Just a few years ago, if you wanted to buy a pair of shoes, you would type that into Google.

But now, many use AI instead.

Nearly two-thirds of Indian consumers now use AI as a part of their shopping journey, according to Boston Consulting Group.

The upsides are obvious. It saves time, the results are more personalised, and you could potentially find a better deal.

But there are trade-offs.

For starters, you don't really know why AI shows you certain options over others.

And now, with AI moving from helping you decide to actually buying for you, how does that change shopping?

To find out, tune in to the latest episode of The Signal Brief, where you'll hear from consumers and experts on this.

The Core produces The Signal Brief. Follow us wherever you get your favourite podcasts.

NOTE: A machine transcribed this episode. A human has looked at this text but there might still be errors. Please refer to the audio above, if you need to clarify something. If you want to give us feedback, please write to us at feedback@thecore.in.

TRANSCRIPT:

Kudrat (Host): Abhishek, a 30-year-old manufacturing professional, told me that he’s the kind of shopper who does his homework.

Abhishek: It's either you rely on word-of-mouth options, which can be a little flawed because of how, how much or how little information you provide to somebody else.

So, that or you spend enough time online and you just keep Googling and you go on forums and you keep searching. So, I think that manual work was how I, how I used to approach things earlier.

Kudrat (Host): But now, he outsources much of that work to AI.

Abhishek: So more specifically, it's Gemini. Like I keep hopping between platforms as such. Currently, I'm working, like, w- a lot of my workflows and a lot of my personal usage is reliant on Gemini.

And how it usually goes for me is, like I have a product in mind, like something I want to add to my regular rotation, or if I'm looking for recommendations for a certain type of product, I will usually just open it up to like a specific chat that I've, trained in that order to kind of seek recommendations or get an aggregate of reviews from various online sources and that kind of takes a lot of the grunt work away from the decision-making aspect of it.

So if I had to give you an example, if I'm looking for, let's say, a laptop bag and, you know, I have certain criteria in mind, like a particular size or this number of pockets or this, like a specific number of liters minimum that I'm looking in terms of storage capacity. So I'll feed all, all of those parameters in, ask it to... ask the LLM to comb through online reviews on Reddit or through websites and kind of give me like a list of options or recommendations or alternative brands for that matter.

Kudrat (Host): Abhishek’s story is part of a larger trend.

Consumers across the world are replacing Google searches with LLM prompts.

Adobe Analytics found that AI-referred traffic to US retail sites jumped 393% year-on-year in the first quarter of 2026.

Nearly 2/3rds of Indian consumers now use generative AI as part of their shopping journey, according to Boston Consulting Group’s Global Consumer Radar.

So, what do we gain from handing over the work of shopping to AI? And what might we lose?

Kudrat (Host): My name is Kudrat Wadhwa and you’re listening to The Signal Brief. We don’t do hot takes. Instead, we bring you deep dives into the how and why of consumer trends.

In today’s episode: how AI is changing the way we shop.

Kudrat (Host): Now, most of us still recognise the old way.

You need to buy a pair of black shoes.

You type it into Google.

Pages appear according to SEO.

Paid placements come with labels.

Retailers like Amazon arrange their own shelves.

The system is imperfect and noisy, but the levers are visible.

But, when the same question moves into an LLM, the levers disappear.

The model does not return ten blue links.

It returns an answer.

Or three options.

Or a size recommendation and a cheaper link.

Kudrat (Host): There are some obvious advantages to this.

For one, it saves time. You don't have to read 100 reviews yourself.

And because the LLM can compare so many options at once, it can potentially help you find a better price.

The results are more personalised.

Prof Animesh Mukherjee, who researches responsible AI at IIT Kharagpur, saw this first-hand.

Prof Animesh: So I did it with, uh, this with ChatGPT, but I'm sure similar things are going to crop up in other platforms also and in some different avatars maybe. So I asked a very vanilla question: "I want to buy a phone, like can you suggest me some options?"

Now, it, it started churning different things. I did not interact with it. It churn- it started churning different things, different, analysis it did.

And then at the end it says, like the final solution it gives, before that it tells, like, "Since you are a professor and since you are from an central government, Indian institution, so I am assuming that your salary range is so and so, therefore you should be able to afford so and so." So I did not give it any information about myself in this session.

Kudrat (Host): The recommendations feel curated and for a lot of people, that’s the appeal.

But there are trade-offs.

When AI gives you three options, you don't see the hundreds it rejected.

You don't necessarily know why a certain product appeared, and why another one didn't.

Much of that decision happens behind the scenes, based on signals most users never see.

Here’s Prof Animesh again.

Prof Animesh: Social scientists have observed that it is driven by something called a choice architecture. So basically a choice architecture would mean, say for instance, you are buying a phone. So the choice architecture would include, different types of attributes like for instance, price, the model, the brand, the color, et cetera, et cetera.

What is happening is that this choice architecture itself is evolving. So it's no longer, you know, guided by a single, prototypical structure designed by a designer at the back end.

It's evolving, and it's evolving through the interactions that you are having with these agents. So for instance, like, if you go-- So we were seeing, like, if you pretend that you do not know anything, about a phone and you are, planning to search, for it, the first thing it picks up, is the price. So, and then it picks up, the brand and so on. So basically it is guiding you through an architecture.

Now, if you, if you change your prompt slightly and say like, "I have some idea about phones, and I like these so and so types of phones," immediately the choice architecture changes. Some other attribute pops up at the beginning.

Kudrat (Host): Because these systems are black boxes…meaning we, the public, don’t know what’s in the algorithm, it’s unclear why certain brands show up and others don’t.

Brands aren’t sure either. Which means they’re faced with a question: how do you make sure yours is the product that the AI chooses to show?

Some are trying to solve that problem. One such organisation is RankInLLM–an agentic AI platform that helps brands become visible inside ChatGPT, Perplexity, Grok and the rest.

I spoke to its founder, Kartik Sharma.

Kartik: What comes up is, in fact, one of the broader categories what we are trying to address at RankInLLM.ai, which is the problem of, generative engine optimization or answer engine optimization.

Back in the day, Google would typically get certain keywords from your question or prompt, and then they would essentially indexed various pages and serve you a best fit result based on their PageRank algorithm, um, and the ways in which Google serves the best, result to a particular user, given the context and memory they also have.

But now in, GEO and AEO, what's happening is there is no single playbook because it's not, just keywords, right?

I mean, essentially users are doing a threaded prompt. Like I'm... I may be saying that, "Hey, today I'm not feeling well," or, "I'm depressed," and hippity-hop, after two, three conversation, the LLM is suggesting me, "Hey, why don't you take a trip to Goa and you will feel better because travel always helps you?" And I'm, here am I sort of booking tickets on a online travel agent, while my initial use case was not travel, it was wellbeing or, or mental health, to that extent.

So I think what comes up, is a combination of various factors which the brands' digital assets, like their website, their social media pages, their YouTube channel, how they are structured, what kind of content is popped up there. Then also a lot of it is on third-party citations, like how consumers are talking about them, their brand on Google reviews, GMB, Google Business Profile, UGC platforms like Reddit, you know, Ambition Box, Salary Box, what the employees are saying, what the media houses are saying, if their PR coverage is about a particular brand. So as I said, there are a lot of signals which a particular LLM would look at, and what finally comes up is pretty dynamic.

It's hard to say that, hey, this is the rule book, if you follow this, you are guaranteed to be served on that particular result.

Kudrat (Host): In other words, the old SEO rulebook is gone.

An ever-shifting mix of on-site content, third-party chatter and conversation context has replaced it.

That opacity cuts both ways.

Brands can’t easily force their way in. And consumers don’t know why AI is showing certain results over others.

Here’s Abhishek again.

Abhishek: There are a couple of occasions where I've had to kind of, like, put in that am-extra amount of work to fine-tune that LLM to kind of, you know, tweak the recommendations a lot because sometimes there is a certain amount of brand pushing that I've noticed.

Because obviously we, we live in a, you know, kind of world where ads are inevitable.

There is no escaping, brand sponsorships and pushes and ads and whatnot. So obviously with any LLM, you are bound to see a certain amount of preferential, behavior, I would say, that, you know, "Why don't you try this brand?" Or, "You should check this one out." And it may not even meet the exact brief that I may have given it, so I wouldn't even call it hallucination, it's just brand pushing.

Kudrat (Host): Now, to be clear, ads inside chatbots are still limited in India.

But even without traditional ads, Abhishek says he feels the AI pushes certain brands over others.

And surprisingly, he's willing to accept some of that bias.

Abhishek: I think you kind of become accustomed to the idea from the get-go that, okay, what I'm going to get is not exactly unbiased.

At the end of the day, if the product's good, I think it's a small price to pay- Mm-hmm. that you will encounter a slight bias.

Kudrat (Host): For now, that trade-off works for Abhishek.

But AI is moving beyond recommendations.

It's starting to make the purchase itself.

Today the question still sounds like this: ‘Which shoes should I buy?’

Tomorrow it could sound like this: ‘Buy me the best shoes under a hundred dollars.’

That’s agentic AI–the system has agency, it will make the purchase for you.

We’ve covered agentic AI before too. I’ll link the episode in the show notes below.

Here’s Kartik of RankInLLM again.

Kartik: So I think with agentic commerce coming into play, shopping will become extremely, seamless because now consumers have an option to almost delegate, their process of thinking what they want, how they want to purchase, do comparative, do pri-be-better price discovery, and also ultimately, you know, consume that product and and service, give reviews, and then maybe get up-selled or, or cross-selled at using AI agents which they themselves have built.

Kudrat (Host): In India especially, the appetite for agentic commerce is high.

Adobe found that in India, 62% of consumers say they are already open to shopping through a virtual AI concierge, and 60% want their own personal AI agent, the highest interest in Asia-Pacific.

But, there are costs to letting agents run the shop.

Kartik: A agent, let's say if in the first turn of the conversation feels that this is a good enough, option for you, it, it'll just make the decision. And they, they are sort of... most of the agents are tuned to be efficient, right? And, and do things better and faster. But humans sometime, at least the shopping, you know, the experience or transaction for most human beings is not very, trivial a transaction, right?

Irrespective of the amount you're spending, it's always about the whole experience of let me, you know, take a look of what's more out there, you know, what more options are there in s- in a certain color, pattern, design, shape or size or whatever.

Kudrat (Host): An AI agent is designed to be efficient.

It wants to find something that fits your criteria and move on.

But shopping isn't always about efficiency.

Sometimes we want to browse. We want to see what's out there. We want to change our minds. We want to find something we didn't know we were looking for.

Kartik: I think that bit or, or that sort of color which a human being has to, to sort of their personality in terms of shopping, a agent may not really have that sort of color in their choices or in terms of probing more for option. They'll be like, "Yeah, this is a good fit. Let me just go ahead and, and buy it. Why sort of keep exploring, keep looking for options and wasting resources?" So I think that's gonna be like a tricky one to see how that plays out.

Kudrat (Host): And that is the tension beneath everything.

AI makes shopping faster and potentially more precise.

It removes the friction of searching, comparing, and second-guessing.

But friction isn’t always a problem.

It’s also where preference forms.

Where we notice what we actually like… change our minds… and sometimes find something better than what we set out looking for.

Hand the whole process over and you may lose the point of shopping itself.

Interestingly, Abhishek's own relationship with AI is starting to change.

Abhishek: I think I'm just trying to go to a more old-fashioned approach of talking to people and getting organic recommendations. It- I think it just makes things a little more personal, I guess. Like I, I, I, I've started to feel a little desensitized by the idea of, you know, going to an LLM and asking for recommendations when I could actually just put in all of that effort, speak to another person and be a little more specific.

Outro: That's all for today. You just heard The Signal Brief. We don't do hot takes. Instead, we bring you deep dives into the how and why of consumer trends. The Core produces The Signal Brief. Follow us wherever you get your favourite podcasts.

To check out the rest of our work, go to www.thecore.in.

If you have feedback, we'd love to hear from you. Write to us at feedback@thecore.in or you can write to me personally at kudrat@thecore.in.

Thank you for listening.