice or meet people with the same interests. Lego’s large-format stores lean heavily into experience. Specialist comic retailers survive in part because knowledgeable staff and a community of enthusiasts can provide something an efficient transaction cannot.
This is the paradox I find most interesting: the easier AI makes buying, the harder stores will have to work to justify shopping. A mediocre store whose only advantage is that it happens to contain products may struggle. A store that saves me time has a reason to exist. So does one that gives me expertise, theatre, serendipity or human connection. The uncomfortable middle gets harder to defend.
The more transactional buying becomes automated, the less transactional good physical retail can afford to be.
Brand loyalty gets an algorithmic middleman
There is another consequence that deserves more attention. AI agents could change the relationship between customers and brands because they insert a new decision-maker into the middle.
Consider the way loyalty works today. You might automatically open the app of your preferred airline, supermarket or fashion retailer because habit, familiarity and rewards keep you coming back. But if your agent is instructed to find the best combination of price, quality, delivery time and past satisfaction, it has little reason to respect habit for habit’s sake.
That could make loyalty more conditional. A customer may be loyal to the agent precisely because the agent is willing to be disloyal to brands on their behalf.
This does not mean brands cease to matter. In fact, trust, reputation and emotional preference may become important signals an agent needs to incorporate. But the route between preference and purchase becomes less direct. Brands will increasingly need to be chosen twice: once by the human setting the intent and again by the machine evaluating the options.
There is a marketing consequence here too. Traditional digital advertising was built around winning human attention. Agentic commerce creates an environment in which at least some decisions may occur without a person seeing the ad, scrolling past the sponsored result or visiting the retailer’s site. McKinsey explicitly identifies pressure on advertising and retail-media models as agents begin to bypass parts of the traditional discovery funnel.
For marketers, the scarce resource may gradually shift from attention to recommendation.
Trust becomes part of the plumbing
There is, however, a large gap between asking an AI what television to buy and giving it permission to spend $2,000 on one.
The more autonomy we hand to agents, the more important trust becomes. Who authorised the purchase? What happens if the agent misunderstood the brief? Who is responsible if a system buys the wrong product, exposes sensitive information or is manipulated into making a decision that benefits someone other than the customer? Can I see why it chose one option over another? Can I reverse what it did?
These questions are why payments companies are working on delegated authorisation, programmable spending rules, agent identity and auditable transaction trails. McKinsey notes that the existing payments infrastructure was largely designed around a human being in the loop. Agentic commerce requires a different trust architecture because the entity arriving at checkout may be software acting on someone’s behalf.
This will probably put a natural brake on the most breathless predictions. Consumers are unlikely to hand over unlimited financial authority simply because an AI assistant is convenient. Autonomy will be earned category by category and transaction by transaction.
My sense is that the winning systems will not be those that remove the human from every decision. They will be those that become good at knowing when we want to be removed and when we want to step back in.
Retail’s next advantage may be knowing when not to automate
There is a tendency with every new technology to assume its highest form is total adoption. If AI can automate a process, surely the goal is to automate all of it. Retail may be a useful reminder that efficiency and value are not always the same thing.
Sometimes the value of a purchase is precisely that I did not have to think about it. I do not need a memorable experience buying laundry detergent. If an agent can make that transaction almost disappear, terrific.
But there are other purchases where removing the effort removes part of the value. Wandering through a bookstore, talking to someone who knows far too much about coffee, trying on a jacket you would never have searched for, choosing a gift that says something about how well you know another person. These experiences contain inefficiency, but that inefficiency is not necessarily a defect.
That is why I suspect the future of shopping will be defined less by a simple march from human to automated commerce and more by a sorting process. We will increasingly outsource the purchases where participation adds little value and become more discerning about the experiences where it does.
For retailers, that means the strategic challenge is not merely to adopt AI faster. It is to decide what should disappear into the background and what deserves to become more distinctive in the foreground.
In the next era of retail, buying may become almost effortless. Shopping will have to earn its place.
Michael McQueen is a globally recognised trend forecaster, change strategist and keynote speaker.
A bestselling author of 10 books, Michael’s latest release was named by Malcolm Gladwell and Adam Grant as one of the top five must-read new leadership books. He is a sought-after media commentator, with his insights regularly featured in Forbes, The Guardian, and CNN.
To find out more about Michael and his work, click here.
NOTES
1. Katharina Schumacher, Roger Roberts, and Katharina Giebel, “The Agentic Commerce Opportunity: How AI Agents Are Ushering in a New Era for Consumers and Merchants,” McKinsey & Company, October 17, 2025, https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-agentic-commerce-opportunity-how-ai-agents-are-ushering-in-a-new-era-for-consumers-and-merchants.
2. Ivan Mehta, “PayPal Partners with OpenAI to Let Users Pay for Their Shopping within ChatGPT,” TechCrunch, October 28, 2025, https://techcrunch.com/2025/10/28/paypal-partners-with-openai-to-let-users-pay-for-their-shopping-within-chatgpt/.
3. Sarah Butler, “‘What to Buy Dad for Christmas’: Is Retail Ready for the AI Shopping Shift?,” The Guardian, December 11, 2025, https://www.theguardian.com/business/2025/dec/10/is-retail-ready-for-ai-shaking-up-how-we-shop.
4. “Shopping in the age of AI: Redefining stores for a new era, 2026. Supplied research materials,” McKinsey – https://www.mckinsey.com/industries/retail/our-insights/shopping-in-the-age-of-ai-redefining-stores-for-a-new-era.
5. “The rise of the agentic shopper: ASOS’s AI investment,” McKinsey – https://www.mckinsey.com/industries/retail/our-insights/the-rise-of-the-agentic-shopper-asoss-ai-investment.
6. “Gap says it will launch checkout within Google’s Gemini, in an AI first from a major fashion company,” CNBC – https://www.cnbc.com/2026/03/24/gap-google-gemini-checkout-ai-platform.html.
7. “Walmart Expands Wing Drone Delivery As Last Mile Story Takes Flight,” Yahoo Finance – https://finance.yahoo.com/markets/stocks/articles/walmart-expands-wing-drone-delivery-232020353.html.
8. Michael McQueen, “How AI Is Quietly Rewriting the Rules of Retail Shopping,” MichaelMcQueen.net, January 31, 2026, https://michaelmcqueen.net/trends-and-insights/how-ai-is-quietly-rewriting-the-rules-of-retail-shopping/.
9. Michael McQueen, “Your AI Shopping Butler Is Ready: How Agentic Commerce Is About to Redefine Retail Forever,” MichaelMcQueen.net, accessed September 2026, https://michaelmcqueen.net/trends-and-insights/your-ai-shopping-butler-is-ready-how-agentic-commerce-is-about-to-redefine-retail-forever/.
