In this week's column, Shirley Marschall looks at the agentic AI 'solve' for shopping, and the danger of losing some of the purchase journey's moments of magic...

Women and shopping. Such a cliché, isn’t it? And yet not entirely wrong. Yes, many women enjoy shopping. And did you know that women make 85% of day-to-day spending decisions for the family?

But this is not a piece about feminism or women’s habits, nor is it speaking on behalf of all women. It’s simply that most of the commentary on agentic shopping so far has come from men. So maybe it is worth adding a female POV to the debate.

What women want? No idea. But it’s safe to say that shopping isn’t something the majority of us want "solved."

And yet, here we are… (allegedly) AI will now "do the shopping for you" and handle the entire buying process. Really? What kind of shopping exactly?

Shopping isn’t really a catch-all term, nor is it a single category. It is an entire ecosystem. It’s shopping, buying, window shopping, bargain hunting, retail therapy. It’s browsing, touching, smelling, wandering, comparing, hesitating… discovering something unexpected because you happened to walk past it.

Yes, shopping is far more complex than "AI will do it for you." 

And then there are all these little moments and decisions that make us buy a certain product, order from a specific online retailer, or go into a store. It might be because the drugstore is next to your dog’s favourite pet shop, or because an online store carries your favourite, unique toothpaste that tastes like blood orange and basil.

It might be because a brand reminds you of childhood, or because another brand betrayed you once and is now dead to you. It might be because a retailer has a return policy that does not make you cry. It might be because a product ticks the box of your love for natural materials. It might be because the mall has the best coffee shop, or the pickup station has the most convenient parking…

Abandoned carts in internet heaven

One thing is for sure: there is a very special place in internet heaven for all the abandoned shopping carts, the items marked as favourites, the customised cars and shoes you never intended to buy. The things you dream about, the things you wish you could afford, or simply things you like but already own plenty of (Yes, you know deep down that you do not need yet another nail polish in coral-red...)

And so many additional factors influence what and where we buy. Do you live in the city or the suburbs? Do you have a car? Do you prefer chains or small businesses? Do you like eating out or cooking? Do you have a family or are you single? The list goes on…

None of that happens by prompting 'Get me the best deal' into a chatbot. None. And training an AI assistant… honestly sounds way more exhausting than shopping. 

These are not simply data points. They’re experiences, emotions, nostalgia, routines, and micro-preferences. They change constantly and at the same time not at all. Some decisions are not even conscious, like driving on autopilot from sheer habit.

Oh and then there of course are all the products we literally don’t care about. The quick-grab, one-click, zero-thought items we would never in our lives discuss with an AI assistant.

Another thing, agentic shopping doesn’t take into account: some people like leaving the house. Yes, browse stores, feel fabrics, smell candles, or simply really not wanting to end up like the humans in Wall-E, ordering our lives from a reclining chair.

Here’s an alternative proposal. Myles Younger argues that "AI could also shop with you, not always for you," and that the winning agentic commerce apps will strike gold on enjoyment, surprise, service, and delight.

So… an AI companion? An AI shopping assistant hybrid? Interesting. Maybe someone will strike gold. Maybe.

But here’s another thought: are solutions what we’re looking for? AI already does a lot of mansplaining. But when women search, we don’t always want an answer or a solution. Sometimes we want inspiration, exploration, discovery. Because shopping isn’t necessarily a chore and therefore doesn’t need to be solved.

So what are the incentives? 

Brands? Why would they care whether you buy in-store or online, with credit card or Apple Pay? They care that you buy. Full stop.

So are tools like Amazon’s Rufus really for shoppers or for shareholders? And isn’t Amazon in a very different position than most online retailers? It is Amazon, the AI company and Amazon the advertising company, after all. If they don’t offer an agent, who will? And in case it’s not obvious, Amazon’s agent won’t tell you, "You know what? You’ll find the exact product you need at Walmart."

OpenAI, continues its spaghetti approach of throwing every product possible at the wall, hoping something will stick. Of course that includes an AI shopping assistant. Though they recently admitted that product responses in ChatGPT Search are accurate only 64% of the time, and their new Shopping Research tool takes a few minutes to respond. Who could possibly resist that?

And even if these problems vanish, we are still left with trust, payments, liability, fraud, returns, regulation, conflict of interest, and everything else.

So maybe autonomous shopping agents will become an option. Maybe they’ll be great for certain tasks. Maybe they’ll quietly disappear like every previous ecommerce hype cycle which was wrong. Choose it, trust it, use it, or don’t.