1 00:00:00,299 --> 00:00:02,849 In this episode, I'm going to explain to you how not to 2 00:00:02,849 --> 00:00:04,619 use AI in retail businesses. 3 00:00:06,059 --> 00:00:08,070 Welcome to the Resilient Retail Game Plan. 4 00:00:08,349 --> 00:00:11,650 I'm Catherine Erdly, and this podcast is here to share retail industry 5 00:00:11,650 --> 00:00:15,780 trends, insights, and best practices for you, the retailer who's serious 6 00:00:15,780 --> 00:00:17,749 about building something that lasts. 7 00:00:21,576 --> 00:00:25,786 This episode is all about how you use generative AI in your business. 8 00:00:26,246 --> 00:00:27,596 Why do I make that distinction? 9 00:00:27,786 --> 00:00:31,006 Well, because there's a lot of chat about AI in retail at the moment, and 10 00:00:31,006 --> 00:00:35,346 a lot of that revolves around agentic AI, which I'm going to be covering 11 00:00:35,376 --> 00:00:37,096 in an episode coming up shortly. 12 00:00:37,516 --> 00:00:41,366 In today's episode, what I really wanted to talk about was you as a business owner 13 00:00:41,456 --> 00:00:43,816 using generative AI in your business. 14 00:00:44,156 --> 00:00:49,496 This is about ChatGPT style tools or Claude, and it's also about AI 15 00:00:49,496 --> 00:00:52,596 built into many of the tools that you've been using, like Shopify, 16 00:00:52,666 --> 00:00:54,626 email platforms, or accounting tools. 17 00:00:55,646 --> 00:01:00,216 And what I'm seeing is that almost everyone is experimenting and 18 00:01:00,346 --> 00:01:02,796 lots of people are wondering what else there is that they could 19 00:01:02,796 --> 00:01:04,746 be doing, and perhaps what... 20 00:01:04,746 --> 00:01:06,886 starting to wonder what the pitfalls might be. 21 00:01:07,776 --> 00:01:12,066 So I'm gonna be talking in this episode about how people are using AI 22 00:01:12,066 --> 00:01:16,316 right now, some of the different use cases, and also some of the reasons 23 00:01:16,316 --> 00:01:20,426 to be cautious and some of the things that I've started to see have become 24 00:01:20,426 --> 00:01:24,246 a bit of an issue in some of the businesses that I work with day to day. 25 00:01:25,616 --> 00:01:27,666 So let's start off with the numbers. 26 00:01:27,776 --> 00:01:33,586 And as usual, these vary wildly depending on how you define using AI. 27 00:01:34,176 --> 00:01:38,856 And what that looks like in practice is that 16% of UK businesses are formally 28 00:01:38,886 --> 00:01:41,696 using AI according to government data. 29 00:01:43,086 --> 00:01:48,546 But if you think about it informally, around 50 to 70% of SMEs are using it 30 00:01:48,916 --> 00:01:53,086 according to surveys like the British Chambers of Commerce and Intuit, who've 31 00:01:53,146 --> 00:01:55,416 done a fair amount of research themselves. 32 00:01:56,546 --> 00:02:01,236 And up to 80% of SME decision-makers are saying they've used it in 33 00:02:01,236 --> 00:02:04,716 some way according to research done by Enterprise Nation. 34 00:02:05,756 --> 00:02:09,756 So what this really means is that people saying that they're using AI 35 00:02:10,116 --> 00:02:13,026 is often about writing emails faster. 36 00:02:13,736 --> 00:02:19,544 It's not as deeply embedded yet, but it is becoming the norm. So what 37 00:02:19,544 --> 00:02:20,934 are people actually using it for? 38 00:02:20,934 --> 00:02:25,864 Well, the first main use case that we're seeing for generative AI in 39 00:02:25,864 --> 00:02:30,294 retail business is as an assistant as opposed to a replacement. 40 00:02:30,774 --> 00:02:33,404 So people are actually using it for writing support. 41 00:02:33,874 --> 00:02:38,964 They are using the Shopify tools to help them with product descriptions. 42 00:02:39,444 --> 00:02:44,154 They are getting it to draft them emails, to come up with subject lines for their 43 00:02:44,154 --> 00:02:49,594 newsletters, to help them create social media captions, and to write blogs. 44 00:02:50,364 --> 00:02:55,214 They also might be using it for things like customer communications, so FAQs. 45 00:02:56,104 --> 00:02:59,644 And a lot of people that I've spoken to actually use AI, the number one 46 00:02:59,644 --> 00:03:04,704 thing that they use it for, is to take awkward customer messages and to help 47 00:03:04,734 --> 00:03:08,634 them rewrite it or to come up with a way of diplomatically responding to them. 48 00:03:09,364 --> 00:03:13,384 People are also using AI for a lot of admin, so meeting summaries, 49 00:03:13,484 --> 00:03:16,574 note-taking, and some basic organization. 50 00:03:17,244 --> 00:03:21,184 Another really important role which I think AI can be very good for is 51 00:03:21,744 --> 00:03:26,994 idea generation, so content ideas, campaign angles, or planning prompts. 52 00:03:27,834 --> 00:03:30,344 So what we're seeing with a lot of independent retailers, and I'm sure this 53 00:03:30,344 --> 00:03:34,404 will be a no surprise to anybody, is that they're using AI to help them save time. 54 00:03:34,404 --> 00:03:37,524 Because if it's one thing that I know about independent retailers and 55 00:03:37,524 --> 00:03:40,024 brands is that you are time poor. 56 00:03:40,024 --> 00:03:46,298 I'm sure that you have felt that a lot in your own business So product 57 00:03:46,298 --> 00:03:50,258 pages, let's just talk about those for a minute because that is a major use case 58 00:03:50,258 --> 00:03:54,098 that we're seeing in terms of what people are actually using generative AI for. 59 00:03:55,048 --> 00:03:59,528 And this is important because of the other side of AI, that agentic AI 60 00:03:59,528 --> 00:04:00,918 that we're gonna talk more about. 61 00:04:00,958 --> 00:04:05,388 But AI tools are increasingly reading your product pages, interpreting your 62 00:04:05,388 --> 00:04:08,738 business from the information it finds there, and then using that information 63 00:04:08,738 --> 00:04:13,098 to help with product recommendations, which means that your product descriptions 64 00:04:13,098 --> 00:04:17,978 now need to explain clearly what the product is, who it's for, why it's 65 00:04:17,978 --> 00:04:19,568 useful, and what makes it different. 66 00:04:19,568 --> 00:04:23,808 So weak product pages or product pages without a lot of information 67 00:04:23,808 --> 00:04:28,308 on it can almost become invisible under agentic AI. 68 00:04:29,408 --> 00:04:33,948 So this is where you can really use AI tools to your advantage because 69 00:04:33,948 --> 00:04:39,508 previously I know before the advent of ChatGPT or other generative AI tools, 70 00:04:39,608 --> 00:04:44,148 that writing product descriptions was one of the most tedious, long-winded tasks 71 00:04:44,188 --> 00:04:45,608 that a lot of retailers went through. 72 00:04:46,188 --> 00:04:50,718 And now with, with these tools, we can have an advantage. 73 00:04:50,748 --> 00:04:55,488 You know your products better than anyone, so you can work on it to come 74 00:04:55,488 --> 00:04:57,858 up with product pages that reflect them. 75 00:04:58,298 --> 00:05:00,418 AI rewards clarity, not fluff. 76 00:05:00,748 --> 00:05:04,688 But the one thing that I would say is even though it is easier than ever to 77 00:05:04,688 --> 00:05:09,278 come up with AI descriptions for your product pages, I still think that you 78 00:05:09,278 --> 00:05:14,138 need to really put into it all of your knowledge and all of your thoughts about 79 00:05:14,138 --> 00:05:18,338 the product, why you picked it, and it's gonna come onto something I'm gonna 80 00:05:18,338 --> 00:05:22,638 touch on later, which is the best use for AI is when you give it really detailed 81 00:05:22,788 --> 00:05:26,698 prompts and it can help a lot from that as opposed to just asking it to come 82 00:05:26,698 --> 00:05:28,608 up with generic product descriptions. 83 00:05:29,448 --> 00:05:34,078 But it is useful because you can give them long prompts, and then you can ask it to 84 00:05:34,078 --> 00:05:39,108 format all of this information in a way that means that the product description 85 00:05:39,108 --> 00:05:45,918 will be read more easily by other AI agents. So another place that we're 86 00:05:45,918 --> 00:05:50,548 starting to see people really leaning into when it comes to AI is summarizing 87 00:05:50,548 --> 00:05:55,558 sales data, spotting patterns, identifying best sellers, flagging slow-selling 88 00:05:55,558 --> 00:05:57,978 stock, sense-checking decisions. 89 00:05:58,068 --> 00:06:01,708 And tools are increasingly built into platforms like Shopify, where the 90 00:06:01,708 --> 00:06:06,148 sidekick has now become, frankly, a lot more helpful when it comes to finding 91 00:06:06,148 --> 00:06:08,038 the information that you actually want. 92 00:06:08,738 --> 00:06:13,578 However, this is an area that I've seen both the most advanced but also some 93 00:06:13,578 --> 00:06:16,928 of the biggest issues that I've seen crop up, and that is because a lot of 94 00:06:16,928 --> 00:06:19,938 people have what I call messy data. 95 00:06:20,278 --> 00:06:24,388 So messy data is where there aren't really clear product hierarchies, the 96 00:06:24,388 --> 00:06:29,678 information is not really categorized very well in a way that means that 97 00:06:29,728 --> 00:06:32,858 the analysis is actually that helpful. 98 00:06:33,298 --> 00:06:37,098 So if your categories are inconsistent, and especially if your stock 99 00:06:37,098 --> 00:06:42,278 information is wrong, then all of these capabilities with the AI, they're 100 00:06:42,278 --> 00:06:44,218 just not actually going to help you. 101 00:06:44,298 --> 00:06:49,268 And a lot of the time we need better inputs to have better outputs, and a 102 00:06:49,268 --> 00:06:52,818 lot of the work that we do with Stock Doctor actually ends up being around 103 00:06:52,818 --> 00:06:57,128 cleaning data, making the setup more strategic, making sure that the stock 104 00:06:57,128 --> 00:07:00,958 information's more accurate so you can actually start to leverage it properly. 105 00:07:02,088 --> 00:07:06,208 I would also say that a lot of people have started using it for 106 00:07:06,208 --> 00:07:11,258 that top-level strategy, and there is definitely a confirmation bias in AI. 107 00:07:11,658 --> 00:07:14,018 So we're seeing lots of people asking it, you know, "Should I do 108 00:07:14,018 --> 00:07:18,468 this?" And it, generally speaking, will tell you that it's a good idea. 109 00:07:19,208 --> 00:07:23,148 I definitely see that AI is still very much more tactical as opposed to 110 00:07:23,148 --> 00:07:27,958 strategic, so it is giving you specific answers to specific questions, but 111 00:07:27,958 --> 00:07:30,348 it's not always seeing the full picture 112 00:07:32,216 --> 00:07:36,536 So customer service is another area that people are really leaning into with AI, 113 00:07:36,536 --> 00:07:41,256 and it's being used for handling repeat questions, delivery updates, returns 114 00:07:41,256 --> 00:07:44,046 information, basic recommendations. 115 00:07:44,926 --> 00:07:50,446 So again, with this particular use of AI, I think it's important to look 116 00:07:50,446 --> 00:07:52,976 at what it can do and what it can't do and where it can really help. 117 00:07:53,416 --> 00:07:57,146 And I think, again, as I've seen in lots of other areas of people's businesses, it 118 00:07:57,146 --> 00:07:59,056 works really well when there's a blend. 119 00:07:59,576 --> 00:08:03,116 So you don't want to risk losing the personal touch, you don't 120 00:08:03,116 --> 00:08:06,856 want to sound generic, and you don't want to damage trust. 121 00:08:07,266 --> 00:08:11,636 Your advantage as an independent retailer or brand is personal connection, so 122 00:08:11,666 --> 00:08:13,776 you don't want to automate that away. 123 00:08:14,256 --> 00:08:19,296 That said, you can look at it as what I would call a triage approach. 124 00:08:19,776 --> 00:08:26,356 So even pre-AI, around 80% of queries, the research shows that around 80% 125 00:08:26,356 --> 00:08:30,466 of customer queries could be handled by some kind of automated response. 126 00:08:30,876 --> 00:08:35,966 You know those, the kind, the where is my parcel, how long is the delivery date, 127 00:08:35,966 --> 00:08:37,456 if I order it today, when will I get it? 128 00:08:37,526 --> 00:08:38,516 All of that kind of thing. 129 00:08:39,296 --> 00:08:43,856 So probably 80% of all of those have always been able to be automated, 130 00:08:43,856 --> 00:08:48,726 but now with a lot of plug-ins, AI-powered chatbots, things like 131 00:08:48,786 --> 00:08:53,146 that, it's probably easier than ever to automate the more basic requests. 132 00:08:54,386 --> 00:09:00,426 But what people are really leaning into is where you're using AI to triage the 133 00:09:00,426 --> 00:09:04,286 most basic questions, but then very quickly being able to put a human on 134 00:09:04,286 --> 00:09:10,486 to deal with that 20% of queries that just can't be solved that quickly. 135 00:09:11,096 --> 00:09:16,986 So earlier this year, I spoke to Sarah from Artichoke, and she won 136 00:09:16,986 --> 00:09:21,906 the Boutique Star Award for Online Retailer of the Year, and she, for 137 00:09:21,906 --> 00:09:26,086 example, and her team, after somebody places an online order, they call them. 138 00:09:26,516 --> 00:09:30,026 They call every single customer who places an order because they're 139 00:09:30,116 --> 00:09:35,446 able to take the efficiencies from technological advancements, but then 140 00:09:35,446 --> 00:09:37,126 they're keeping that human element. 141 00:09:37,176 --> 00:09:40,596 And I think that that's where AI works really, really well in retailers is 142 00:09:41,046 --> 00:09:46,386 where you can use those advancements, as I said, to really triage the majority 143 00:09:46,386 --> 00:09:50,026 of those queries, but then make sure that you have got that human touch 144 00:09:50,336 --> 00:09:54,166 ready to go and maybe even more ready to jump in because they have got the 145 00:09:54,166 --> 00:09:58,546 time to focus on a fewer number of maybe trickier problems to solve. 146 00:10:00,152 --> 00:10:03,532 One area that is really controversial, and I think the jury is still out 147 00:10:03,532 --> 00:10:06,432 on in retail, is AI for imagery. 148 00:10:07,832 --> 00:10:12,632 It has its use cases for sure, things like mock-ups, things like backgrounds, 149 00:10:12,782 --> 00:10:17,292 campaign visuals potentially, and maybe using it for concepts. 150 00:10:17,292 --> 00:10:23,072 I've seen a lot of people use AI to almost create very bespoke vision boards 151 00:10:23,142 --> 00:10:29,202 for a remodel or a new store or an idea of what a new product range might look 152 00:10:29,202 --> 00:10:31,292 like, and I think that's really useful. 153 00:10:31,762 --> 00:10:37,272 I think obviously AI imagery is appealing in terms of saving time and saving money, 154 00:10:37,922 --> 00:10:42,972 but I think that there's still a real concern here with things like is it 155 00:10:43,052 --> 00:10:44,982 accurately representing your product? 156 00:10:45,462 --> 00:10:48,192 So it would be brilliant if we could get to the point where you could take 157 00:10:48,242 --> 00:10:56,012 a photo of one product once and then be able to use AI to replicate that 158 00:10:56,012 --> 00:10:57,542 across all of the different colors. 159 00:10:57,972 --> 00:11:02,352 But at the moment, I think that still you would have to be so certain that 160 00:11:02,352 --> 00:11:06,292 something that you did that with would then be correctly represented 161 00:11:06,292 --> 00:11:09,612 so that when the customer receives that item, they weren't disappointed. 162 00:11:10,802 --> 00:11:14,732 That could lead to higher returns, that could lead to loss of trust. 163 00:11:15,152 --> 00:11:18,012 And also, there's this feeling of these off visuals that 164 00:11:18,012 --> 00:11:20,132 customers can't quite explain. 165 00:11:20,572 --> 00:11:26,182 And I've been seeing this rise of certain websites where people have 166 00:11:26,182 --> 00:11:31,112 used AI for all of the photography, including putting up fake photos of the 167 00:11:31,332 --> 00:11:34,652 shop owners to make it look like it's an independent retailer when actually 168 00:11:34,742 --> 00:11:39,112 it's ... Well, it may be an independent retailer, but it's not UK-based even 169 00:11:39,112 --> 00:11:40,302 though it says that in the name. 170 00:11:41,542 --> 00:11:47,042 And I think that there is a slight fear that if you find a, a website 171 00:11:47,082 --> 00:11:52,092 and it feels like the images are AI produced, then it really does give a 172 00:11:52,122 --> 00:11:56,562 level of trust or it drops a level of trust from the consumer right away. 173 00:11:57,042 --> 00:12:01,472 So I think that AI imagery is just something that we have to be really 174 00:12:01,472 --> 00:12:06,422 careful with and to really, really understand what it can be used for, 175 00:12:06,422 --> 00:12:08,082 what's working, and what's not. 176 00:12:08,922 --> 00:12:12,192 And it may well be that it's something that is great for doing touch-ups, 177 00:12:12,442 --> 00:12:16,582 but not yet a replacement for real product imagery, and certainly not real 178 00:12:16,582 --> 00:12:19,642 photographs of you as the shop owner. 179 00:12:20,232 --> 00:12:23,772 And there's also this uncanny valley effect, which is that sort 180 00:12:23,772 --> 00:12:27,582 of visuals being slightly off, which can really affect people's trust. 181 00:12:27,582 --> 00:12:31,512 Because people are really looking for signals that you're a real brand, that 182 00:12:31,512 --> 00:12:34,972 you're a real person behind the brand, that it's a real company, that it's 183 00:12:34,972 --> 00:12:38,882 not a spoof website, that their money's gonna be safe if they spend it with you. 184 00:12:39,252 --> 00:12:43,642 So imagery is a, a really strong, powerful way to prove that. 185 00:12:43,952 --> 00:12:48,792 So I would say that of all of the uses of AI in independent retailers, imagery 186 00:12:48,792 --> 00:12:54,072 is one that is probably potentially the one that we have to really look at the 187 00:12:54,072 --> 00:12:59,834 most and think carefully about what you want to do So most retailers are not 188 00:12:59,834 --> 00:13:04,714 necessarily using AI to build systems, even though now what we're seeing, 189 00:13:04,714 --> 00:13:06,474 it's crazy the speed of things change. 190 00:13:06,744 --> 00:13:11,064 People are able to build apps by themselves, for example, things like that. 191 00:13:11,114 --> 00:13:13,694 They're not necessarily looking to automate everything, but they do 192 00:13:13,694 --> 00:13:18,524 want less admin, faster output, and definitely help getting started. 193 00:13:18,524 --> 00:13:22,684 So if I had to describe the mood of a lot of independent retailers when 194 00:13:22,684 --> 00:13:25,974 it comes to AI, I would say that people, a lot of people are curious, 195 00:13:26,084 --> 00:13:30,374 slightly skeptical, and some people are feeling kind of over the hype, and 196 00:13:30,534 --> 00:13:31,744 maybe that's how you're feeling too. 197 00:13:32,144 --> 00:13:35,424 It's an interesting topic, and we haven't even really touched on the environmental 198 00:13:35,424 --> 00:13:40,544 concerns or people's ethical concerns about using AI, which a lot of people 199 00:13:40,544 --> 00:13:43,514 have very strong feelings about. 200 00:13:43,514 --> 00:13:47,584 So I would say that at the moment, AI in independent retail, it feels like a 201 00:13:47,584 --> 00:13:51,444 lot of people, as I said, are exploring, trying things, seeing what works. 202 00:13:52,164 --> 00:13:55,764 So what would I suggest that you don't use AI for? 203 00:13:56,674 --> 00:14:00,934 Well, I do think that you need to be careful with fully AI written 204 00:14:01,044 --> 00:14:04,794 content, because people are much better at spotting AI now. 205 00:14:04,864 --> 00:14:08,724 Even subconsciously, people notice if the tone feels off, if the 206 00:14:08,724 --> 00:14:12,884 structure feels repetitive, or if the language feels generic. 207 00:14:13,484 --> 00:14:16,954 So for example, I'll use this example from my own business as well. 208 00:14:16,984 --> 00:14:21,474 A couple of years ago, I used one of the early AI tools to help me 209 00:14:21,534 --> 00:14:27,654 revise my onboarding email sequence, and I was able to make it sound more 210 00:14:27,654 --> 00:14:31,084 enthusiastic, which is something that I sometimes find a little bit hard 211 00:14:31,084 --> 00:14:34,754 in the written word to really convey excitement, passion, et cetera, et cetera. 212 00:14:35,084 --> 00:14:39,784 So I'd used this tone adjuster, as it were, from AI, and recently 213 00:14:39,784 --> 00:14:42,614 I was re-editing it to make sure it was all still accurate, and 214 00:14:42,704 --> 00:14:44,344 honestly, I was completely mortified. 215 00:14:45,174 --> 00:14:48,634 And, and in fact, if you've ever received these emails from me, then 216 00:14:48,634 --> 00:14:55,084 I apologize, because it... The shift already from what feels acceptable, it 217 00:14:55,084 --> 00:14:56,864 is now... It was so clearly AI written. 218 00:14:56,864 --> 00:15:01,144 It had, it had rocket emojis and confetti cannon emojis all over it, 219 00:15:01,144 --> 00:15:05,904 and I feel like what passed even a year ago doesn't necessarily pass now. 220 00:15:05,934 --> 00:15:09,354 So at the time, everyone, when it first came out, we were all just so excited. 221 00:15:09,354 --> 00:15:10,774 It was like magic. 222 00:15:10,964 --> 00:15:12,734 I mean, it still is like magic, a lot of it. 223 00:15:13,364 --> 00:15:17,784 But a lot of people used it for a lot of things, and now because it's so 224 00:15:18,154 --> 00:15:22,764 ubiquitous, everyone just knows that there's so much AI out there, that I 225 00:15:22,764 --> 00:15:26,824 think that w- I'm actually seeing a lot of people reverse AI text in their business. 226 00:15:26,824 --> 00:15:28,774 In other words, take it out and write it instead. 227 00:15:29,414 --> 00:15:31,964 So I would say watch your numbers closely. 228 00:15:31,964 --> 00:15:35,054 If you're using AI heavily, check your email open rates, check your 229 00:15:35,114 --> 00:15:36,734 click rates, check your conversion. 230 00:15:37,294 --> 00:15:42,144 I'm already seeing people, as I said, removing AI written text and actually 231 00:15:42,144 --> 00:15:43,734 seeing their performance improve. 232 00:15:43,734 --> 00:15:46,584 So if something drops, I would just ask yourself, "Does this actually 233 00:15:46,584 --> 00:15:49,644 sound like me, or does it sound like everyone or everything else?" 234 00:15:50,344 --> 00:15:55,824 So what I think AI has absolutely meant is that a lot of retailers should no 235 00:15:56,204 --> 00:15:59,854 longer or should ever have a situation of been sitting there and thinking, "I 236 00:15:59,854 --> 00:16:03,164 don't know what to do. I don't know what ideas I should do. I, I can't think of 237 00:16:03,164 --> 00:16:09,374 any ideas for my next email, my next campaign." You absolutely 100% can use 238 00:16:09,484 --> 00:16:11,334 AI to help you generate those ideas. 239 00:16:11,334 --> 00:16:12,604 It's really, really good for ideas. 240 00:16:12,954 --> 00:16:16,994 It's really good for structure, and it's really great for getting started. 241 00:16:17,064 --> 00:16:20,704 But your voice is your differentiation and is your brand. 242 00:16:21,134 --> 00:16:24,344 My favorite approach or what I've seen people using and that works 243 00:16:24,344 --> 00:16:26,464 really well is to talk it out. 244 00:16:26,644 --> 00:16:31,484 So voice note it, brain dump it, and then get AI to tidy it up. 245 00:16:31,824 --> 00:16:36,504 Because there's a huge world of difference between spending 10 or 15 minutes brain 246 00:16:36,504 --> 00:16:41,192 dumping your thoughts around a topic, thinking what you feel about something, 247 00:16:41,192 --> 00:16:44,902 what you'd like to say, all of the kind of things that you'd like to get across to 248 00:16:45,022 --> 00:16:51,222 your customer, and then asking AI to put it into some post ideas or to clean it up. 249 00:16:51,862 --> 00:16:55,402 That will get you a lot closer to where you want to be than if you just say to 250 00:16:55,492 --> 00:17:00,042 a generic tool, marketing tool, "Create me a marketing campaign." The more you 251 00:17:00,042 --> 00:17:04,782 can put into it, the more likely you are to get something out that sounds close. 252 00:17:06,092 --> 00:17:09,952 I would say as well that I would be very careful with using 253 00:17:10,032 --> 00:17:11,492 AI to- for decision-making. 254 00:17:11,532 --> 00:17:13,942 It definitely has a confirmation bias problem. 255 00:17:14,152 --> 00:17:17,332 If you say, "I think I should do X," it will say, "Yes, that's a great 256 00:17:17,332 --> 00:17:20,912 idea." It struggles with nuance, it struggles with context, and it 257 00:17:20,912 --> 00:17:23,682 struggles with seeing the whole picture. 258 00:17:23,682 --> 00:17:26,642 And it can miss really obvious practical issues. 259 00:17:27,372 --> 00:17:31,462 You can ask it to challenge you, to ask questions, and you can ask 260 00:17:31,462 --> 00:17:33,512 it to summarize your thinking. 261 00:17:33,772 --> 00:17:36,902 It can be a really powerful tool to get it ... To, to say to it, "I want you 262 00:17:36,902 --> 00:17:41,102 to challenge me. I want you to ask me a whole load of questions," and for you to 263 00:17:41,212 --> 00:17:44,902 verbally process whatever decision it is that you want to make, rather than just 264 00:17:44,972 --> 00:17:47,272 asking it to give you the right answer. 265 00:17:48,222 --> 00:17:52,752 So using it for thinking and rather than deciding is a better 266 00:17:52,812 --> 00:17:55,282 case, in my opinion, for using AI. 267 00:17:56,182 --> 00:18:00,152 So it's very different to talk for 20 minutes and process out loud and 268 00:18:00,152 --> 00:18:04,952 then get it to summarize it for you than it is to tell me what to do. 269 00:18:05,672 --> 00:18:08,702 So this is how I put it in 2026. 270 00:18:08,852 --> 00:18:12,252 There is no reason at all to sit in front of a blank page. 271 00:18:12,282 --> 00:18:15,992 But there is every reason to sound like yourself. 272 00:18:16,362 --> 00:18:18,382 The retailers who get the most from AI won't be the 273 00:18:18,382 --> 00:18:19,862 ones who hand everything over. 274 00:18:19,862 --> 00:18:24,762 They'll be the ones who use it to think better, move faster, and stay distinct. 275 00:18:25,582 --> 00:18:26,932 Thank you so much for listening. 276 00:18:27,102 --> 00:18:30,002 Don't forget to like, follow, or subscribe, depending on where you 277 00:18:30,002 --> 00:18:33,132 listen to your podcast, to be the first to know about each new episode. 278 00:18:33,402 --> 00:18:37,072 Why not come over to Instagram at Resilient Retail Club and tell me 279 00:18:37,112 --> 00:18:38,592 what you thought of today's episode. 280 00:18:39,162 --> 00:18:42,212 And then in a couple of weeks' time, we're going to be talking 281 00:18:42,312 --> 00:18:46,942 about agentic AI, which is when it's actually the robots doing the shopping