1 00:00:15,396 --> 00:00:15,826 Speaker 1: Pushkin. 2 00:00:20,316 --> 00:00:23,476 Speaker 2: Hello, hello, Malcolm Gladwell here. Over the past several months 3 00:00:23,616 --> 00:00:27,376 Speaker 2: across revisionist history and business history, we've been exploring the 4 00:00:27,496 --> 00:00:31,476 Speaker 2: idea of trust, where it comes from, the moment it's earned, 5 00:00:32,016 --> 00:00:35,396 Speaker 2: and where we decide to place it. This episode is 6 00:00:35,436 --> 00:00:40,226 Speaker 2: sponsored by PayPal, trusted by consumers, built for all business. 7 00:00:40,866 --> 00:00:44,066 Speaker 2: For over 25 years, they've been at the forefront of commerce, 8 00:00:44,596 --> 00:00:48,776 Speaker 2: helping merchants of all sizes grow. So wherever commerce moves next, 9 00:00:49,136 --> 00:00:52,676 Speaker 2: you can rest assured PayPal has your back. And now 10 00:00:52,836 --> 00:00:55,276 Speaker 2: we want to leave you with one final discussion on 11 00:00:55,316 --> 00:00:59,806 Speaker 2: this special episode of What's Your Problem? Today, Jacob Goldstein 12 00:01:00,106 --> 00:01:04,166 Speaker 2: hosts a conversation about how consumers and companies are trying 13 00:01:04,206 --> 00:01:07,576 Speaker 2: to figure out trust in the era of agentic AI. 14 00:01:08,296 --> 00:01:08,796 Speaker 1: Let's listen. 15 00:01:20,556 --> 00:01:22,876 Speaker 3: I'm Jacob Goldstein, and this is What's Your Problem? My 16 00:01:22,916 --> 00:01:26,746 Speaker 3: guest today is Kartik Hosanagar. Kartik is a professor of 17 00:01:26,946 --> 00:01:30,306 Speaker 3: operations and marketing at the Wharton School at the University 18 00:01:30,346 --> 00:01:33,646 Speaker 3: of Pennsylvania. And I wanted to talk with him because 19 00:01:34,146 --> 00:01:37,146 Speaker 3: we are at this moment when the way people use 20 00:01:37,226 --> 00:01:40,476 Speaker 3: AI is shifting. We're starting to go beyond just having 21 00:01:40,616 --> 00:01:44,946 Speaker 3: AI answer questions and toward having AI actually do stuff 22 00:01:45,086 --> 00:01:49,146 Speaker 3: in the world for us, including having AI agents go 23 00:01:49,286 --> 00:01:53,306 Speaker 3: out onto the internet and buy stuff. Karthik has been 24 00:01:53,366 --> 00:01:55,926 Speaker 3: studying this shift. He's been trying to figure out how 25 00:01:55,966 --> 00:01:59,706 Speaker 3: both customers and merchants can solve the problems that arise 26 00:01:59,826 --> 00:02:03,246 Speaker 3: in the era of shopping with agentic AI. What will 27 00:02:03,286 --> 00:02:05,906 Speaker 3: it take for customers to trust AI agents to make 28 00:02:05,986 --> 00:02:09,106 Speaker 3: purchases on their behalf? And what do merchants need to 29 00:02:09,166 --> 00:02:11,426 Speaker 3: do to prepare for a world where some of their 30 00:02:11,506 --> 00:02:15,506 Speaker 3: customers are AI agents? In the first part of the conversation, 31 00:02:15,686 --> 00:02:18,346 Speaker 3: Karthik and I talked about what this all means from 32 00:02:18,366 --> 00:02:21,106 Speaker 3: the consumer's point of view. In the second part, we 33 00:02:21,146 --> 00:02:22,946 Speaker 3: talked about what it means for merchants. 34 00:02:23,426 --> 00:02:28,276 Speaker 1: You know, I will often begin my shopping research, having 35 00:02:28,436 --> 00:02:31,016 Speaker 1: LLMs do the research for me. I'll send them on 36 00:02:31,196 --> 00:02:35,676 Speaker 1: a deep research task and have them do all of that. 37 00:02:35,716 --> 00:02:39,016 Speaker 1: I'll have them synthesize it for me. In some cases, 38 00:02:39,096 --> 00:02:41,736 Speaker 1: I might be patient enough to read that long report. 39 00:02:41,776 --> 00:02:45,076 Speaker 1: Sometimes I'll say, convert this into a table and just 40 00:02:45,136 --> 00:02:47,776 Speaker 1: give me like, you know, what I need to know. 41 00:02:48,076 --> 00:02:50,996 Speaker 3: This is pretty hardcore shopping. Like what kind of item 42 00:02:51,136 --> 00:02:54,816 Speaker 3: are you having deep research reports written for? Tell me 43 00:02:54,836 --> 00:02:55,576 Speaker 3: it's a t-shirt. 44 00:02:58,196 --> 00:03:00,596 Speaker 1: Yeah, well, you know, sometimes it's like, you know, I'm 45 00:03:01,316 --> 00:03:04,536 Speaker 1: looking to buy a car right now and trying to 46 00:03:04,556 --> 00:03:10,936 Speaker 1: figure out what new electric vehicles are around. But honestly, 47 00:03:10,976 --> 00:03:14,836 Speaker 1: I don't even know the research about whether electric vehicles 48 00:03:15,556 --> 00:03:19,456 Speaker 1: are as environmentally friendly as originally touted to be because 49 00:03:19,476 --> 00:03:22,836 Speaker 1: there's all this criticism that, hey, there's all this rare 50 00:03:22,856 --> 00:03:25,596 Speaker 1: earth mining and the environmental damage of that is very high. 51 00:03:25,636 --> 00:03:28,676 Speaker 1: So I'm not even saying which car should I buy. 52 00:03:28,716 --> 00:03:31,456 Speaker 1: I start with, tell me The pros and cons of 53 00:03:31,496 --> 00:03:34,316 Speaker 1: electric vehicles. In what sense is it good for the 54 00:03:34,376 --> 00:03:36,956 Speaker 1: environment and what sense isn't? And in what sense is 55 00:03:37,016 --> 00:03:39,976 Speaker 1: it convenient to own one and in what sense it 56 00:03:40,056 --> 00:03:43,416 Speaker 1: is not? And then from there, it would be, hey, 57 00:03:43,516 --> 00:03:45,976 Speaker 1: I've seen on the streets people driving Rivian. What do 58 00:03:45,996 --> 00:03:50,256 Speaker 1: you think about Rivian? Is that any good? At times, 59 00:03:50,336 --> 00:03:53,616 Speaker 1: it may not be deep research, but nonetheless, you know, 60 00:03:53,646 --> 00:03:55,406 Speaker 1: I have a startup as well that I started a 61 00:03:55,426 --> 00:03:59,786 Speaker 1: year ago. We're trying to make decisions for the startup like, hey, which. 62 00:04:01,236 --> 00:04:05,596 Speaker 1: Software should I use for cap table management? Which software 63 00:04:05,616 --> 00:04:09,456 Speaker 1: should I use for payroll? Which vendors? Who's right for 64 00:04:09,496 --> 00:04:12,676 Speaker 1: a company our size? We are very small. We're just 65 00:04:12,736 --> 00:04:16,956 Speaker 1: about 10, 12 people. Who's right for California? Who's right 66 00:04:16,996 --> 00:04:18,886 Speaker 1: if we grow to 100 people and so on? 67 00:04:19,386 --> 00:04:23,846 Speaker 3: I do feel like this distinction between using AI, using 68 00:04:23,906 --> 00:04:26,756 Speaker 3: a language model to do research for shopping and and 69 00:04:26,996 --> 00:04:31,096 Speaker 3: actually handing off the purchase decision to the agent is 70 00:04:31,156 --> 00:04:35,016 Speaker 3: really important right now. Like I basically trust the AI 71 00:04:35,056 --> 00:04:38,256 Speaker 3: to give me information about, you know, product reviews and 72 00:04:38,296 --> 00:04:41,996 Speaker 3: that sort of thing. And I'm curious, like when you 73 00:04:42,156 --> 00:04:45,596 Speaker 3: see people who are using AI on the research side, not, 74 00:04:45,636 --> 00:04:47,276 Speaker 3: you know, we'll get back to agentic in a second, 75 00:04:47,316 --> 00:04:49,516 Speaker 3: but on the research side, like how are they using 76 00:04:49,556 --> 00:04:51,256 Speaker 3: it differently than traditional search? 77 00:04:53,026 --> 00:04:56,506 Speaker 1: Well, I think the biggest difference with traditional search is 78 00:04:56,606 --> 00:05:01,706 Speaker 1: that the search engines are merely curating the information and 79 00:05:02,006 --> 00:05:07,906 Speaker 1: then you are taking the trouble to synthesize it. And 80 00:05:07,986 --> 00:05:12,446 Speaker 1: you get exposed to a lot of information about various brands, 81 00:05:12,566 --> 00:05:15,206 Speaker 1: even if you decide not to eventually buy from them, 82 00:05:15,226 --> 00:05:18,466 Speaker 1: or even if it's not in your shortlist, you still 83 00:05:18,486 --> 00:05:23,096 Speaker 1: get exposed to a lot. Yeah. In contrast, when it 84 00:05:23,156 --> 00:05:28,056 Speaker 1: comes to AI for shopping, I'm outsourcing a lot of 85 00:05:28,106 --> 00:05:32,246 Speaker 1: the synthesizing. And then depending on the type of buyer, 86 00:05:32,846 --> 00:05:37,626 Speaker 1: the effort level varies where some might say, synthesize it 87 00:05:37,666 --> 00:05:40,076 Speaker 1: for me. And then there might be a lot of 88 00:05:40,176 --> 00:05:42,436 Speaker 1: back and forth, but what about this? What about that? 89 00:05:43,236 --> 00:05:46,676 Speaker 1: And in other instances, you might outsource the decision itself 90 00:05:47,076 --> 00:05:51,376 Speaker 1: in terms of, bring it down to the short list. 91 00:05:51,436 --> 00:05:54,036 Speaker 1: And so you're starting off with a short list of 92 00:05:54,096 --> 00:05:56,596 Speaker 1: three brands and a comparison table, and that's where the 93 00:05:56,656 --> 00:05:59,946 Speaker 1: human work begins. A lot of it is based for 94 00:06:00,026 --> 00:06:03,546 Speaker 1: me in terms of the risk of making a wrong decision. 95 00:06:03,826 --> 00:06:05,606 Speaker 1: And if the risk is high and I'm spending a 96 00:06:05,646 --> 00:06:08,166 Speaker 1: lot of money, then I'll actually go back and forth 97 00:06:08,206 --> 00:06:10,886 Speaker 1: with the LLMs. But if the risk is low, then 98 00:06:10,946 --> 00:06:15,516 Speaker 1: I'm pretty much, for the most part, outsourcing everything except 99 00:06:15,576 --> 00:06:17,756 Speaker 1: the final approval decision to the LLM. 100 00:06:18,136 --> 00:06:21,356 Speaker 3: And that seems pretty common, right? It seems like, by 101 00:06:21,416 --> 00:06:23,816 Speaker 3: and large, a lot of people are comfortable using AI 102 00:06:23,856 --> 00:06:26,746 Speaker 3: for doing the research, but they're a lot more hesitant 103 00:06:26,786 --> 00:06:30,326 Speaker 3: to let the agent take that last step. They're wary 104 00:06:30,386 --> 00:06:35,226 Speaker 3: of letting AI actually buy the thing. And, you know, 105 00:06:35,386 --> 00:06:38,396 Speaker 3: it reminds me a little bit of the early days 106 00:06:38,456 --> 00:06:41,256 Speaker 3: of the web, basically the 90s, which I am old 107 00:06:41,296 --> 00:06:43,896 Speaker 3: enough to remember. And, you know, you would go and 108 00:06:44,136 --> 00:06:46,956 Speaker 3: companies had websites and you'd look at stuff. But the 109 00:06:46,996 --> 00:06:50,416 Speaker 3: idea of putting your credit card into the web was like, no, 110 00:06:50,496 --> 00:06:52,516 Speaker 3: there is no way I'm going to put my credit 111 00:06:52,556 --> 00:06:56,326 Speaker 3: card into some random website. And of course, you know, 112 00:06:56,506 --> 00:06:58,706 Speaker 3: PayPal is in fact an example of one of the 113 00:06:58,786 --> 00:07:02,986 Speaker 3: early pioneers here. They were navigating these shifts in real time. 114 00:07:03,446 --> 00:07:07,646 Speaker 3: But now, whatever, 25, 30 years later, we have solved that. 115 00:07:07,926 --> 00:07:11,006 Speaker 3: Everybody buys stuff online all the time. And, you know, 116 00:07:11,676 --> 00:07:13,676 Speaker 3: I would say at this moment, I am not ready 117 00:07:13,736 --> 00:07:16,736 Speaker 3: to just have agentic AI go off and buy stuff 118 00:07:16,976 --> 00:07:21,166 Speaker 3: for me without my input. I could imagine getting there. 119 00:07:21,186 --> 00:07:24,546 Speaker 3: I just don't trust it yet. And so I'm curious, 120 00:07:24,966 --> 00:07:27,646 Speaker 3: you know, we've solved the trust problem before. How do 121 00:07:27,686 --> 00:07:29,946 Speaker 3: we solve it this time? Like what has to happen 122 00:07:30,306 --> 00:07:33,176 Speaker 3: to get to the point where consumers trust agents to 123 00:07:33,276 --> 00:07:36,136 Speaker 3: buy stuff on their behalf, on our behalf? 124 00:07:36,716 --> 00:07:40,256 Speaker 1: Yeah, I think there's a few things that, need to happen, 125 00:07:40,336 --> 00:07:45,456 Speaker 1: some of which is probably the tech infrastructure needed to 126 00:07:45,516 --> 00:07:50,016 Speaker 1: support something like this. The second is a behavior change. 127 00:07:50,276 --> 00:07:55,986 Speaker 1: People doubted in 1998, 1999, whether it would ever happen 128 00:07:56,026 --> 00:07:58,986 Speaker 1: that we would put our credit cards and buy. And 129 00:07:59,006 --> 00:08:01,246 Speaker 1: there were others who said, no, no, no, it'll happen 130 00:08:01,286 --> 00:08:03,566 Speaker 1: next year. And there were folks who said, no, it'll 131 00:08:03,586 --> 00:08:07,826 Speaker 1: happen in five years. And so that's the speculative piece, 132 00:08:07,906 --> 00:08:10,956 Speaker 1: when it happens and for what it'll happen and so on. 133 00:08:11,536 --> 00:08:14,896 Speaker 1: I remember back in the day, I had my own 134 00:08:14,956 --> 00:08:17,596 Speaker 1: students who, you know, I teach at Wharton. So my 135 00:08:17,636 --> 00:08:20,466 Speaker 1: students at Wharton who started the company Warby Parker, where 136 00:08:20,486 --> 00:08:24,926 Speaker 1: you buy eyeglasses from, when they first came and actually 137 00:08:24,966 --> 00:08:28,746 Speaker 1: their term paper in my course was a business plan 138 00:08:28,786 --> 00:08:30,986 Speaker 1: for Warby Parker. And at the end of the semester, 139 00:08:31,106 --> 00:08:34,406 Speaker 1: I chatted with them. And to me, it felt like 140 00:08:34,606 --> 00:08:36,816 Speaker 1: at that point when I chatted with them, it was like, 141 00:08:36,956 --> 00:08:41,376 Speaker 1: I can see some people buying their eyeglasses online. But 142 00:08:41,736 --> 00:08:44,276 Speaker 1: I think when you're buying eyeglasses, you're wearing one, I'm 143 00:08:44,296 --> 00:08:46,436 Speaker 1: wearing one. I feel like I want to try it on. 144 00:08:46,456 --> 00:08:49,536 Speaker 1: I want to see how it feels. And that is 145 00:08:49,606 --> 00:08:52,886 Speaker 1: extremely important. So I kind of my reaction was, it 146 00:08:52,966 --> 00:08:55,666 Speaker 1: makes sense, but I still don't see how it's mainstream. 147 00:08:56,616 --> 00:08:57,366 Speaker 1: But they changed it. 148 00:08:57,416 --> 00:09:00,606 Speaker 3: Well, they sent you the glasses. The thing of sending 149 00:09:00,636 --> 00:09:03,026 Speaker 3: you the tester pair, was that because of your feedback? 150 00:09:03,406 --> 00:09:04,426 Speaker 3: Did you get equity? 151 00:09:05,186 --> 00:09:07,786 Speaker 1: No, I didn't. I'm sure others brought it up to them. 152 00:09:07,866 --> 00:09:09,706 Speaker 1: I won't take credit for that, but I will give 153 00:09:09,746 --> 00:09:13,926 Speaker 1: them credit for kind of saying, we hear you on 154 00:09:13,966 --> 00:09:18,456 Speaker 1: the problem, but we disagree that there is no solution 155 00:09:18,496 --> 00:09:22,156 Speaker 1: to the problem. And so that's the part that I 156 00:09:22,176 --> 00:09:26,166 Speaker 1: was saying is speculative in terms of, when it will 157 00:09:26,186 --> 00:09:29,686 Speaker 1: happen and what needs to happen for that full behavior 158 00:09:29,746 --> 00:09:34,106 Speaker 1: change to happen. I think someone will crack it. And 159 00:09:34,186 --> 00:09:36,936 Speaker 1: I think it'll happen in phases just like e-commerce did. 160 00:09:37,126 --> 00:09:39,716 Speaker 1: We first felt comfortable buying books and we said, hey, 161 00:09:40,056 --> 00:09:42,136 Speaker 1: a book is a book. I don't need to worry 162 00:09:42,176 --> 00:09:44,136 Speaker 1: about how it feels in my hand and so on. 163 00:09:44,916 --> 00:09:48,496 Speaker 1: And then it became electronics. Hey, I bought books for $ 20. 164 00:09:48,036 --> 00:09:52,286 Speaker 1: Maybe I can buy a $ 200 or $ 500 electronics online 165 00:09:52,326 --> 00:09:55,236 Speaker 1: and it's going to be okay. And then it became clothing. 166 00:09:55,296 --> 00:09:57,656 Speaker 1: And then it just kept expanding. And so I think 167 00:09:57,696 --> 00:10:00,996 Speaker 1: it's going to be similar with agentic where there are 168 00:10:01,036 --> 00:10:04,886 Speaker 1: some categories which will be easier than others. And there'll 169 00:10:04,926 --> 00:10:07,986 Speaker 1: be some customer segments where it'll be easier than the others. 170 00:10:08,146 --> 00:10:11,006 Speaker 1: And that's where things will start. And then how that 171 00:10:11,326 --> 00:10:16,636 Speaker 1: percolates or spreads out and becomes more mainstream across all categories, 172 00:10:16,756 --> 00:10:19,456 Speaker 1: that's the subjective piece. And that's what I was kind 173 00:10:19,496 --> 00:10:21,296 Speaker 1: of saying. It's hard to say when that will happen. 174 00:10:22,116 --> 00:10:25,396 Speaker 3: You mentioned books. I mean, famously, Amazon started out selling books. 175 00:10:25,996 --> 00:10:26,296 Speaker 1: Yeah. 176 00:10:27,266 --> 00:10:29,006 Speaker 3: Do you have a bet for what's going to be 177 00:10:29,046 --> 00:10:30,906 Speaker 3: the books of agentic? What's going to be the first 178 00:10:31,006 --> 00:10:36,186 Speaker 3: category where people are comfortable having an agent go make 179 00:10:36,206 --> 00:10:37,336 Speaker 3: the purchase decision for them? 180 00:10:38,216 --> 00:10:42,716 Speaker 1: Yeah, I actually feel like groceries are a really good 181 00:10:42,756 --> 00:10:45,516 Speaker 1: candidate for me. You know, and this is something that 182 00:10:45,696 --> 00:10:48,436 Speaker 1: I actually ask my students on a regular basis. I've 183 00:10:48,456 --> 00:10:51,756 Speaker 1: been doing this for about two, three years where anytime 184 00:10:51,826 --> 00:10:56,996 Speaker 1: I'm teaching my MBA, undergrad, or executive students, every session 185 00:10:57,026 --> 00:11:00,636 Speaker 1: I have on agentic shopping, AI shopping, I ask them, 186 00:11:01,616 --> 00:11:06,476 Speaker 1: would you use an AI agent to shop for groceries? 187 00:11:07,376 --> 00:11:11,776 Speaker 1: Just your usual grocery list. And perhaps you could argue 188 00:11:11,856 --> 00:11:14,756 Speaker 1: that if I'm using Instacart or if I'm using even 189 00:11:15,496 --> 00:11:18,316 Speaker 1: the Whole Foods or the Amazon app to order my 190 00:11:18,416 --> 00:11:20,736 Speaker 1: groceries where the list is there and I'm saying go 191 00:11:20,756 --> 00:11:24,276 Speaker 1: buy it, at some level, I'm kind of saying, hey, 192 00:11:25,716 --> 00:11:28,876 Speaker 1: here's my usual list. You buy it. You make the 193 00:11:28,916 --> 00:11:33,936 Speaker 1: decisions on substitutions and do it on my behalf. So 194 00:11:33,956 --> 00:11:37,996 Speaker 1: I've already outsourced that. That's not a big deal. Tomorrow, 195 00:11:38,056 --> 00:11:40,936 Speaker 1: if someone were to say, here's an AI agent that 196 00:11:40,976 --> 00:11:43,876 Speaker 1: can do end-to-end, just upload your grocery list once, and 197 00:11:43,916 --> 00:11:45,916 Speaker 1: then you don't have to ever ask again, that's one 198 00:11:45,966 --> 00:11:49,426 Speaker 1: I'd be willing to do. The Instacart, the human making 199 00:11:49,466 --> 00:11:54,526 Speaker 1: that decision, for me, is not that different than an 200 00:11:54,806 --> 00:11:59,466 Speaker 1: AI agent saying, hey, I'll substitute this particular brand of 201 00:11:59,506 --> 00:12:01,666 Speaker 1: milk with that particular brand of milk and so on. 202 00:12:02,196 --> 00:12:05,096 Speaker 1: I know what this customer wants. And I'll know in 203 00:12:05,696 --> 00:12:09,456 Speaker 1: two orders that this is working for me or not. 204 00:12:10,376 --> 00:12:12,356 Speaker 1: Am I willing to try two, three times? Yeah. 205 00:12:12,556 --> 00:12:15,716 Speaker 3: And it's relatively low stakes, right? It's not like buying 206 00:12:15,776 --> 00:12:17,676 Speaker 3: a car. If you do it and it kind of sucks, 207 00:12:17,696 --> 00:12:20,256 Speaker 3: you're like, well, I didn't get the right milk this week. 208 00:12:20,596 --> 00:12:21,996 Speaker 3: Not that big of a deal. 209 00:12:22,176 --> 00:12:25,826 Speaker 1: Exactly. So I'll happily try that a few times. So 210 00:12:25,846 --> 00:12:28,206 Speaker 1: I feel that's a good one. I also think the 211 00:12:28,266 --> 00:12:32,186 Speaker 1: other one is where individuals are doing a lot of 212 00:12:32,506 --> 00:12:35,556 Speaker 1: research and the research is a big part of it. 213 00:12:35,806 --> 00:12:39,236 Speaker 1: Insurance is a good example. If I were buying car insurance... Mm-hmm. 214 00:12:39,826 --> 00:12:42,666 Speaker 1: I would be happy to have Muse or somebody else 215 00:12:42,746 --> 00:12:46,076 Speaker 1: go to all the different websites, fill in my information, 216 00:12:47,026 --> 00:12:49,416 Speaker 1: get me a bunch of quotes, and then pick one. 217 00:12:49,796 --> 00:12:54,996 Speaker 1: Maybe with car insurance, I might still want the final decision, 218 00:12:55,056 --> 00:12:58,536 Speaker 1: like an approval gate. To me, that's still agentic if 219 00:12:59,076 --> 00:12:59,936 Speaker 1: it's doing everything. 220 00:12:59,996 --> 00:13:02,576 Speaker 3: I mean, that's a big purchase, and it's a lot 221 00:13:02,616 --> 00:13:04,796 Speaker 3: of money. And if it gets it a little bit wrong, 222 00:13:05,916 --> 00:13:08,876 Speaker 3: you're paying... you know, or buys the wrong policy. Like 223 00:13:09,516 --> 00:13:11,676 Speaker 3: that one feels way scarier to me than milk. 224 00:13:12,216 --> 00:13:13,936 Speaker 1: Yeah, but give me the chance to approve it. And 225 00:13:13,956 --> 00:13:15,816 Speaker 1: that's all I need. You give me all the data. 226 00:13:15,876 --> 00:13:20,236 Speaker 1: You've done all the work. In effect, from the brand's perspective, 227 00:13:20,356 --> 00:13:22,436 Speaker 1: AI is the customer. It's doing all of the work. 228 00:13:22,576 --> 00:13:25,016 Speaker 1: And I'm just kind of saying, okay, everything looks good. 229 00:13:25,096 --> 00:13:28,966 Speaker 1: Let's proceed. That's the one right I want with insurance. 230 00:13:28,986 --> 00:13:31,096 Speaker 1: But with groceries, I don't even need to approve it. 231 00:13:31,126 --> 00:13:33,446 Speaker 1: I'd say just do it. If it breaks two out 232 00:13:33,466 --> 00:13:35,186 Speaker 1: of three times or three out of three times, I'm 233 00:13:35,226 --> 00:13:37,536 Speaker 1: not repeating it. But if it works all three times, 234 00:13:37,596 --> 00:13:38,296 Speaker 1: I'll do it forever. 235 00:13:38,956 --> 00:13:43,456 Speaker 3: So let's talk about what happens when something goes wrong. 236 00:13:43,896 --> 00:13:45,836 Speaker 3: If the agent does something I don't want it to do, 237 00:13:45,876 --> 00:13:48,646 Speaker 3: it makes a mistake. If it goes off the rails, 238 00:13:49,566 --> 00:13:53,666 Speaker 3: we know that businesses are split over who should be 239 00:13:53,726 --> 00:13:56,716 Speaker 3: responsible in this kind of case when there's a dispute 240 00:13:56,746 --> 00:14:01,206 Speaker 3: between the AI platform and the customer. And it's clear 241 00:14:01,346 --> 00:14:04,866 Speaker 3: that most businesses think that the merchant who fulfilled the 242 00:14:04,986 --> 00:14:08,846 Speaker 3: order should not be the one who's financially responsible when 243 00:14:08,886 --> 00:14:11,766 Speaker 3: things go wrong. And so the question is, who should 244 00:14:11,786 --> 00:14:14,606 Speaker 3: be on the hook if the AI, say, makes an 245 00:14:14,786 --> 00:14:15,726 Speaker 3: unauthorized purchase? 246 00:14:16,346 --> 00:14:20,206 Speaker 1: That's a great question. I think anyone who's pushing agents 247 00:14:20,826 --> 00:14:25,346 Speaker 1: to make them mainstream for shopping, any company that's doing 248 00:14:25,386 --> 00:14:28,386 Speaker 1: that will have to assure customers, we are on the hook, 249 00:14:28,446 --> 00:14:32,426 Speaker 1: we'll take care of it. So that's when it goes mainstream. 250 00:14:32,466 --> 00:14:34,606 Speaker 1: I think that's when you get past the trust hurdle. 251 00:14:35,366 --> 00:14:39,886 Speaker 1: So if Muse wants to do that for me, or Instinct, 252 00:14:39,986 --> 00:14:42,826 Speaker 1: another app or startup that wants to do this for me, 253 00:14:43,266 --> 00:14:46,166 Speaker 1: if they want me to go full agentic, they've got 254 00:14:46,186 --> 00:14:50,746 Speaker 1: to take on the liability risk and say that if 255 00:14:50,846 --> 00:14:54,906 Speaker 1: our agent does something that deviates from this intent that 256 00:14:54,936 --> 00:14:56,836 Speaker 1: you had, then we will reimburse you. 257 00:14:57,766 --> 00:15:01,106 Speaker 3: You mentioned Muse. I mean, that's the meta... agent that 258 00:15:01,166 --> 00:15:05,846 Speaker 3: is new and buzzy and instinct, which is sort of 259 00:15:05,866 --> 00:15:09,866 Speaker 3: a startup, right, that wants to do something similar. Invite only, 260 00:15:09,946 --> 00:15:11,596 Speaker 3: I believe. I haven't been invited yet. 261 00:15:12,356 --> 00:15:13,076 Speaker 1: I can invite you. 262 00:15:13,176 --> 00:15:13,536 Speaker 3: Okay. 263 00:15:13,596 --> 00:15:14,016 Speaker 1: I'll do that. 264 00:15:14,036 --> 00:15:16,376 Speaker 3: I mean, I'm curious, what do you think it's going 265 00:15:16,396 --> 00:15:18,956 Speaker 3: to take? You know, there is this leap that a 266 00:15:19,036 --> 00:15:22,796 Speaker 3: lot of consumers haven't yet made. That leap is letting 267 00:15:22,866 --> 00:15:26,706 Speaker 3: an AI agent buy something for them. And I'm curious, 268 00:15:27,466 --> 00:15:29,006 Speaker 3: what is it going to take for that to happen? 269 00:15:29,226 --> 00:15:32,546 Speaker 3: And And do you think agents like Muse are going 270 00:15:32,566 --> 00:15:34,866 Speaker 3: to be part of the key to this transition? 271 00:15:36,086 --> 00:15:39,306 Speaker 1: Yeah, I think agents like Muse are key to the transition. 272 00:15:39,386 --> 00:15:42,026 Speaker 1: So what Muse and other agents can do is you 273 00:15:42,066 --> 00:15:45,306 Speaker 1: can articulate your preference, just like you do in a 274 00:15:45,346 --> 00:15:48,846 Speaker 1: prompt on an LLM. They can go to different websites, 275 00:15:48,926 --> 00:15:53,226 Speaker 1: gather the information, reason through that, fill forms if need be, 276 00:15:54,206 --> 00:15:57,866 Speaker 1: make a decision, present it to us. So all of that, 277 00:15:57,966 --> 00:16:00,846 Speaker 1: those systems do. And like I said, I use it 278 00:16:01,026 --> 00:16:04,006 Speaker 1: up to that point. They can also go and transact 279 00:16:04,046 --> 00:16:06,356 Speaker 1: and that's where they need to get past the trust hurdle. 280 00:16:06,436 --> 00:16:11,416 Speaker 1: And I think for that to happen, first, they need 281 00:16:11,456 --> 00:16:14,536 Speaker 1: to insure me about things going wrong, as you were 282 00:16:14,576 --> 00:16:17,516 Speaker 1: just alluding to earlier. So they have to tell me 283 00:16:17,576 --> 00:16:21,136 Speaker 1: that there is some protection. Just like, by the way, 284 00:16:21,196 --> 00:16:23,676 Speaker 1: a credit card company, when it's giving us a credit card, 285 00:16:23,696 --> 00:16:28,206 Speaker 1: they're saying if there's some fraudulent transaction, You're not liable. 286 00:16:28,266 --> 00:16:30,586 Speaker 1: You can appeal and you can say, this was not me. 287 00:16:31,346 --> 00:16:35,846 Speaker 1: Something's gone wrong. And they will make sure that no 288 00:16:35,866 --> 00:16:39,286 Speaker 1: harm is done. And they're providing that guarantee. That's what 289 00:16:39,346 --> 00:16:41,986 Speaker 1: needs to happen in this world. So someone needs to 290 00:16:42,026 --> 00:16:46,556 Speaker 1: be the guarantor or the trust builder here. So that's 291 00:16:46,656 --> 00:16:49,356 Speaker 1: the piece that is needed beyond just the technology needs 292 00:16:49,396 --> 00:16:49,716 Speaker 1: to work. 293 00:16:50,796 --> 00:16:53,776 Speaker 3: And it seems like that will happen, right? It seems 294 00:16:53,816 --> 00:16:59,096 Speaker 3: like the nature of the history of finance is... lowering 295 00:16:59,216 --> 00:17:03,356 Speaker 3: friction is the way things go. There's less and less 296 00:17:03,436 --> 00:17:06,086 Speaker 3: friction to buy things. A credit card is lower friction 297 00:17:06,106 --> 00:17:08,606 Speaker 3: than cash. Shopping online is lower friction than going to 298 00:17:08,646 --> 00:17:12,086 Speaker 3: a store. And it seems like ultimately, agentic AI will 299 00:17:12,126 --> 00:17:15,566 Speaker 3: be lower friction still, and someone will figure out how 300 00:17:15,606 --> 00:17:19,126 Speaker 3: to solve the trust problem, and people will use agents 301 00:17:19,186 --> 00:17:21,346 Speaker 3: to buy things on their behalf as often as we 302 00:17:21,396 --> 00:17:22,676 Speaker 3: buy stuff online now. 303 00:17:23,256 --> 00:17:25,536 Speaker 1: Yeah. I mean, look, I'm not a betting man, but 304 00:17:25,576 --> 00:17:28,666 Speaker 1: if you were to say, hey, What's your prediction here? 305 00:17:29,186 --> 00:17:32,306 Speaker 1: Pick a side. For me, it would be, you know, 306 00:17:32,346 --> 00:17:36,326 Speaker 1: there's no doubt the history of shopping shows that we're 307 00:17:37,026 --> 00:17:43,436 Speaker 1: always willing to adopt technologies that add convenience while guaranteeing trust. Now, 308 00:17:43,876 --> 00:17:45,736 Speaker 1: is it going to happen for every kind of shopping? 309 00:17:46,056 --> 00:17:49,776 Speaker 1: Maybe not. Is it going to happen for every person 310 00:17:49,856 --> 00:17:53,136 Speaker 1: on earth? Maybe not. But is it going to be mainstreamed? 311 00:17:53,616 --> 00:17:56,676 Speaker 1: For large pockets of shopping, my bet is that it 312 00:17:56,716 --> 00:17:57,696 Speaker 1: will happen. 313 00:18:01,196 --> 00:18:14,906 Speaker 3: We'll be back in just a minute. So we've been 314 00:18:14,946 --> 00:18:18,906 Speaker 3: talking essentially about the consumer point of view, about what 315 00:18:18,946 --> 00:18:22,186 Speaker 3: it's going to take for people to let agents buy 316 00:18:22,246 --> 00:18:25,446 Speaker 3: stuff on their behalf. I want to talk now about 317 00:18:25,546 --> 00:18:28,106 Speaker 3: the merchant point of view, about the retailer point of view. 318 00:18:28,726 --> 00:18:30,986 Speaker 3: This is a thing you have written about. You wrote 319 00:18:31,006 --> 00:18:35,086 Speaker 3: an article about it recently in the Harvard Business Review. And, 320 00:18:35,946 --> 00:18:40,666 Speaker 3: you know, it does seem like most business owners largely 321 00:18:40,806 --> 00:18:45,066 Speaker 3: trust AI platforms to represent their brand to customers. But 322 00:18:45,136 --> 00:18:47,456 Speaker 3: it is clear that there is, you know, to some extent, 323 00:18:47,566 --> 00:18:52,576 Speaker 3: a trust issue here, that the AI won't accurately represent 324 00:18:52,616 --> 00:18:55,036 Speaker 3: the brand. What are you seeing here? What are you 325 00:18:55,076 --> 00:18:56,876 Speaker 3: hearing from the merchants you talk to? 326 00:18:57,356 --> 00:19:01,696 Speaker 1: Yeah, I think merchants have to worry about this too. 327 00:19:01,796 --> 00:19:05,526 Speaker 1: And there's many layers of the merchant concerns. One is, 328 00:19:05,726 --> 00:19:09,986 Speaker 1: are the agents representing the brands correctly? And so I'll 329 00:19:10,006 --> 00:19:13,526 Speaker 1: give you an example. I was talking to this seller 330 00:19:14,246 --> 00:19:19,246 Speaker 1: of vintage clothes in London. And so this person owns 331 00:19:19,326 --> 00:19:22,626 Speaker 1: this brand. They sell in stores. They also sell online. 332 00:19:23,676 --> 00:19:27,676 Speaker 1: And they sell vintage women's clothing. He actually bought the 333 00:19:27,716 --> 00:19:31,166 Speaker 1: brand off of the previous owner. And he was kind 334 00:19:31,206 --> 00:19:35,926 Speaker 1: of saying, it's shocking that we are the number one 335 00:19:36,246 --> 00:19:41,846 Speaker 1: seller of vintage clothes in the UK. And you go 336 00:19:41,886 --> 00:19:45,346 Speaker 1: to ChatGPT, you go to Gemini, and you ask this question. 337 00:19:45,406 --> 00:19:49,096 Speaker 1: I'm in London. Where can I buy vintage clothes with 338 00:19:49,216 --> 00:19:52,566 Speaker 1: 70s look, 60s look, and things like that? or Vintage 339 00:19:52,666 --> 00:19:55,836 Speaker 1: Inspired rather is their brand, and his brand would never 340 00:19:55,876 --> 00:19:59,156 Speaker 1: show up. And he was like, what's going on? And 341 00:19:59,216 --> 00:20:02,396 Speaker 1: so I actually went through this a little bit. We 342 00:20:02,476 --> 00:20:05,616 Speaker 1: found what was happening was that there was a post 343 00:20:05,696 --> 00:20:09,166 Speaker 1: on Reddit where someone had mentioned that this brand is 344 00:20:09,186 --> 00:20:13,546 Speaker 1: shutting down. And the history there is that the previous 345 00:20:14,266 --> 00:20:19,056 Speaker 1: owner had put it up for sale And probably the 346 00:20:19,096 --> 00:20:21,756 Speaker 1: word got around that the previous owner was struggling to 347 00:20:21,796 --> 00:20:24,526 Speaker 1: make it work. And then, of course, new buyers took over. 348 00:20:24,566 --> 00:20:27,906 Speaker 1: But there was this post online. So ChatGPT had probably 349 00:20:27,966 --> 00:20:30,746 Speaker 1: decided that it shut down or something like that and 350 00:20:30,786 --> 00:20:33,846 Speaker 1: was not showing the brand. And so that's an example 351 00:20:34,286 --> 00:20:39,906 Speaker 1: of the brand being misrepresented by an LLM. 352 00:20:40,966 --> 00:20:43,946 Speaker 3: What do you do about that? Like, this is a problem, 353 00:20:44,026 --> 00:20:46,706 Speaker 3: and presumably it's not in the LLM's interest, right? The 354 00:20:46,746 --> 00:20:49,006 Speaker 3: maker of the model doesn't want it to be wrong. 355 00:20:49,436 --> 00:20:52,196 Speaker 3: The merchant doesn't want it to be wrong. Is this 356 00:20:52,256 --> 00:20:53,356 Speaker 3: a solvable problem? 357 00:20:54,476 --> 00:20:56,756 Speaker 1: It's a tough one to solve because the maker of 358 00:20:56,796 --> 00:21:00,116 Speaker 1: the LLM is not going to say, every time you're 359 00:21:00,156 --> 00:21:02,426 Speaker 1: not happy with an answer, appeal to us, and then 360 00:21:02,466 --> 00:21:04,846 Speaker 1: we'll have a human look through it and correct it. 361 00:21:04,886 --> 00:21:09,046 Speaker 1: That doesn't work for them. And so while it is 362 00:21:09,106 --> 00:21:13,666 Speaker 1: in their interest, there's just too much information on the web, 363 00:21:14,606 --> 00:21:18,706 Speaker 1: too many vested interests. And then how do you also 364 00:21:18,766 --> 00:21:22,746 Speaker 1: address people who might complain about subjective things and say that, 365 00:21:22,806 --> 00:21:26,846 Speaker 1: how come I'm not listed as the top three vintage brands? 366 00:21:26,906 --> 00:21:29,426 Speaker 1: And in fact, our survey shows that we're among the 367 00:21:29,466 --> 00:21:33,266 Speaker 1: top three. And so they're not going to provide resources 368 00:21:33,326 --> 00:21:36,126 Speaker 1: for that. So unfortunately, what it ends up becoming is 369 00:21:37,136 --> 00:21:39,616 Speaker 1: This is what I tell a lot of brands, that 370 00:21:39,676 --> 00:21:44,186 Speaker 1: this world is a lot like your customers are sometimes misinformed. 371 00:21:45,086 --> 00:21:48,146 Speaker 1: Your customers also need to be persuaded. And just like 372 00:21:48,226 --> 00:21:51,926 Speaker 1: you've spent decades thinking about, how do I ensure I 373 00:21:52,086 --> 00:21:55,566 Speaker 1: inform my customers about my brand? How do I make 374 00:21:55,606 --> 00:21:59,386 Speaker 1: sure I persuade my customers about your brand? You have 375 00:21:59,426 --> 00:22:02,106 Speaker 1: to do the same for AI agents now. And you 376 00:22:02,146 --> 00:22:04,886 Speaker 1: have to start thinking about the AI agents as another customer, 377 00:22:05,416 --> 00:22:08,186 Speaker 1: And the same questions you had to have for human customers, 378 00:22:08,276 --> 00:22:12,626 Speaker 1: now you need to answer those for AI agent customers. 379 00:22:12,706 --> 00:22:16,866 Speaker 1: And it's a new problem for marketers and brands and 380 00:22:16,886 --> 00:22:18,066 Speaker 1: so on. So, right. 381 00:22:18,086 --> 00:22:20,846 Speaker 3: So, I mean, the kind of central theme of the 382 00:22:20,886 --> 00:22:26,086 Speaker 3: merchant point of view section of your article is companies, merchants, 383 00:22:26,196 --> 00:22:29,316 Speaker 3: marketers need to think about the agent as a customer. 384 00:22:29,796 --> 00:22:32,436 Speaker 3: And sort of you dig into that and you lay 385 00:22:32,536 --> 00:22:37,416 Speaker 3: out these three common misconceptions that you have heard, you know, 386 00:22:37,496 --> 00:22:40,026 Speaker 3: in your work with retailers and marketers. And so I 387 00:22:40,056 --> 00:22:42,546 Speaker 3: actually want to walk through those and have you kind 388 00:22:42,586 --> 00:22:44,726 Speaker 3: of talk us through them. So I'm going to read 389 00:22:44,766 --> 00:22:45,926 Speaker 3: them to you and then I'm going to ask you 390 00:22:45,946 --> 00:22:49,626 Speaker 3: what they mean. So the first one, misconception number one, 391 00:22:49,986 --> 00:22:53,306 Speaker 3: is there will always be a human at the other end. 392 00:22:53,986 --> 00:22:56,246 Speaker 3: Tell me about that. Like, what are you hearing? Why 393 00:22:56,626 --> 00:22:57,636 Speaker 3: is that a misconception? 394 00:22:58,756 --> 00:23:01,986 Speaker 1: Yeah, so I think a common one that a lot 395 00:23:02,006 --> 00:23:06,546 Speaker 1: of marketers and brands think is, okay, AI is probably 396 00:23:06,586 --> 00:23:10,986 Speaker 1: another channel like search, where in search engines, we have 397 00:23:11,026 --> 00:23:14,416 Speaker 1: to think through how we show up in Google's results pages, 398 00:23:14,506 --> 00:23:17,156 Speaker 1: but there's always a human reading it and making the decision. 399 00:23:18,026 --> 00:23:19,756 Speaker 1: And so if you think about it that way, then 400 00:23:19,776 --> 00:23:21,956 Speaker 1: you would say, hey, maybe the human is doing the 401 00:23:21,996 --> 00:23:25,636 Speaker 1: research through chat GPT, but then the human's making the 402 00:23:25,676 --> 00:23:30,506 Speaker 1: final decision. But I would argue that the direction in 403 00:23:30,546 --> 00:23:33,926 Speaker 1: which we are headed is where people will say there 404 00:23:33,966 --> 00:23:38,246 Speaker 1: are certain decisions where, you know, I don't need the 405 00:23:38,286 --> 00:23:40,396 Speaker 1: headache of going through all of this stuff. It's not 406 00:23:40,436 --> 00:23:43,736 Speaker 1: a lot of joy for me to sit and comb 407 00:23:43,776 --> 00:23:46,296 Speaker 1: through the information and make sense of it. And I'd 408 00:23:46,356 --> 00:23:47,266 Speaker 1: rather outsource it. 409 00:23:47,896 --> 00:23:50,176 Speaker 3: Outsource it to an AI agent. 410 00:23:50,236 --> 00:23:54,056 Speaker 1: To an AI agent. Yeah. Right. So groceries, we already 411 00:23:54,096 --> 00:23:57,326 Speaker 1: talked about as one example of that. Yeah. There could 412 00:23:57,346 --> 00:23:59,636 Speaker 1: be many others. I gave the example in the article 413 00:23:59,676 --> 00:24:04,256 Speaker 1: about like tickets to a music concert. So I know 414 00:24:04,416 --> 00:24:08,416 Speaker 1: I'm interested in, let's say, Coldplay concert. I want that 415 00:24:08,476 --> 00:24:12,186 Speaker 1: ticket or in my daughter's case, you know, Taylor Swift, 416 00:24:12,426 --> 00:24:14,906 Speaker 1: and we want that ticket. So I just tell an 417 00:24:14,966 --> 00:24:17,226 Speaker 1: AI agent, hey, listen, just keep searching on the web. 418 00:24:17,286 --> 00:24:19,816 Speaker 1: If it's under a certain price, just go buy it. 419 00:24:19,946 --> 00:24:23,726 Speaker 1: If it's anywhere in my vicinity, go buy it. Because 420 00:24:23,806 --> 00:24:26,236 Speaker 1: if you're going to find it and then flag it 421 00:24:26,276 --> 00:24:28,156 Speaker 1: for me, but I'm busy in a meeting and I 422 00:24:28,196 --> 00:24:31,216 Speaker 1: don't respond fast enough. And then the tickets are gone. 423 00:24:31,336 --> 00:24:34,056 Speaker 1: I'd just rather just go search. If you get one, 424 00:24:34,116 --> 00:24:36,756 Speaker 1: then go just buy it. So those kinds of things, 425 00:24:36,816 --> 00:24:39,066 Speaker 1: I kind of feel like, you know, they're going to 426 00:24:39,106 --> 00:24:42,386 Speaker 1: train people more and more to use agents. So this 427 00:24:42,486 --> 00:24:46,426 Speaker 1: idea that a human being is making the decision is 428 00:24:46,626 --> 00:24:51,286 Speaker 1: not quite there in those instances. So You know, it's 429 00:24:51,326 --> 00:24:54,146 Speaker 1: the agent that is getting informed, the agent that is 430 00:24:54,166 --> 00:24:56,666 Speaker 1: getting persuaded to make choices for the most part, the 431 00:24:56,726 --> 00:24:57,286 Speaker 1: AI agent. 432 00:24:58,846 --> 00:24:59,806 Speaker 3: Did you see Coldplay? 433 00:25:00,906 --> 00:25:05,026 Speaker 1: I saw both of those concerts, actually. 434 00:25:05,046 --> 00:25:07,916 Speaker 3: Okay, back to your article, back to that list of misconceptions. 435 00:25:08,696 --> 00:25:12,496 Speaker 3: We are at merchant misconception number two, and that misconception 436 00:25:12,616 --> 00:25:16,416 Speaker 3: is we just need AI crawlers to read our content. 437 00:25:17,216 --> 00:25:20,056 Speaker 3: And in fact, most businesses say that they have started 438 00:25:20,216 --> 00:25:23,906 Speaker 3: structuring their catalogs for AI, although for the most part, 439 00:25:23,946 --> 00:25:26,266 Speaker 3: they're not all the way there yet. They're not fully ready. 440 00:25:26,326 --> 00:25:28,966 Speaker 3: But why is this a misconception? Why is it a 441 00:25:29,066 --> 00:25:32,966 Speaker 3: misconception that merchants think we just need AI crawlers to 442 00:25:33,066 --> 00:25:34,806 Speaker 3: read our content? What's going on here? 443 00:25:35,066 --> 00:25:39,286 Speaker 1: Well, I would say it's necessary but not sufficient to 444 00:25:39,366 --> 00:25:42,386 Speaker 1: succeed in a world where AI agents might be shopping 445 00:25:43,026 --> 00:25:46,716 Speaker 1: on behalf of customers. It's necessary because, yes, you want 446 00:25:46,736 --> 00:25:49,416 Speaker 1: to make sure AI crawlers can read your content and 447 00:25:49,456 --> 00:25:53,176 Speaker 1: understand how you show up. But that's not enough because 448 00:25:53,236 --> 00:25:57,076 Speaker 1: their answers to customer questions or the way they reason 449 00:25:57,146 --> 00:26:01,926 Speaker 1: through all of the information is not just informed by 450 00:26:02,566 --> 00:26:04,866 Speaker 1: reading through your page and making sense of it. They're 451 00:26:05,006 --> 00:26:07,646 Speaker 1: also reading a lot of other stuff. For example, they're 452 00:26:07,686 --> 00:26:11,786 Speaker 1: reading third-party reviews. Some may be biased. Some may be objective. 453 00:26:12,306 --> 00:26:15,596 Speaker 1: They're reading through those, making sense of it. they're increasingly 454 00:26:15,636 --> 00:26:20,036 Speaker 1: starting to process videos. So I increasingly see that Google 455 00:26:20,096 --> 00:26:25,436 Speaker 1: AI Answers and Gemini Answers are shaped by YouTube videos increasingly. 456 00:26:26,356 --> 00:26:28,936 Speaker 1: They're shaped by Reddit discussions. So there's a lot of 457 00:26:28,996 --> 00:26:31,816 Speaker 1: other things that goes into them. So the fact that 458 00:26:31,856 --> 00:26:37,376 Speaker 1: they read your content does not mean that they are 459 00:26:37,436 --> 00:26:40,926 Speaker 1: going to make decisions that are favorable to the brand. 460 00:26:40,946 --> 00:26:43,996 Speaker 1: So just like every marketers had to think through how 461 00:26:44,036 --> 00:26:46,706 Speaker 1: do you persuade humans? They have to think about how 462 00:26:47,226 --> 00:26:49,746 Speaker 1: do you persuade AI? But with humans, we understand a 463 00:26:49,786 --> 00:26:54,906 Speaker 1: few things. We understand $ 9. 99 is not treated the same 464 00:26:54,966 --> 00:26:57,726 Speaker 1: as $ 10. It doesn't make any rational sense, but it 465 00:26:57,806 --> 00:27:02,066 Speaker 1: is true that people sort of react to these prices differently. 466 00:27:02,106 --> 00:27:07,036 Speaker 1: We also know that social proof matters and saying that 467 00:27:07,076 --> 00:27:09,516 Speaker 1: your friends have bought this matters or you know that 468 00:27:09,676 --> 00:27:13,436 Speaker 1: scarcity matters for people. Only three items in stock can 469 00:27:13,496 --> 00:27:14,846 Speaker 1: influence somebody to buy it. 470 00:27:14,916 --> 00:27:15,716 Speaker 3: Buy now. 471 00:27:16,076 --> 00:27:16,896 Speaker 1: Buy now, right? 472 00:27:16,956 --> 00:27:17,456 Speaker 3: Exactly. 473 00:27:17,496 --> 00:27:19,316 Speaker 1: So we know those things matter for humans. But if 474 00:27:19,376 --> 00:27:22,576 Speaker 1: I were to ask any marketer, what are these biases 475 00:27:22,666 --> 00:27:25,906 Speaker 1: that AI agents have? Or how do you persuade an 476 00:27:25,966 --> 00:27:28,746 Speaker 1: AI agent? The answer would be a blank because none 477 00:27:28,786 --> 00:27:32,406 Speaker 1: of us truly understand it. So that's why it's important 478 00:27:32,446 --> 00:27:36,006 Speaker 1: to understand that misconception and really start thinking hard about 479 00:27:36,046 --> 00:27:36,826 Speaker 1: how we correct that. 480 00:27:37,686 --> 00:27:41,406 Speaker 3: Does anybody know anything about how to persuade an AI 481 00:27:41,486 --> 00:27:43,966 Speaker 3: agent or a language model making recommendations? 482 00:27:45,346 --> 00:27:50,106 Speaker 1: We're starting to understand some basics. And so, for example, 483 00:27:50,266 --> 00:27:54,166 Speaker 1: I just mentioned that with Google, I have enough data 484 00:27:54,206 --> 00:27:58,806 Speaker 1: to suggest that YouTube videos matter. And so in your 485 00:27:58,846 --> 00:28:05,376 Speaker 1: marketing strategy, you should be thinking about influencers and you know, 486 00:28:05,416 --> 00:28:10,836 Speaker 1: the relevant, who are the relevant influencers? Which influencers are shaping, say, 487 00:28:11,096 --> 00:28:14,636 Speaker 1: Gemini or AI answers in my business vertical? And who 488 00:28:14,676 --> 00:28:20,126 Speaker 1: should I go talk to? Or understanding, you know, sometimes 489 00:28:20,226 --> 00:28:24,646 Speaker 1: even page layouts and how they matter. For example, and 490 00:28:24,686 --> 00:28:26,366 Speaker 1: this could change, by the way, what I'm about to 491 00:28:26,386 --> 00:28:28,666 Speaker 1: say could change in six months because new models can 492 00:28:28,706 --> 00:28:31,606 Speaker 1: come in and they may reason through information differently. But 493 00:28:31,626 --> 00:28:35,696 Speaker 1: I've already seen, for example, that presenting information in the 494 00:28:35,736 --> 00:28:41,016 Speaker 1: form of tables somehow has a greater impact than having 495 00:28:41,686 --> 00:28:46,266 Speaker 1: the information in paragraphs, especially when you're comparing things. So why, 496 00:28:46,366 --> 00:28:50,526 Speaker 1: who knows? I mean, but for some reason, if that 497 00:28:50,606 --> 00:28:54,546 Speaker 1: same comparison is presented as a table versus three paragraphs, 498 00:28:55,186 --> 00:28:59,256 Speaker 1: it somehow seems to persuade AI differently. So there's a 499 00:28:59,316 --> 00:29:00,856 Speaker 1: lot of small things and big things. 500 00:29:01,256 --> 00:29:05,006 Speaker 3: So at least for now, language models love tables. 501 00:29:05,946 --> 00:29:08,926 Speaker 1: At least for now, they love tables. Don't ask me why, 502 00:29:08,986 --> 00:29:10,466 Speaker 1: but that's what we find. 503 00:29:11,246 --> 00:29:13,726 Speaker 3: Okay, so we're going back to your list of common 504 00:29:13,766 --> 00:29:18,526 Speaker 3: misconceptions that marketers, retailers, merchants have about AI. The last one, 505 00:29:18,586 --> 00:29:23,876 Speaker 3: misconception three, is AI is just another channel. Why is 506 00:29:23,936 --> 00:29:25,156 Speaker 3: that a misconception? 507 00:29:27,296 --> 00:29:29,796 Speaker 1: Yeah. I mean, I think the most common way in 508 00:29:29,836 --> 00:29:33,286 Speaker 1: which brands perceive the impact of AI is to think 509 00:29:33,346 --> 00:29:36,346 Speaker 1: of it as another channel. Your customers can come to 510 00:29:36,406 --> 00:29:40,446 Speaker 1: you directly. They can come to you via social. They 511 00:29:40,466 --> 00:29:43,246 Speaker 1: can come to you via search. And now they can 512 00:29:43,286 --> 00:29:47,686 Speaker 1: come to you via AI search or via AI agents. 513 00:29:48,486 --> 00:29:52,166 Speaker 1: And that perspective just says that, okay, we've got a 514 00:29:52,206 --> 00:29:54,196 Speaker 1: bunch of channels and we need to show up in 515 00:29:54,246 --> 00:29:58,676 Speaker 1: this other channel. And what I worry about that is 516 00:29:58,816 --> 00:30:02,896 Speaker 1: that you are mistaking something that I view as a 517 00:30:03,016 --> 00:30:09,466 Speaker 1: system level shift in how buying is happening And you're 518 00:30:09,506 --> 00:30:13,806 Speaker 1: mistaking that for another distribution channel that emerged. The analogy 519 00:30:13,866 --> 00:30:17,706 Speaker 1: I would give is that when Netflix got started, all 520 00:30:17,746 --> 00:30:20,936 Speaker 1: the movie studios said, okay, here's another channel through which 521 00:30:21,036 --> 00:30:24,296 Speaker 1: I can distribute my content. I can get my movies 522 00:30:24,416 --> 00:30:27,616 Speaker 1: out via the movie theater. I can get it out 523 00:30:27,696 --> 00:30:31,816 Speaker 1: via pay-per-view on TV. I can also give it to 524 00:30:31,836 --> 00:30:34,166 Speaker 1: various TV channels or I can give it to Netflix 525 00:30:34,566 --> 00:30:38,156 Speaker 1: through streaming. And this is yet another channel And what 526 00:30:38,176 --> 00:30:41,816 Speaker 1: they realized later was that, whoa, it's not just another 527 00:30:41,896 --> 00:30:45,076 Speaker 1: channel because what's happening is it's changing my customer behavior 528 00:30:45,176 --> 00:30:48,916 Speaker 1: and training them to work in entirely different ways. And 529 00:30:48,976 --> 00:30:51,636 Speaker 1: it's changing who has the power and so on. And 530 00:30:51,696 --> 00:30:55,716 Speaker 1: ultimately Netflix took over that industry because Netflix said, hey, 531 00:30:55,736 --> 00:30:57,876 Speaker 1: maybe I don't need your content anymore. I can just 532 00:30:57,936 --> 00:31:01,556 Speaker 1: make my own content and deliver it and so on. 533 00:31:01,756 --> 00:31:04,076 Speaker 1: And so I think there are similar analogies here, which 534 00:31:04,166 --> 00:31:06,776 Speaker 1: is to say that If you just think about it 535 00:31:06,856 --> 00:31:09,736 Speaker 1: as another channel, you're missing two things. You're missing one, 536 00:31:10,376 --> 00:31:13,336 Speaker 1: the fact that your customer behavior is changing in big ways. 537 00:31:14,656 --> 00:31:16,956 Speaker 1: And as a result, you could lose market share because 538 00:31:17,016 --> 00:31:19,286 Speaker 1: if all you're thinking is it's another channel and I'm 539 00:31:19,326 --> 00:31:23,146 Speaker 1: on that channel, your competitor that is thinking smartly about 540 00:31:23,326 --> 00:31:26,846 Speaker 1: how to inform and persuade this new type of customer 541 00:31:27,066 --> 00:31:29,846 Speaker 1: AI agents will be ahead of you and will steal 542 00:31:29,896 --> 00:31:33,256 Speaker 1: market share from you. The second thing is you're also 543 00:31:33,356 --> 00:31:37,996 Speaker 1: missing how the power dynamics might change. If people are 544 00:31:38,036 --> 00:31:43,676 Speaker 1: doing more shopping through, I'll pick an LLM like say ChatGPD, 545 00:31:44,176 --> 00:31:48,076 Speaker 1: then does it shift the power balance and those LLMs 546 00:31:48,116 --> 00:31:51,176 Speaker 1: become more powerful? By the way, Amazon has acknowledged that 547 00:31:51,216 --> 00:31:54,256 Speaker 1: this could be a huge impact on them, which is 548 00:31:54,316 --> 00:31:58,656 Speaker 1: why Amazon is reluctant to provide its full product feed 549 00:31:58,736 --> 00:32:02,506 Speaker 1: to the ChatGPDs of the world. And instead, Amazon is 550 00:32:02,566 --> 00:32:07,986 Speaker 1: building its own AI agentic shopping experience, which is called Rufus, 551 00:32:08,686 --> 00:32:12,586 Speaker 1: and trying to persuade people, hey, have AI do your 552 00:32:12,626 --> 00:32:15,846 Speaker 1: shopping for you, but do it within our Amazon environment. 553 00:32:16,426 --> 00:32:19,886 Speaker 1: That's because Amazon recognizes that the power balance will shift 554 00:32:20,586 --> 00:32:22,986 Speaker 1: if the agent doing the shopping is not their agent, 555 00:32:23,026 --> 00:32:25,646 Speaker 1: but a third party, because then they're at the mercy 556 00:32:25,686 --> 00:32:28,766 Speaker 1: of the third party deciding whether it'll finally buy it 557 00:32:28,786 --> 00:32:32,006 Speaker 1: from Walmart or whoever else. And so they recognize that. 558 00:32:32,046 --> 00:32:35,326 Speaker 1: And that's an example of saying AI is not another channel. 559 00:32:36,186 --> 00:32:39,306 Speaker 3: Well, I mean, that last example is compelling, right? Amazon 560 00:32:39,366 --> 00:32:43,756 Speaker 3: is one of the few retailers with the scale and 561 00:32:43,836 --> 00:32:47,456 Speaker 3: the money and the technical expertise to be able to say, oh, no, 562 00:32:47,476 --> 00:32:50,936 Speaker 3: we're not going to let these outside AI agents shop 563 00:32:50,976 --> 00:32:54,656 Speaker 3: with us. We're going to build our own. Like, clearly, 564 00:32:54,696 --> 00:32:58,416 Speaker 3: businesses are worried that If they don't adopt agentic commerce, 565 00:32:58,596 --> 00:33:02,476 Speaker 3: then they risk losing customers or sales to competitors. So 566 00:33:03,096 --> 00:33:05,066 Speaker 3: if you're a small company, what should you do? 567 00:33:05,606 --> 00:33:08,166 Speaker 1: Yeah, I think you need a holistic strategy. So by 568 00:33:08,206 --> 00:33:10,366 Speaker 1: the way, when we talked about e-commerce, it's not like 569 00:33:10,406 --> 00:33:13,806 Speaker 1: stores don't exist today. They are there. And a good strategy, 570 00:33:13,866 --> 00:33:16,356 Speaker 1: like we talked about Warby Parker, figures out how do 571 00:33:16,396 --> 00:33:22,396 Speaker 1: we mix stores for humans who want that experience walking 572 00:33:22,456 --> 00:33:25,756 Speaker 1: into a store and trying on glasses with an online 573 00:33:25,776 --> 00:33:28,756 Speaker 1: strategy for humans who want the convenience of going to 574 00:33:28,776 --> 00:33:32,116 Speaker 1: a website and with four clicks getting their eyeglasses. And 575 00:33:32,196 --> 00:33:35,186 Speaker 1: so it may be a strategy where you say, how 576 00:33:35,226 --> 00:33:39,526 Speaker 1: do we have a strategy where we provide an experience 577 00:33:39,566 --> 00:33:43,046 Speaker 1: for humans who want to shop themselves? And also an 578 00:33:43,086 --> 00:33:46,896 Speaker 1: experience for agents who have already taken over a big 579 00:33:46,976 --> 00:33:49,976 Speaker 1: chunk of the shopping work on behalf of humans. And 580 00:33:50,036 --> 00:33:50,996 Speaker 1: so how do you support both? 581 00:33:51,076 --> 00:33:54,376 Speaker 3: Although now you're walking back to the It's Another Channel, right? 582 00:33:54,416 --> 00:33:57,916 Speaker 3: Like that is the It's Another Channel story. But the 583 00:33:57,996 --> 00:34:02,966 Speaker 3: Netflix analogy suggests that you might end up getting eaten 584 00:34:03,226 --> 00:34:03,686 Speaker 3: in the end. 585 00:34:04,826 --> 00:34:07,106 Speaker 1: Good pushback. But by the way, notice that I did 586 00:34:07,146 --> 00:34:11,066 Speaker 1: not say, how did you, how do you support humans 587 00:34:11,126 --> 00:34:13,526 Speaker 1: who want to shop on their own with humans who 588 00:34:13,546 --> 00:34:15,826 Speaker 1: want to shop through AI. What I said was, how 589 00:34:15,846 --> 00:34:19,146 Speaker 1: do you support humans who want to shop and AI 590 00:34:19,246 --> 00:34:22,956 Speaker 1: agents who are shopping? And so I'm really saying it's 591 00:34:23,026 --> 00:34:26,196 Speaker 1: another type of customer. It's another audience and you need 592 00:34:26,236 --> 00:34:29,436 Speaker 1: a real strategy for that audience. And so for that, 593 00:34:29,696 --> 00:34:32,596 Speaker 1: it might mean, I'll give you again an example. You 594 00:34:32,616 --> 00:34:36,726 Speaker 1: might say that Either the way I respond to it, 595 00:34:36,986 --> 00:34:40,726 Speaker 1: I have to make shopping so enjoyable and so luxurious 596 00:34:40,796 --> 00:34:43,876 Speaker 1: and so high-end that it's worth it for the human 597 00:34:43,896 --> 00:34:48,016 Speaker 1: to show up. And a very high-end luxury brand would 598 00:34:48,056 --> 00:34:51,256 Speaker 1: have to ask, how do I make it so amazing 599 00:34:51,296 --> 00:34:54,256 Speaker 1: that the person says, I'm not asking any AI agent 600 00:34:54,296 --> 00:34:56,736 Speaker 1: to do anything. I want to show up at the 601 00:34:56,776 --> 00:34:58,976 Speaker 1: car dealership and I want to sit in this car 602 00:34:59,036 --> 00:35:02,736 Speaker 1: and experience it. And that is what shopping is going 603 00:35:02,756 --> 00:35:06,766 Speaker 1: to be. So you either make that call or you've 604 00:35:06,806 --> 00:35:09,626 Speaker 1: got to make the other call, which is to say 605 00:35:09,826 --> 00:35:14,226 Speaker 1: our primary customers will be AI agents. And so what 606 00:35:14,266 --> 00:35:18,356 Speaker 1: I'm going to do now is have a team that 607 00:35:18,416 --> 00:35:21,976 Speaker 1: is solely focused on winning with AI agents. And that 608 00:35:22,056 --> 00:35:25,336 Speaker 1: is to ask, how do we inform these new customers? 609 00:35:25,396 --> 00:35:28,816 Speaker 1: How do we persuade these new customers? And unleash this 610 00:35:28,896 --> 00:35:32,436 Speaker 1: new team to go chase after that business And make 611 00:35:32,476 --> 00:35:34,696 Speaker 1: that work. I think you've got to first make that 612 00:35:34,736 --> 00:35:38,796 Speaker 1: strategic choice. And once you've made that, your playbook is 613 00:35:38,836 --> 00:35:42,226 Speaker 1: going to be extremely different for these two strategies. But 614 00:35:42,246 --> 00:35:45,386 Speaker 1: that second strategy is, again, I'm still saying it's not 615 00:35:45,446 --> 00:35:48,526 Speaker 1: another channel. It's here's your new customers. How do you 616 00:35:48,546 --> 00:35:53,646 Speaker 1: win those customers? That now means you hire new researchers 617 00:35:53,686 --> 00:35:56,356 Speaker 1: who are figuring out how are these AI agents reasoning? 618 00:35:57,336 --> 00:36:00,256 Speaker 1: I've actually just created a new startup called Bodium Labs, 619 00:36:00,296 --> 00:36:02,296 Speaker 1: and what we are doing is doing that on behalf 620 00:36:02,336 --> 00:36:06,756 Speaker 1: of brands. We are basically building systems that brings in 621 00:36:06,876 --> 00:36:10,936 Speaker 1: interpretability or explanations for how AI systems reason. So it's 622 00:36:11,016 --> 00:36:13,576 Speaker 1: no longer this channel. It's almost just like you have 623 00:36:14,296 --> 00:36:16,396 Speaker 1: a data science team that's thinking about how do you 624 00:36:16,456 --> 00:36:19,776 Speaker 1: inform and persuade humans. You have an AI research team 625 00:36:19,816 --> 00:36:21,966 Speaker 1: that's thinking about how do you inform and persuade AI. 626 00:36:26,316 --> 00:36:28,136 Speaker 3: We're going to take an ad break now and we 627 00:36:28,176 --> 00:36:40,126 Speaker 3: will be back in just a minute. Your HBR article 628 00:36:40,166 --> 00:36:44,206 Speaker 3: lands with this idea that AI is a new kind 629 00:36:44,246 --> 00:36:47,256 Speaker 3: of customer, right? That the, at this moment, we're in 630 00:36:47,296 --> 00:36:49,936 Speaker 3: this hybrid moment where people are using AI, but people 631 00:36:49,996 --> 00:36:55,026 Speaker 3: are still clearly the customer. Um, we will probably soon 632 00:36:55,066 --> 00:36:57,706 Speaker 3: be in a place where the AI is actually making decisions, 633 00:36:57,786 --> 00:37:01,566 Speaker 3: actually buying things. And so like, ultimately, what does that mean? 634 00:37:01,606 --> 00:37:03,006 Speaker 3: What does that world look like? How do you think 635 00:37:03,066 --> 00:37:03,346 Speaker 3: about it? 636 00:37:04,086 --> 00:37:07,866 Speaker 1: I'll share one thought I have about how it plays out. 637 00:37:09,146 --> 00:37:13,296 Speaker 1: By the way, in software settings and in B2B settings, 638 00:37:13,316 --> 00:37:15,656 Speaker 1: we already see it. You asked me for early examples 639 00:37:15,696 --> 00:37:19,436 Speaker 1: where this is already happening. Many people are building their 640 00:37:20,416 --> 00:37:27,886 Speaker 1: online storefronts through tools like Cloud Code. And they're not 641 00:37:28,206 --> 00:37:30,916 Speaker 1: hiring developers to go build it. They'll just say, hey, 642 00:37:30,956 --> 00:37:33,336 Speaker 1: build a storefront for me. I want to sell XYZ 643 00:37:33,996 --> 00:37:39,356 Speaker 1: and Cloud Code or Lovable or these other no-code tools 644 00:37:39,416 --> 00:37:41,956 Speaker 1: that will have AI write the code, no humans writing 645 00:37:41,976 --> 00:37:45,016 Speaker 1: the code, will build the storefront. When they're doing that, 646 00:37:45,056 --> 00:37:47,696 Speaker 1: they're making decisions. They're making, for example, how are you 647 00:37:47,716 --> 00:37:52,356 Speaker 1: going to accept payments? They, the agents. Yeah. 648 00:37:52,476 --> 00:37:55,586 Speaker 3: So the agents are making decisions about things like payments, 649 00:37:55,626 --> 00:37:58,346 Speaker 3: these fundamental business trust things. 650 00:37:58,386 --> 00:38:01,566 Speaker 1: Yeah. Yeah. And if you now take that to the 651 00:38:01,606 --> 00:38:06,296 Speaker 1: consumer world, I think increasingly we will have a shopping assistant. 652 00:38:06,476 --> 00:38:08,596 Speaker 1: I don't know who wins that. Is that meta with 653 00:38:08,656 --> 00:38:11,716 Speaker 1: Muse or is that going to be a next generation 654 00:38:11,776 --> 00:38:16,516 Speaker 1: thing that Apple launches with Siri and Maybe in the future, 655 00:38:17,176 --> 00:38:19,296 Speaker 1: or Google does something, but it might be on the 656 00:38:19,356 --> 00:38:22,816 Speaker 1: phone itself. It might come built into the phone. But 657 00:38:22,856 --> 00:38:25,966 Speaker 1: a system where you might say, hey, I want to 658 00:38:26,006 --> 00:38:33,526 Speaker 1: buy this kitchen timer for when I'm cooking in the kitchen, 659 00:38:33,866 --> 00:38:36,186 Speaker 1: or I want to buy a new phone, or I 660 00:38:36,226 --> 00:38:39,336 Speaker 1: want to buy an alarm clock because I don't want 661 00:38:39,356 --> 00:38:41,796 Speaker 1: to set it on my phone, whatever the reason. But 662 00:38:42,296 --> 00:38:45,646 Speaker 1: you might just speak that into the AI. And you 663 00:38:45,686 --> 00:38:49,946 Speaker 1: might either say, do all the work, make a decision, 664 00:38:50,066 --> 00:38:52,946 Speaker 1: confirm and get my approval before buying, or you might say, 665 00:38:53,526 --> 00:38:57,926 Speaker 1: proceed to buy it and you outsource that. And it's 666 00:38:57,966 --> 00:39:01,026 Speaker 1: going to be that seamless for a customer. But in 667 00:39:01,106 --> 00:39:04,676 Speaker 1: that world, we then have to think through what does 668 00:39:04,716 --> 00:39:07,976 Speaker 1: that mean for businesses? You know, how do you stay 669 00:39:07,996 --> 00:39:12,296 Speaker 1: relevant as a business when you're branding? What does brand 670 00:39:12,316 --> 00:39:14,356 Speaker 1: even mean now in a world where AI agents are 671 00:39:14,396 --> 00:39:18,496 Speaker 1: deciding that? What does persuasion mean when AI agents are 672 00:39:18,516 --> 00:39:22,616 Speaker 1: making the choices? What does advertising look like in that 673 00:39:22,676 --> 00:39:26,636 Speaker 1: kind of environment? How do you advertise to AI is unclear, 674 00:39:26,656 --> 00:39:28,436 Speaker 1: but all of those will have to evolve. And I 675 00:39:28,476 --> 00:39:31,046 Speaker 1: think there will be a whole AI economy coming up. 676 00:39:32,146 --> 00:39:34,206 Speaker 1: When AI is buying and selling, you might have a 677 00:39:34,586 --> 00:39:38,286 Speaker 1: whole new layer of the economy open up with services 678 00:39:38,406 --> 00:39:42,416 Speaker 1: targeting as either helping brands sell to AI or helping 679 00:39:43,136 --> 00:39:46,546 Speaker 1: AI make better decisions when buying on behalf of some people. 680 00:39:47,266 --> 00:39:50,006 Speaker 1: And so I think that's an interesting new era that's 681 00:39:50,266 --> 00:39:52,346 Speaker 1: potentially coming up in the next several years. 682 00:39:53,266 --> 00:39:54,906 Speaker 3: It was great to talk with you. Thank you so 683 00:39:54,926 --> 00:39:55,626 Speaker 3: much for your time. 684 00:39:56,226 --> 00:39:57,836 Speaker 1: Thank you for having me. It was my pleasure. 685 00:40:05,696 --> 00:40:09,546 Speaker 3: Karthik Hosanagar is a professor of operations and marketing at 686 00:40:09,566 --> 00:40:13,346 Speaker 3: the Wharton School at the University of Pennsylvania. Today's show 687 00:40:13,406 --> 00:40:17,176 Speaker 3: was produced by Gabriel Hunter Chang, engineered by Sarah Bruguier, 688 00:40:17,496 --> 00:40:20,976 Speaker 3: and our showrunner and editor is Jake Harper. I'm Jacob Goldstein.