Do you remember the days when we’d take brisk walks to our favourite stores to go shopping? Then came e-commerce, which boomed during the pandemic, bringing a seemingly endless stream of parcels to our doorsteps. Not long after, social media became our newest shopping destination, with TikTok Shop and increasingly sophisticated algorithms influencing what we discovered, desired and eventually bought.
Well, now we’re entering another era: Agentic AI shopping, or in slightly more technical terms, agentic commerce. Essentially, it means having an AI assistant that can search for products, compare options and even make purchases on your behalf.
Last month, Meta launched its personal AI agent, Muse, in North America. It behaves much like your personal assistant, one you can hand over all the grunt work to, including shopping. Unsurprisingly, Muse quickly rocketed to the top of Apple’s App Store charts in the US and Canada, overtaking ChatGPT, indicating that AI agents could become a much more familiar part of our everyday lives.
I cannot quite fathom having a set of computer codes do the shopping for me. While I’d certainly give it a try for the experience, I’m not sure I could outsource something which is much of a retail therapy.
After all, isn’t part of the fun in browsing? In discovering a product you weren’t necessarily looking for, changing your mind along the way, or finding something that catches your eye for no particular reason?
I’ve used AI to recommend products to me, compare prices and check whether an item is back in stock. In fact, when I tried Amazon’s AI shopping assistant, Rufus, I found it surprisingly useful for comparing product features and prices, saving me a considerable amount of time.
But there’s a difference between letting technology help you shop and letting it do the choosing for you.
Perhaps that distinction matters more to someone like me, who genuinely enjoys the process of shopping. For someone else trying to find the best deal on a dishwasher, it might be a different story.
On a quick side note, something else that caught my attention was Muse’s icon. Surprisingly, it looks nothing like what I’d expect from an AI agent. It’s warm, welcoming and teddy-bear-like, with a certain cuteness that makes you want to try it.
I found that contrast quite interesting. Behind this rather innocent-looking icon is a sophisticated piece of technology that can learn about us, access information and carry out tasks on our behalf. Perhaps that’s precisely why Meta chose such a friendly design, making something potentially intrusive feel far less intimidating.
When it comes to fashion, as consumers, this essentially means we’ll increasingly be shown products that AI considers worth recommending.
For example, when we look to buy new running shoes, what do we think of? Is it comfort, innovation, cushioning, price, or customer reviews? Speaking from my end, I’d say in most cases, probably a sound mix of all.
But if I were to ask an AI agent to find me the “best running shoe”, there’s a chance it might focus heavily on just one of those features, rather than considering everything that makes a shoe worth buying. Let me break it down.
Demonstrating the above, I came across a study by Jellyfish, a digital marketing company, that explored how AI shopping agents are changing the way we discover brands and products. The researchers tested 50 different shopping scenarios involving running shoes to understand how AI recommends products, compares prices and decides which sources to trust.
The findings suggested that AI tends to favour products based on the features most frequently talked about online, potentially overlooking other qualities.
For example, take Nike. A running shoe might offer excellent cushioning, innovative technology and features designed for speed. But if AI mostly associates that shoe with cushioning, it might recommend it to someone looking for comfortable running shoes but overlook it when someone else searches for innovative footwear, even though the shoe does both equally well.
In other words, a product could have much more to offer than what AI recognises it for.
If we continue relying on AI agents to do our shopping, I wonder whether we risk developing a rather limited understanding of the brands and products we buy?
Don’t get me wrong, I can see how useful they are for discovering new products. But their recommendations depend heavily on the information they can find online, whether that’s customer reviews, product descriptions or other sources.
And that information doesn’t always tell the whole story. A product might have far more to offer than what’s being said about it online. If certain qualities aren’t talked about enough, or the relevant information isn’t easily accessible to AI, a perfectly good product might never make it onto our list of recommendations.
In a way, AI shopping agents could end up narrowing our perception of a brand before we’ve even had the chance to explore it ourselves.
This puts emerging designers with less digital visibility at a disadvantage as Agentic AI’s recommendations consistently favour products with abundant reviews, accessible ‘crawlable’ information, and established online recognition.
From a business perspective, this means that brands are no longer competing solely for the attention of customers. They also need to be recognised, understood and recommended by AI, giving birth to another emerging economy of what is known as ‘generative engine marketing’.
In September 2026, Mastercard published a report predicting that more than 1 in 10 online shoppers could regularly use AI agents to purchase products on their behalf by 2030.
For businesses, this creates an entirely new challenge: It is no longer enough to invest in beautiful marketing if the information behind it is difficult for AI to find, understand or compare.
Every few months, it feels as though technology is accelerating faster than we have the chance to fully understand or adjust to it. Businesses spent years learning how to appeal to social media algorithms, only to find themselves navigating yet another shift, this time shaped by AI.
The challenge is becoming as much about how clearly a machine can understand a product as how beautifully a brand can present it.
I’m curious to see how brands will adapt to this new world, where they must appeal as much to the human eye as to machines.
In making shopping more efficient, are we losing some of the very things that made the ‘hunt’ enjoyable in the first place?