In this episode of Retaili$tic, John Harmon, Head of Technology Research delves into the transformative impact of agentic AI on the retail landscape, exploring how technologies like ChatGPT are reshaping consumer shopping experiences. He discusses the competitive dynamics between major retailers and the implications of AI-driven shopping, including trust issues and the future of personalized retail. The conversation also touches on the potential for agentic AI to revolutionize how consumers interact with products and make purchasing decisions, especially during the upcoming holiday season.
Video version of this episode is here
Takeaways
Agentic AI is set to revolutionize retail by enabling seamless transactions within chat interfaces.
The integration of AI in shopping is becoming more prevalent, with major retailers like Etsy and Shopify leading the way.
Consumer trust in AI-driven shopping experiences is crucial for widespread adoption.
The future of shopping will likely involve personalized agents that cater to individual preferences and needs.
AI can enhance the shopping experience by providing tailored recommendations and price comparisons.
Drones and advanced delivery systems are emerging as viable options for last-mile logistics.
The distinction between GenAI and agentic AI is significant, with the latter capable of executing actions based on user input.
Retailers may need to adapt their strategies as AI changes consumer behavior and expectations.
The upcoming holiday season is expected to see a surge in agentic shopping as consumers seek value and convenience.
Trust and security will be paramount as consumers share personal information with AI shopping agents.
Chapters
00:00 The Rise of Agentic AI in Commerce
03:39 Consumer Behavior and AI Shopping Dynamics
05:43 Retailer Integration and Market Competition
08:21 Logistics and Delivery Innovations
11:05 Understanding Agentic AI vs. GenAI
13:27 Personal Shopping Agents and Future Trends
16:00 The Role of IoT in Shopping
18:11 Dynamic Pricing and Consumer Trust
20:54 The Future of Shopping and Consumer Experience
Philip Moore (00:00)
Welcome to Retaili$tic, the official podcast of Coresight Research for October 28, 2025. This week, we have a fascinating conversation with John Harmon, our head of technology research, about the groundbreaking developments in agentic AI and e-commerce integrations into AI platforms like ChatGPT and Perplexity. But before we chat with John, Isla is here with her update on the new research publishing this week.
Isla Meldon (00:27)
Coming up this week, we're diving into some of the biggest stories shaping retail and technology. First, our latest playbook explores the fast-evolving world of generative and agentic AI, unpacking what these technologies mean for the future of retail and business strategy. Then, in our head-to-head series, we're putting two department store giants under the microscope, Kohl's versus Macy's, to see how they stack up in today's competitive landscape.
And we'll continue our US Consumer Survey Insights series, delivering fresh proprietary data on shopper sentiments and trends for the 2025 holiday season. From AI innovation to retail rivalries and consumer behavior, this week's reports have it all.
Philip Moore (01:09)
Thanks, Isla. Now let's bring in John.
Philip Moore (01:12)
welcome, John. Glad to have you back on Retaili$tic this I am really excited to have you because I feel like I need to get caught up. Like, what is going on with agentic AI? The headlines are all over the place, and I just need to, like, what is up?
John Harmon (01:26)
This will be in the history books going forward that the age of agenic commerce began in October 2025.
Philip Moore (01:34)
Wow. So, like, what are some of the things that are happening for those that aren't, like, up on it like you are?
John Harmon (01:40)
So consumers have been using, call them AI chat bots like ChatGPT for product search and discovery and information for quite some time. But now, starting this month, you can actually complete a transaction within ChatGPT. That means you can look for products, you can collect information, and you can hit a buy button and pay for them completely within ChatGPT.
Philip Moore (02:04)
Wow, so that's like frictionless one button from within program that more and more people are using all the time, like their primary computer interface almost.
John Harmon (02:07)
Hmm
Well, exactly. So this summer we had a couple summer interns, a high school student and a kid who's now a college sophomore. And for those guys, ChatGPT is their interface to the internet. They log on, they go to their browser and they go straight there. Whereas many of us, others may have gone to individual websites or used Google.
in addition to chat GPT, which is increasing in popularity.
Philip Moore (02:40)
So inside of chat GPT, if they find something and buy it, is that, know Walmart announced that that's where they're gonna be. Are there other retailers that are moving into chat GPT to make their inventories available through that interface?
John Harmon (02:55)
Yes, the first two retailers or marketplaces announced were Etsy, which operates a marketplace platform, and Shopify, which provides the platform for a lot of retailers, relatively small retailers to conduct commerce. So those were the first two. And some other marketplaces will be getting on board.
Philip Moore (03:14)
Okay, yeah Shopify is huge. they've got it's the long tail, but it's a long tail of retailers that are that sell through Shopify. So that's a lot of available inventory.
John Harmon (03:24)
Right, and since then Salesforce announced an integration with OpenAI as well and Salesforce there, it used to be called Commerce Cloud, now it's Agent Force Commerce, provides the e-commerce platform for a lot of large global retailers. So we've got the small ones, we've got the large ones, we've got Etsy, we've got Walmart, and many more to come.
Philip Moore (03:39)
Okay.
So this sounds like by this time next year, will literally be on par with Amazon for a channel for making purchases.
John Harmon (03:55)
It's definitely going to give Google some competition. Amazon wants to do its own thing. thus far, it's blocked the chat bots from accessing its website. So it looks like kind of a horse race with Google, OpenAI, and Amazon. But it's already here in one of our recent consumer surveys. 50 % of the consumers that we surveyed said for this holiday season,
they were either going to use AI tools frequently or they may use them occasionally. So I'm not really going out on a limb by saying this is going to be the holiday of chat GPT shopping.
Philip Moore (04:30)
Now, there other other than
chat GBT, are there other AI platforms that are moving in this direction too? Are you going to be able to buy something inside of perplexity or I mean, I know Elon Musk said he wanted X to be the everything platform. So obviously Grok, however Grok is integrated through X will have the opportunity to buy stuff. I mean, that's his already announced ultimate vision. But what about
perplexity or any of these other... Have you heard about any of them doing the same thing?
John Harmon (05:01)
Yes, I believe Perplexity has a shopping platform for its pro paid customers and US retail sales are $6 trillion. This is a big prize. It's logical for everyone to be moving in this direction. It's just that OpenAI got here early and got a lot of publicity around it.
Philip Moore (05:05)
Okay.
John Harmon (05:22)
ChatGPT is an extremely popular AI chatbot.
Philip Moore (05:26)
So what are the dynamics that the LLM platform, Perplexity or OpenAI or Grok or whoever, what are the dynamics, the kind of level of influence they have over the marketplace when the network effect kicks in and everybody's shopping through the LLM?
John Harmon (05:43)
Well, they already have had a lot of influence. Again, a lot of consumers have been using the LLMs for shopping to being able to buy anything due to the ease of use that it offers. And it was another firm put out some figures recently that said 20 % of Walmart's referral traffic was coming from ChatGPT. I think it was 15 % for Target.
even though no one could buy anything there. And you know, I want to just say this clearly. Well, a couple of things.
Even though I cover technology and I love technology and I'm generally an early adopter, I was kind of skeptical about using LLMs for shopping, probably because we spend so much time covering hallucination and governance and guardrails and model drift. But a friend showed me how to shop using ChatGPT. And my personal opinion, I don't own any ChatGPT stock because no one does. It's a nonprofit.
But the experience is extremely good. I was looking for a new backpack I put in. I'm looking for a slim backpack, a designer under $100. And it shows you several images of products. It summarizes some of the key selling points for each one, tells you what you should look for, and makes recommendations. You don't have to follow the recommendations. But it's a really rich experience that you get that...
is in my opinion superior to other search engines. And you know, what's new is not every retailer is connected with ChatGPT to be able to check out yet, but some are. And for some products, you got a buy button right there within ChatGPT, you hit it, you enter your credit card information, bang, you bought it and the information goes to the retailer to ship it and handle all of that. If the retailer isn't connected,
Philip Moore (07:27)
Mm-hmm.
John Harmon (07:30)
then there's a visit button, which unfortunately takes you out of GPT and dumps you on their website. But you're right there on the product page. We talk a lot about friction in retail. It's relatively low friction to shop this way.
Philip Moore (07:43)
Yeah.
I wonder if this makes more business for the USPS if people start moving some of their purchases off of Amazon and using Amazon delivery. And I know Walmart has their logistics set up pretty well and they're 90 % of the population is within 10 miles of the Walmart store. So they, they have already a footprint that lets their delivery logistics work. You know, maybe even better than Amazon, but
I guess for a lot of and the others, the Etsy's, they may be shopping through US Postal Service.
John Harmon (08:14)
It completely varies. You generally don't know which shipper you're going to get and it could be a combination of them. Amazon used to use USPS for lot of deliveries. Sometimes it can be a mix of both. Walmart has opened its delivery infrastructure platform to other retailers, but it's probably not widely known that Walmart uses Amazon for some delivery now.
Philip Moore (08:21)
⁓ interesting.
interesting.
John Harmon (08:38)
Amazon
offers these services fulfilled by Amazon, with Prime, where Amazon handles the packing and shipping for other retailers, including Walmart.
Philip Moore (08:49)
Well, that's a good thing. guess if they're cooperating, I know it's always frustrating to me when I see an Amazon truck come in the morning with something for my wife and then a UPS truck come a few hours later with something for my mother-in-law and then the mailman drops off a package. I'm like, if you guys could cooperate, you could save a lot of gas. Just make one trip with all three boxes.
John Harmon (09:10)
Well, and create a lot less pollution for the environment. If there are any entrepreneurs listening, I think there's a huge opportunity here because you can have multiple carriers coming to your house each day. And sometimes Amazon comes more than once, I think, because the goods are coming out of different fulfillment centers. Wow, there's an opportunity to optimize that, to make fewer deliveries, ⁓ fewer trips.
Philip Moore (09:13)
Right? Exactly.
And I live out in the country,
it's not a trivial thing to be making three or four trips out here. It's a pretty long turnaround. I'm always surprised when we get stuff the next day, just because we're not that close to a distribution center. So it's pretty amazing what they've got done, but it does seem like it's pretty wasteful.
John Harmon (09:55)
Well, I really don't intend to be doing an Amazon commercial. Amazon has introduced a couple of new things. Like you can pick your quote Amazon day. I mean, for me, it's always Saturday where you can have everything delivered one day a week and they'll give you a 2 % discount for having things delivered on your day. But also Amazon has announced they're investing a lot to broaden their delivery infrastructure.
to cover rural areas better, to be able to offer the fast shipping that people in cities get.
Philip Moore (10:25)
And I know, I mean, it's a whole nother. I don't know that it necessarily ties into agentic AI, but the drones now with Last Mile is starting to pick up too. I've seen some really interesting stories about drone deliveries. So I mean, that's sort of a few years from now, it'll be click a button and chat GPT and a few hours later, a drone drops a box on your roof.
John Harmon (10:46)
We've been talking about drone deliveries for years. Jeff Bezos jumped the gun five, six years ago talking about Amazon Air on 60 Minutes. But in the meantime, there have been some pilot programs of drone deliveries by Amazon and Walmart testing in some cities. I think there are issues with it, but I think it's...
appropriate for some things and drones have proven themselves in a lot of applications like surveilling outdoor pipelines and delivering medicine to people in remote locations. ⁓
Philip Moore (11:17)
Yep. I saw
a documentary about the company that I guess the government of Uganda, an American company that the government of Uganda sort of sponsored to do blood deliveries to their hospital network. So they went over to Uganda and built out there was sort of their prototyping stage and they built out their whole system over there and perfected it. And now they're bringing it back to the U S cause they got it working really tight. I can't remember the name of the company, but it's one that
John Harmon (11:27)
you
Philip Moore (11:45)
you know, uses launchers, they're winged drones, I guess they use launchers and then they drop the blood with parachutes or something and come back and it's highly efficient. They can go 60 miles on an electric charge or something because of the way they launched the drones and they don't have to land them. They just go to where they're supposed to go and drop a paper parachute with a box of blood out of the bottom of
John Harmon (11:46)
you
Especially
in developing countries with poor infrastructure, maybe where the roads aren't good or they don't go everywhere, drones can travel everywhere.
Philip Moore (12:09)
Right. Exactly. Yeah.
So, and we probably should have started our conversation with this, but the difference between Gen.ai and a Gentic.ai. I want to just touch that real quick.
John Harmon (12:23)
Sure,
Agenic AI builds on GenAI. It builds on large language models which interpret what you want it to do and translate it into actions. Agenic AI does include other features like the ability to reason. So once it generates an answer or an action, it checks it to make sure it's really the right thing to do. I mean, the one big difference is Agenic AI
can take action on your behalf. It can forecast scenarios or create estimates or make recommendations. Some of the language models can kind of do that. And if the humans let it, it could execute those actions. with the genetic AI in the early stages, it's best to have a human in the loop, let the human push the button to do the ultimate action.
But in the case of aogenic shopping, again, the agents are executing the transaction for the person once you press the buy button. And again, making recommendations and searching for products for you and putting it all together.
once you tell it to do that. Now.
Philip Moore (13:27)
That
sounds like a
interesting demo someday we should probably go over a shopping trip using all agents like how little input can you give it and actually end up with what you want.
John Harmon (13:40)
Well, at this point, is kind of like an AI chatbot conversation, right? Because you tell it what you're looking for and what you like, and it makes some suggestions. And it can ask you questions to refine the search or the suggestions. I have this science fiction view of agents, though, where shoppers have their own personal agents. I mean, I'll give you, here's an example.
They're hard to find items that I'm always looking for. So to find them, I would have to go on search engines or bang on marketplace websites continually until these things pop up. an agent, my agent, could work on my behalf and be out there in the internet and be constantly looking and then notify me when these items become available. Another example is, ⁓
We talk in retail about personalization all the time and it really seems like an elusive goal because it's not there yet. Some retailers are much better than others, but other retailers send me daily blast emails for things I don't want. And with other retailers, I go to their app or their website and I see the things that I buy every day or that I like. there are promotions, they're on sale and the retailer didn't tell me.
I think another kind of shopping agent, you I can tell it the things I like and it could just look for them to go on sale and let me know. This could be done by the retailer or it could be done by the consumer. But I'm optimistic that we citizens, people, adults, and consumers can have our own agents that do things for us as well.
Philip Moore (15:09)
Seems like with the new IoT and fridges with sensors, you could actually develop a personal demand forecasting agent for your refrigerator that would measure the pace with which food goes in and out of a particular type, the frequency that it's there, and automatically make sure that you have your Instacart order include the thing that's about to run out so that you never...
You never run out of your whatever that you normally eat.
John Harmon (15:38)
That technology is really here today. But I don't know if we're ready for it. I mean, there are a couple ways to do it. Using traditional machine learning. I don't know, say you use a gallon of milk every two weeks. If you buy a gallon of milk every two weeks, pretty easy to forecast you're going to probably need a gallon of milk every two weeks based on your historical patterns. But there's another step, a lot of the...
advanced refrigerators have the capability you can put a camera inside and monitor your milk and detect when you're getting low and ask, you want to order milk in a day or two? And or suggest recipes based on what you've got in the fridge.
Philip Moore (16:18)
Or with some of these new Avery Dennison cold and wet resistant RFID labels, if you had the right sensor in the door of your fridge, it would know every time you pulled something out and put it back in if it read the RFID on the label of the food. Assuming it was packaged goods, I guess. The camera works for everything and the RFID only works for things that have tags on
John Harmon (16:39)
Well, it's interesting you say that I haven't heard that use case recently, but Avery Genison views food as a huge opportunity. ⁓ They demonstrated at the NRF Big Show at the beginning of 2025. They're working with Kroger to put RFID tags on bakery items so that the store associate could scan the shelf and figure out which loaves of bread are about to expire.
Philip Moore (16:45)
Mm-hmm.
John Harmon (17:01)
put them on sale or pull the ones that have expired, you know, so that no one buys them and figure out what to do with it. And Avery Geneson announced just this week, they're working with Walmart. Walmart's been a huge fan of RFID. They've expanded the use of RFID a couple of years ago to include all kinds of items, you know, even down to like an 83 cent box of crayons. Walmart has put RFID tags on it.
And it makes sense to put an RFID tag on a steak, which is a high value item, but then you know when it's about to go bad, then maybe you could make hamburger out of it and cook it, or again, figure out which ones are expiring sooner and put them at the front of the shelf. So we're really seeing RFID expanding through grocery, which can just do a lot of things to reduce food waste.
Philip Moore (17:48)
mean, the car business is the same thing. If a particular car, either new or used, hasn't had a test drive in a week, they automatically reduce the price. So it's the same. If your stake's been sitting there an extra hour longer than you think it should, could automatically, particularly with electronic shelf labels, could automatically reduce the price on the stake to try and move it out of inventory.
John Harmon (18:11)
Absolutely. Yeah, I think that's a great application. People are a bit nervous about electronic shelf labels thinking that grocers are going to implement surge pricing and gouge them. But the the grocery retailers are using dynamic pricing so far, they're only bringing prices down. You know, I know that a bakery where cookies are half off after 3pm, you can do lot of interesting things like that or yeah, discount.
food that's on its way towards expiration.
Philip Moore (18:37)
Exactly. I'm sure grocers, just like any other retailer starts high and then brings it down to move it out of inventory. That's sort of retailing 101. you're usually, if you're finding something on a discount, it's because it's been there for a while or there's too much in the warehouse and they know they've got to turn it over before the next season or whatever. So yeah. The idea that people would increase prices using electronic shelf labels is a little
John Harmon (18:47)
Amen.
Philip Moore (19:03)
a little unusual. Like they have lost leader items, right? So they may have something that gets people in the store, but the idea that it's, know, because people come at 6.30 at night, they're gonna increase the price from 6.30 to 7.30 seems a little odd.
John Harmon (19:05)
Well this is, this is, is... Right.
Absolutely true. Yeah, you would probably want to reduce things at the end of the day to get them out of the store, but it's not this is Not going to be popular for me to say it but maybe Retailer should raise prices sometimes. I mean, you there are they have competitors their supply and demand dynamics They do see price increases from their suppliers. We see dynamic pricing already in airline tickets and
Philip Moore (19:24)
Yeah, yeah.
John Harmon (19:43)
hotel rooms, you go on amazon.com and products can have different prices at different times during the day for testing purposes and again for supply and demand. Grocers have notoriously thin margins like 2 % net margins. A little bit of price controls would probably help them. Unfortunately, it's become a political issue.
You know, with the inflation that we've seen over the last several years, a lot of it has hit the food industry. We don't really see it in grocers' margins, but there are some politicians here in the U.S. have, you know, really tried to make hay with that.
Philip Moore (20:18)
Yep. And you're in New York, so you know exactly what that looks like with your new mayor. So anyway.
John Harmon (20:25)
Yes, I'm looking forward to the city owned grocery stores. I think they're gonna work really well. He's not our new mayor yet. Early voting starts tomorrow in New York City.
Philip Moore (20:34)
Okay.
Well, that's exciting. Well, the podcast comes out on Tuesday. early voting starts, ⁓ three days ago. okay. That's fun. I think the, it's, it's interesting to just contemplate what it's going to be like to be a shopper a year and a half from now. I mean, it's just seems like things are changing so fast.
John Harmon (20:39)
Hahaha
Philip Moore (20:55)
I guess there'll always be people. One other interesting dynamic is there'll always be people on different places on the adoption curve, the AI adoption curve. And like you mentioned earlier, I think, you know, there's a whole generation that's grown up in it and they're already at the mass adoption level. And then there's other people coming along later. So that may also have some dynamic implications for what products are moving on chat GPT.
one button by the, you know, the gen alphas and the gen z's and the early millennials are already there and the gen x's and boomers are trailing on the adoption curve. It could be interesting in terms of our early reads on what kinds of products are sold through those channels.
John Harmon (21:39)
Not to mention these language models, they take the prompts, the queries that people type into them and feed them into the models as well. So think about these, TikTok generation who are watching 10 second videos and we see these micro trends that emerge, these, I call it fads, right, for things. They really could skew the language models. It can be, it'll be really interesting to see.
Philip Moore (22:03)
Mmm.
John Harmon (22:04)
what happens once people start shopping like that. But we have to remember that in-store sales are still about, slightly less, 80 % of US retail sales. So really, we're talking about the 20 % that happens online, which is the, it's the sizzle, it's the fastest moving part, it's the sexy part. But not the whole picture.
Philip Moore (22:23)
So do these just in terms of the...
What are we calling it now? We used to call it Omni, but now it's something else. It's not Omni channel anymore. What do we call it now? unified commerce. That's right. So what do you think this new agentic AI does there? Because I know even, you know, when
John Harmon (22:33)
Unified commerce. It took me a minute to remember that.
you
Philip Moore (22:43)
When the cell phone started bringing Amazon into the retail space, it introduced this whole idea of showrooming. Does this accelerate that even more or is this more of the same or is there anything dynamically fundamentally different about the AI e-com versus the marketplace e-com?
John Harmon (23:01)
You know, this AI chat bot shopping, it's really just another channel because on one hand it works similarly to how people shop with social media today, right? You see something on social media, sometimes you can click a buy button there, but other times it may send you to the retailer's website. It kind of works like that in a way. But I think way people shop is changing because the way we've shopped before a month ago,
You would go to the website, the retailer's homepage, and you type what you're looking for in the search window, and then it would show you some product pages, and you'd click on the one you're interested in and maybe buy the item. But now, while one of two things, either the chat bots will take you straight to the product page, you'll be bypassing the homepage. That has implications for retail media, because you're not going to, the eyeballs aren't going to be on.
the media on the homepage, but also the retailer is not gonna get that search data like before. not gonna have as good a picture of what consumers are searching for. I mean, it kind of relegates the retailer to just.
being the inventory provider in a way, ⁓ really gonna change a lot of things in shopping. Also, I saw a presentation a couple of weeks ago from this fraud prevention firm, Forder. New technologies bring new opportunities for criminals, right? You're gonna see people finding gaps and loopholes in the chat bots to steal money from people or steal the goods or do something else.
Philip Moore (24:07)
Interesting.
John Harmon (24:28)
else. consumers are going to have to think about do I trust this chatbot with my credit card number. I mean it goes through a secure form, right? They don't see it these days. But what about my loyalty information or my personal information? I talked about chatbots that are looking for things that you may want to buy. You'll be entrusting them with that information and hope that they don't get hacked or it gets out or it gets leaked somehow. There are a lot of really interesting
issues that are going to come out in this new world of agenic shopping.
Philip Moore (24:53)
Mm.
Yeah. So if you set up an agent to automatically search Craigslist, Facebook marketplace, Etsy, Amazon, wherever else for some particular item that you're would otherwise be searching the internet yourself for, you're saying that bot itself can be
hacked and interpreted and associated with you so everybody knows you're looking for this specific thing.
John Harmon (25:22)
Well, we've essentially seen that basically everything can be hacked. The US government can be hacked, the telecom carriers can be hacked, several retailers have been hacked, hacked, cloud service providers have been hacked. puts an additional burden on the LLM providers, right? They're really good at LLMs and AI. Are they as good in cybersecurity? It's not their main focus, but it is, you know, of...
Philip Moore (25:25)
Okay. All right.
John Harmon (25:46)
extremely high importance.
Philip Moore (25:48)
Yeah, I guess one big breach and then people lose trust and then you're, you know, in a, a highly competitive environment where Google's there and Amazon's doing their thing. Apple's trying to pull something together. Grok is there. And then if chat GPT has a big data breach and everybody figures out that John was looking for a 1947 Corvette or what, don't know what, you're looking for.
John Harmon (25:55)
Yeah
you
Yeah, it's a Rolling Stones concert t-shirt. But it's interesting. this dates us, but back to Internet 1.0, around 2000, around that time, a lot of consumers a bit leery entering their credit card information on a website, right? But fast forward to today, and we think nothing of it. But I think the thing about chat bots is...
Philip Moore (26:15)
there you go.
John Harmon (26:39)
Because they're so personalized and they're so conversational, think, you know, people are more likely to have trust with them, which could open the door to a lot of issues, But I think, but I think, yeah, because you feel like it's talking just to you, you're more likely to believe, you know, what it tells you. We know that that large language models have hallucinations that they do their best to give you the
Philip Moore (26:48)
Okay.
John Harmon (27:02)
the best answer they can, there no guarantees that it's correct or...
Philip Moore (27:05)
When they, so a shopping agent who is gonna go out and search for stuff for you, theoretically could also be reading all of the reviews. I know that's the other shopping step that I always feel guilty when I skip when I buy something is, do you go and read 500 reviews of this product from wherever you can find them before you make the purchase? But the agent's gonna do that, right? The agent can go out and do that easily.
John Harmon (27:29)
Well, that's an easy job for LLMs. you don't need an agent, but the agent can activate the LLMs to go out and read all the reviews and give you a summary of them or even evaluate the quality of them.
Philip Moore (27:31)
Right.
Yeah, that that in and of itself might be the biggest reduction in friction. Like if you were out there telling me go find the best X, Y or Z and look at a thousand reviews and give me a, you know, use the use the reviews on all of the competing products from actual people. to the degree that actual people can be identified anymore on the Internet. But yeah, that's that seems like a.
John Harmon (28:00)
you
Philip Moore (28:06)
handy tool to use when you're shopping.
John Harmon (28:09)
I mean, I would say that's already happening today on Amazon when you look at a product. It gives you an analysis of the reviews. Like, you know, most people liked it, or these were the most common problems that people had with it, or this product was highly, has been returned frequently. So it kind of does give you a way to benchmark the reviews, because we know there are fake reviews out there too.
Philip Moore (28:30)
Right. Seems like the LLMs might be able to even score the reviews they read for the probability that it was real versus fake. like you could put some kind of threshold and say analyze all the reviews for the probability that it was an actual human that wrote the review, a sincere human or whatever prompt you wanted to put in the LLM and then
eliminate anything that's below 50 % probable that it's really a human being and do the analysis based only on those reviews that you've flagged as most probable a sincere human being.
John Harmon (29:05)
I think that's feasible today. The fake ones are probably really short reviews based on the language they use. don't know, this is the best toothbrush I've ever seen in my whole life, you know.
Philip Moore (29:07)
Yeah.
Three words and a five stars not good. Well, great conversation as usual any other parting top of mind things we should all be thinking about this week as we consider the future of our of our world.
John Harmon (29:17)
Hahaha
Well,
I think this will be the holiday of eugenics shopping. Let's see how it turns out. It seems that consumers are ready, willing, and able to use LLMs. Certainly this economic environment, are a lot of consumers who are looking for value or to save money. These LLMs give you another means for performing price comparison, right? You can certainly put in...
this really hasn't been possible before, but you could certainly give it your budget for a present and it'll search around and find it for you. I searching multiple retailers in a very sophisticated way is new. So I think consumers are in a good position to use it this holiday season.
Philip Moore (30:07)
Interesting.
So I tell
the LLM my granddaughter is seven years old, her favorite color is purple, I want to spend $200, what's the best gift to buy her? And then it'll go off and find things and make suggestions. Yeah.
John Harmon (30:23)
Bingo, try it out
after this. I mean, that can be done. I did this example, I was looking for a new backpack. I mean, I mentioned it. I'm looking for a slim, fashionable designer backpack under $100. And look, five or six examples right there and a lot of information.
Philip Moore (30:39)
That's exciting. Sounds like.
John Harmon (30:41)
It
really provides a a richer experience for consumers.
Philip Moore (30:45)
Yeah, even for people who don't typically like to shop for stuff. This could be fun.
John Harmon (30:49)
Especially for us.
Philip Moore (30:50)
All right. Thanks
again, John. It's always a pleasure. We'll look forward to your next visit.
John Harmon (30:56)
Same here, good to be here again, bye.
Philip Moore (30:57)
Thanks, John, and thank you for joining us this week on Retaili$tic. Coresight Research serves our members with industry trends, consumer insights, and data resources. If you need research and data from the intersection of retail and technology, visit us at coresight.com. Have a lovely day, and we'll see you next week.