Episode Categories:
Resources:
Join the discussion: Ask Jordan and Bryan questions on AI in retail
Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineup
Connect with Jordan Wilson: LinkedIn Profile
Try Our Free AI Prompting Course: Register for our free Prime, Prompt, and Polish AI Course!
The Progressive Surge of AI in Retail
The incorporation of Artificial Intelligence (AI) into the retail sector offers tremendous opportunities for retailers seeking innovation. AI has seen exponential growth, and its impact on elements of retail such as personalization, search, content creation, and more, has proven significant. From forecasting and supply chain to pricing and now, text and image-based personalizations, AI has deeply nested itself in the heart of retail technology.
Transformative or Tool? AI Perspective in Retail
While viewing AI as a transformative technology may instigate excitement, it's critical to remember that AI serves optimally as a problem-solving tool. AI’s potential may be more effectively seized when used to incrementally tackle specific challenges instead of seeking a revolution overnight. Coupled with good training, clear governance, defined responsibilities, and a purpose-driven plan, AI can usher in efficiencies in retail like no other.
Accounting for the Challenges in AI Implementation
Yet harnessing the capabilities of AI does not come without its challenges. The integration of AI into retail operations may seem alluring, but it's vital to note that such integration necessitates extensive training, infrastructure, and understanding. Moreover, knowing the degree of training specific to business needs and identifying competent data sources can be daunting tasks for businesses.
AI: The "New Employee" Analogy
Analogous to managing a new employee, AI requires training, pertinent information, governance, and well-defined tasks. The process of training AI should be gradual and incremental. This step-by-step training approach ensures that AI effectively learns its roles and responsibilities at a measured pace, allowing for productive symbiosis between humans and AI in the retail industry.
Fashioning an Effective AI Strategy for Retail
To begin, identify the significant gaps in user and shopper experience and the commercial benefits offered by solving these gaps. Instead of diving headlong into large-scale issues, it's advisable to solve smaller, singular problems, testing and monitoring these solutions before connecting them.
Beyond the customer experience, AI offers opportunities to optimize business processes, which may not directly impact customers but offer immense commercial benefits. It's worth considering these 'back-end' advancements as part of a holistic AI strategy.
The Rise of Generative AI (GenAI) in Retail
The emerging usage of GenAI in retail holds interesting potential. Enhanced chatbots, seeming human-like and culturally attuned pave the road for genuinely personalized experiences. Developed further, these more 'human' AI could also begin predicting needs and automating operations, revolutionizing how retailers interact with consumers.
GenAI's Influence on Search and Advertising
With digital assets like search and AI-driven search and answers engines evolving, AI's impact on search and advertising can't be underestimated. The integration of AI with these functions can change how products are discovered and presented to the consumers based on their behavior, creating a more intuitive and engaging shopping experience.
In conclusion, AI holds immense potential for retailers. As the retail industry continues to adapt to this rapidly evolving technology, understanding and carefully implementing AI strategies could offer unprecedented advantages. All the while, the focus should remain on personifying the customer experience, increasing operational efficiency, and enhancing overall profitability. AI is not merely a technological innovation; it’s a business problem-solving model.
Topics Covered in This Episode
1. Current Commerce Ecosystem
2. AI in Retail
3. Potential Dangers of AI in Retail
4. Role of AI in Content Creation
Podcast Transcript
Jordan Wilson [00:00:16]:
You probably know this by now, but AI actually has a big impact on a lot of how we work, but it probably also has a big impact on how you buy things. Right? How you live your life out in the real world. And AI is changing things very very quickly when it comes to retail. Where we spend our money, how we spend our money. And although it has changed quickly in the last couple of years with the big chat gbt moment, I think that we're in for a lot more innovation than we've experienced so far. So we're gonna be talking about that today and a lot more on everyday AI. What's going on y'all? My name is Jordan Wilson, and I'm the host of everyday AI, and this thing is for you. It is your daily podcast, livestream, free daily newsletter helping us all keep up and get ahead with everything that's happening in the world of generative AI.
Jordan Wilson [00:01:08]:
So if that sounds like you, you are in the right place. Thank you for tuning in. Before we get started, have to give a shout out to our sponsors at Microsoft. So do you know about Microsoft WorkLab? Well, why should you listen to the WorkLab podcast from Microsoft? Because it's made for leaders who know they must adapt to say ahead. WorkLab is the place to find real world lessons and actionable insights to prepare you for the next phase of AI at work. That's worklab, no spaces, available wherever you get your podcast. Alright. Well, this podcast, like I said, is everyday AI.
Jordan Wilson [00:01:45]:
If you haven't already, please go to your everydayai.com. Sign up for that free daily newsletter. We're gonna be recapping all of the AI news and today's podcast in to the newsletter. So make sure you go check that out. Alright. Before we talk about AI's impact on retail and y'all, it is a big one. Let's first quickly do as we do every single day and start off with the AI news. So first, there's some meta privacy confusion online when it comes to AI.
Jordan Wilson [00:02:11]:
Who would have thought? So recently, a viral post has emerged across Meta's platforms claiming users can protect their personal data from being used by the company and its AI. This message has gained a lot of traction, but it lacks any real late and wait, and it's not gonna be effective. So the statement which begins with goodbye meta AI suggests that all users must post it to prevent the company from using their information and photos. So this copypasta appears to have started in early September and follows a similar wave of misinformation from May where users share posts about not permitting data access. So Meta's terms and conditions clearly state that users allow the company to use their publicly shared posts, including photos and text for training AI when they create an account. So, yeah, you can't block them now. Sorry. Alright.
Jordan Wilson [00:03:00]:
Next piece of AI news. NotebookLM from Google has enhanced its learning experience with new audio and video features. So users for Google's notebookLM can now upload public YouTube URLs and audio files into notebookLM, expanding the range of source materials available for analysis. So if you don't know, the tool provides inline citations linked to video transcripts allowing for deeper exploration of key concepts presented in lectures and videos. So the new feature, in addition to this, also we talked about this in the show is sharing, audio overviews that can essentially generate a personalized podcast. So yeah. I'm I'm personally huge fan of notebook l m out of the 1,000 plus tools I've used over the last, year that our AI, this is definitely one of the, one of my favorites. Alright.
Jordan Wilson [00:03:50]:
Our last piece of AI news. UK regulators have cleared Amazon's 4,000,000,000, billion with a b, $4,000,000,000 AI partnership with Anthropic. So the UK's Competition and Markets Authority, or CMA, has decided not to investigate Amazon's partnership with the AI startup, Anthropic, which includes a significant $4,000,000,000 investment. So this decision is pretty noteworthy as it reflects the regulator's stance on competition in the rapidly evolving AI landscape. So the CMA concluded that Amazon's collaboration with Anthropic does not raise any competition concerns, so therefore, it's avoiding a deeper probe under Britain's merger regulations. An Amazon spokesperson welcomed the CMA's decision, emphasizing its acknowledgment of jurisdiction limitations regarding the partnership. So Infropic obviously is cofounded by former OpenAI executives, and it has attracted substantial investments from various tech giants including Amazon. Alright.
Jordan Wilson [00:04:50]:
So there's a lot more AI news. If you haven't already, please make sure you go sign up for the daily newsletter at your everydayai.com. Hey, John and Marie and Jackie and everyone else. Thanks for joining us to our livestream audience. Podcast audience, you can always, come back and watch this video. But, I'm very excited for today's conversation. We have a leader in the retail, in the commerce industry. I'm extremely excited, to have today's guest on the show.
Jordan Wilson [00:05:17]:
So please help me welcome y'all. There we have him, Bryan Gildenberg, the founder and the founder CEO of Confluencer Commerce. Bryan, thank you so much for joining the
Bryan Gildenberg [00:05:26]:
show. Thanks for having me on, Jordan. It's a great pleasure to be here. So
Jordan Wilson [00:05:30]:
Alright. I'm excited for this one. And, hey, everyone joining us live. Please, if you have questions for Bryan about the future of AI in retail, get them in now. But let's start at the top. Can you tell us a little bit more, Bryan, about what you do at Confluence or Commerce?
Bryan Gildenberg [00:05:44]:
Yeah. Sure. So, so, first of all, for those of you I've never met, which is from the comments, most of you, it's a pleasure to be here. Bryan Gillenberg, so, founder of, Confluence of Commerce and also the managing director for Retail Cities, a, retail research firm. I've been around the retail industry for, you know, probably close to 30 years now. And, most of my career was in a business called Kantar, where we studied the evolution of, global retail, going back to the pre Internet world. So, so I started studying retail the day Jeff Bezos wrote his first letter to shareholders in July 15, 1997. So, so, yes, I've been been around the space for a while.
Bryan Gildenberg [00:06:22]:
Worked at Omnicom for a couple of years, helping, an agency kinda build its commerce strategy. And for the last couple of years, been on my own, mostly helping large brands and large retailers understand the transformation of the commerce, media, and content landscape and the confluence of those three things and how that's gonna change the way they go to market. And obviously, over the last 18 months or so, this AI thing has become kind of a big deal. So, we're paying it pay more attention to this than it would have, say, 2 years ago. So
Jordan Wilson [00:06:50]:
Yeah. Exactly. Right. And, artificial intelligence is not new. So, but this whole generative AI movement is. But, Bryan, maybe can you catch us up for those of us that aren't super familiar with with retail industries, commerce industries? How has AI traditionally been used? And then where are we at now with this generative AI boom? I know that's a lot to ask, but give us give us a high level there.
Bryan Gildenberg [00:07:14]:
Oh, sure. Well, that's simple. Well, I would say that I would say, look. I think there's, layers to AI. Right? You know this better than I do. AI as a concept has been used in retail for decades, as an extension of machine learning to help scenario plan and model a operationally complex and granular business. Retailers were using AI to model things like forecasting and supply chain and pricing and all the things that you would think, that from, from a mathematical point of view, quantitative AI has been a big part of the retail landscape forever. I think when you look at GenAI now, the more text and image text content and image based AI that sort of, become all the rage since ChatGPT entered our collective consciousness, you've got, I think, use cases today, which are, I think, conceptually quite strong, particularly around the topic of personalization.
Bryan Gildenberg [00:08:04]:
Right? And personalization has been a parallel theme in retail for years as well. So if you look if you look at a large grocery retailer like Kroger in the US or Tesco in the UK, they've been personalizing promotions for people for years. Right? Like, you know, Tesco was sending out 8 or 10,000,000 different circulars every week when you were mailing circulars to people's homes based on your purchase behavior, based off of the data that they would collect off that relatively innocent piece of plastic you have called a loyalty card. So, so I could then take that, figure out what you were most likely to buy, and then tailor a target promotion based on that. So the so quantitative personalization of retail has been a big deal for a long time. The, I think the application and potential use cases today are to take all of that data that I have on somebody, be able to layer GenAI on top of that, and deliver them a genuinely personalized experience. And, to go back to Jeff Bezos, I think we are very much in day 1 of that rather than day 2 or day 3. So I think right now, a lot of the use cases for GenAI in retail, they get a lot of buzz and people like to talk about them in industry conferences.
Bryan Gildenberg [00:09:10]:
But right now, what you're looking at is sort of basically an enhanced chatbot, for the most part. So, you know, how do I use AI to be able to provide a technical service experience for somebody that feels more like a human being? GenAI allows you to do that by targeting that messaging, targeting the content, in many cases, making it culturally appropriate so that, you know, the AI sounds like a teenager or sounds like a Latino consumer or whatever it is. You can do a bunch of different things within the context of that. But right now, I still think we're in pretty early days.
Jordan Wilson [00:09:40]:
So no, Bryan, you bring up an interesting point. Right? Like, is AI right now in in retail and commerce essentially just a personalized chatbot? But, you know, maybe even asking. So when this kind of chat gbt moment happened, I think everyone in every industry was, like, you know, looking at the long run and saying, oh, this is gonna shake up all of these trees. I guess have all of those trees been shaken yet? Is there still a lot that generative AI can do, Or do you think for the most part, it has just been like, okay. Well, we have more personalized chatbots that are, you know, driving sales?
Bryan Gildenberg [00:10:13]:
Well, I think there are, the short answer is no. I don't think we're anywhere near where this is going to go. Right? So, you know, it's a I think Bill Gates once famously said the most common mistake people make is to overestimate the pace of tech change in the short term and dramatically underestimate it in the long term. I think that's exactly where we are with, with GenAI at the moment. I think right now, the most interesting use cases for AI on the retail side and both Walmart talked about this a lot in their earnings call last quarter, aren't on the consumer facing side, but on the business facing side. Because as it's particularly Walmart and Amazon, which have large third party, reseller marketplaces that they're bringing to market, that involves 100 of 1,000 or in some cases, millions of sellers. The ability to put AI as a layer so that those sellers can access the platform and vary content and buy advertising more quickly and effectively. That's been the single biggest commercial use case for AI and Commerce today.
Bryan Gildenberg [00:11:09]:
Isn't on the consumer facing side, but on the business side. I think on the consumer facing side, there's so many interesting applications that could come out. I'll just give you one simple one. Right? So there's a company called Fetch in our space today that basically takes all of the receipts that you have, collects them all, you know, aggregates that data and resells it, but then also provides you basically discounts based on how much you spent. Right? So, basically, they're paying you to give them data so they can resell it. So it's a good business model. It works for everybody. Cool company.
Bryan Gildenberg [00:11:40]:
Imagine if somebody decided to do that with all your receipts, but instead of that, get all of your receipts through an AI engine. And now all of a sudden, you've got an AI engine that's basically your auto replenishment for your groceries for a year. So, so that you've got the so that you've got this whole thing now where basically, I've got what I've bought in the grocery sector for the last 52 weeks or something from everywhere, and then you've got a predictive AI engine that basically tells you what your replenishment cycles look like for various and sundry things so that you take 80 to 90% of the thinking and decision making out of your everyday grocery purchase. I think those are the types of applications that are gonna get really interesting over time, and that those are gonna fundamentally, shape and transform how people decide, where the where and when they're gonna shop.
Jordan Wilson [00:12:28]:
So, you know, Bryan, you brought up a very interesting point there at the beginning of your response, you you know, drawing on this, you know, Bill Gates, quote about overestimating in the short term and underestimating the long term. So how do you think maybe retailers are underestimating the long term of generative AI? Maybe where are they missing that big opportunity?
Bryan Gildenberg [00:12:51]:
Well, I I think right now I don't know if this is an underestimation, but I think it's just a a lack of clarity on where the single biggest, digital asset today that retailers have to create demand and sell advertising is search. So if you think about what GenAI at a simple level can and will over time view to search and how ad supported that may be versus how organic that may be, that's gonna be a foundational question as this unfolds. So, you know, you've already got Perplexity today who's looking to try to figure out how to monetize their GenAI platform from, from an advertising perspective. Now they're they're not gonna introduce, as of now, ads into the search results. They're putting up ads around the search results. So that's a, that's a thing. But over time, every every content engine in the history of the world has sowed the seeds of its own destruction by by then embedding advertising within that. And, I do think that over time, that balancing act is gonna be interesting.
Bryan Gildenberg [00:13:51]:
But the whole the whole notion of how search evolves and then how GenAI influences that and and how people use GenAI instead of search, or in addition to search or how well to search is a really interesting question. I also think, and, that as you look today, so much of product discovery today in the world is being driven by AI, but it's an AI that isn't a retailer one or even necessarily a tech platform like Amazon. It's TikTok's algorithm. Right? Like, so much new product discovery takes place on TikTok today, and all of that is algorithmically served up to individuals based on their content consumption behavior. And today, you know, TikTok for people under 35, TikTok's the primary way that people find out about anything. So, so, so that right now is the single biggest use case, that we see for AI in the world is just trying to understand TikTok's content algorithm to the degree that you could then, that you can then put new product in front of people. So
Jordan Wilson [00:14:48]:
Yeah. I'm I'm glad to be that in that over 35 group because I don't understand TikTok and dancing and pointing at things. But, yeah, apparently, that works, and that's the future of, you know, buying products. But, you know, Bryan, I I wanna go back to something that you talked about that I think is huge here. Right? So with commerce, retailers, like, searches is huge. Right? It can't be understated. And, you know, just so everyone knows, background. Right? I'm I'm 15 years I've been in and out of SEO, and I've never once said, you know, traditional search is dead.
Jordan Wilson [00:15:17]:
Right? Even though they've been saying that for 15, 15 years. But with AI, with, these answers engines like perplexity and AI overviews from Google, I mean, you have to start thinking about is traditional search going to work the same way? So maybe for people out there, whether they're they're retailers or consumers even, right, using these perplexity or, you know, now you see Google is just, you know, kind of auto suggesting these things with these AI overviews. How does that impact big retailers and consumers, and what should they be aware of, right, when they're, oh, now all of a sudden just being served these things by AI?
Bryan Gildenberg [00:15:53]:
Yeah. Well, fortunately, having the blessing of being markedly over 35, I feel like I've been to the I feel like I've been to something like this movie a couple of times before, and, the closest parallel I can think of is, you know, let's call it 10, 12 years ago, when people started to figure out that, that Instagram was replacing Facebook. Right? And at the time, no one could really figure out how to, you know, no one could really figure out how to to buy Instagram search for the most part. Right? So, so that what people were, so that what people were doing was that they were, they were trying to figure out how to get the how to get, how to take the organic, environment and figure out how to adapt how to how to adapt to the search. So I think today that what you've got is so that for a long time on the, on the TikTok side, you've got a, you've got a bunch of people now that are trying to use, trying to figure how to crack the code on that organically without being able to without being able to buy not being able to buy search that way. So, I think there's an enormous amount of work being done right now in the commerce ecosystem around, around figuring out how do you get content that gets picked up by that algorithm, how do you then how do you then embed that content? And then most importantly, for bigger companies, which are where I spend most of my time, how do you then scale that?
Jordan Wilson [00:17:09]:
Yeah. So, you know, one other thing I want to talk about is we, you know, said, hey. Is AI right now, for the most part, retailers and commerce, is it just kind of a glorified chatbot, you know, a better chatbot? Alright. How do you think, this has played out so far? Right? Because at least for me even, I think a big part of of, you know, retail and commerce for me is I'm asking questions about products either before I buy them or after, you know, with, you know, following up on support and those things. Has generative AI actually been a, quote, unquote, game changer for at least that one specific part of this process, or would it still just be something you'd say is incremental?
Bryan Gildenberg [00:17:50]:
I think right now, it's still in, I think a lot of it's still early days on the on from a piloting point of view. I think people are still learning how to harness, how to harness the technology and then, figuring out that rather than trying to replace a person with AI, which often doesn't work quite as well as you would hope it would, is providing access points that a person wouldn't be able to do or get to, and then layering AI in to create a better experience than nothing. Right? So when that's sort of, you know, as I, as I got fond of teaching my kids when they were doing math at an early it's, you know, when you do it when you divide something by 0, literally anything is infinitely better than nothing. So, so, so and that that I think is where the most interesting start cases are where I can reach more people and more use cases more easily, create more touch points to create a semblance of a personal relationship where there wasn't one before rather than trying to replace a personal relationship with AI, which almost never goes well. Like, if I you know, and, like an example that a friend of mine uses all the time is Erica, the Bank of America sort of AI. You know, Erica is incredibly clear that it is not a person. Right? So it's like, hi. I'm an AI engine.
Bryan Gildenberg [00:19:07]:
I can help you do things that otherwise would be a pain in the neck. So so here's how we're gonna do that if you need to talk to a person click here like I think that's a very real way to set the problem up as opposed to calling into a cut or logging into a customer service experience thinking you're going to get a person and then getting AI back. I think that creates disappointment. So, so I think GenAI allows you to get to a broader set of use cases from a from a back and forth perspective and to create the sense that somebody's getting their question answered, which I think is great. So so, I also think too that, the other interesting application for AI as you look at the consumer facing world is not from a text point of view, but from a content point of view. So Amazon talks a lot about this with its 3rd party sellers that they're creating an ecosystem where it's much easier for sellers to create to create images in AI, to vary their images in AI. Like, if you look at anybody that knows anything about, selling things on Amazon, the frequency with which you refresh your content is an important part of winning consumer attention out of winning organic search on Amazon. So the ability to do that quickly and chiefly through AI then creates a better experience overall, not just for the user because they get something new and refreshed, but for the brand because they it allows them to vary their content without having to have an individual deployed against that, that content variation.
Bryan Gildenberg [00:20:31]:
So
Jordan Wilson [00:20:32]:
Yeah. And and what you know, I'm glad we can get into this because I even personally think that's one of the more exciting and, you know, low hanging fruit use cases of generative AI for commerce and retail. Yeah. It's like we don't need that same, you know, 1 or 2 product photos that look like they're from 1990, but maybe what dangers are there in that? Because I, you know, I remember reading a lot of stories early on in the generative AI phase when the, you know, AI image generators weren't that great and people really didn't know how to use the large language models. But, you know, are there some maybe dangers in, you know, retailers just blanket, you know, putting gen a out GenAI out in the wild without enough human in the loop. What what dangers are there, and and what should retailers be looking at to avoid some of those common pitfalls?
Bryan Gildenberg [00:21:20]:
Well, yeah. I mean, I think there are there are clearly there are clearly dangers that, that come from, you know, poorly trained and hallucinating AI. And, you know, every every everybody on this knows more about this than I do. But, but, yeah, you wanna try to avoid the, you know, what some of some of the challenges of things like Gemini have obviously run into over the, you know, in the recent past that are well publicized. Yeah. We're all learning. Right? So, so so, yes, I think there's a I think there's certainly a risk for, you know, hallucinatory content or I think the bigger I think the actual bigger risk, is another friend of mine, describes this as, not so much that the content's wrong, it's that it's homogenized. So the the idea that the biggest risk from AI is not, you know, black popes everywhere, but that we're drowning in a sea of sameness.
Bryan Gildenberg [00:22:11]:
And, and that, that AI is going to optimize for content that's just going to be remarkably similar to itself. Right? And the metaphor that always sticks in my head, again being old, is that, you know, I'm old enough to remember when everybody discovered clip art in PowerPoint, and everybody had the same 9 pieces of artwork in their PowerPoint slides because they thought they're being creative, but it was always the same 9 things with the the same the same stock photos of people with their arms in the air excited about stuff. And, I do worry that we are from a content point of view going to be headed towards a homogenized and AI produced content ecosystem to the degree that the phrase artisanal content popped into my head one day. So I think that you're going to see agencies in the future. They're gonna talk about artisanal content created by humans for humans. So so much so that I bought the URL, artisanalcontent.com. So it's okay. It's in there.
Jordan Wilson [00:23:09]:
I love it. I love, like like, grabbing those, like, like, little URLs. I've been doing this for a long time. Yeah. Me too. That's that's a good one for the future. You bring up a good point. I was actually talking with a a good friend about this, you know, maybe 6 months ago.
Jordan Wilson [00:23:22]:
This concept of, yeah, will things created by humans in the future almost be, yeah, artisanal. Right? So, yeah,
Bryan Gildenberg [00:23:29]:
it'll be it'll be organic content, but not organic like the way we talk about organic content today, but organic, like, handcrafted content. Like, here is your curated heirloom image. Right? So, you know, like an heirloom tomato at a farmer's market.
Jordan Wilson [00:23:44]:
So Yeah. Yeah. A wild world to think about, but it's it's it's not actually too hard to see that becoming the truth. Yeah. And if so, you you have the URL there, Bryan. So, you you know, one thing that you you talked about, a couple of minutes ago is kind of retailers layering in AI. And I'm wondering if maybe, especially after you just gave this clip art example, if they're just layering it in too quickly, too haphazardly, you know, maybe where they're just trying to sprinkle some AI on everything and not necessarily you know, training it or, you know, making sure it's it's, you know, using rag to bring in their own data to to kind of tune different models. Are there problems with that? Are are are companies maybe not training, AI like they should be and just using it at scale?
Bryan Gildenberg [00:24:30]:
Oh, god. Yeah. So, yeah, there's, like, hundreds of them. Yeah. I think there's I think there's a series of interesting issues. I think that, you know, to your good point, I think most of the challenges that people have with AI is they just don't really understand the value creation process in AI. They don't understand the enormous amount of work that needs to go into training AI, both from a workflow point of view, but also from a data point of view. So where and where and how am I gonna decide the degree to which I need the AI that's powering my business to be trained in a specific way to me? Right? And then if I'm gonna do that, where's the data gonna come from to do that? Like, you know, what's what you know, what's what fuels this content engine? Like, you know, so many of the clients that I work with today saying, oh, AI, we need data scientists.
Bryan Gildenberg [00:25:19]:
We gotta go figure out to say I think it's like, you know, like data scientists do one thing what you actually need are data engineers right like you need people like the way I was the way I put this to a client once they were like we're having a bunch of data scientists not doing anything. It's like it's like opening a restaurant having 12 chefs but no food so, like, you know, data scientists are chefs. Data engineers make food. And, what you've gotta do is you've gotta figure out where and how you're going to use the publicly available or, you know, licensable LLMs, the degree to which you wanna custom train them, the resources that it takes to custom train them to get to the outcomes that you need for your business, the data requirements that that's got, and then, of course, the management of that process. So, you know, one of the things that's always occurred to me with AI, it's like, well, you need to train it. You need to make sure it's got the right information. You need to govern it. You know, you know, early on, you really gotta make sure that it's doing the right things.
Bryan Gildenberg [00:26:12]:
It does well with a well scripted, well defined job description. In the end, for all the talk about how AI is a techno technological transformation, it sounds more like an employee than anything. That's right. Like, the way people talk about AI is exactly the way I would talk about a new hire. Right? Like, you know, 90 days in, you gotta make sure this person knows what they're doing. You gotta train them right. You gotta gotta make sure they've got governance so they don't do something really weird and stupid. Mhmm.
Bryan Gildenberg [00:26:35]:
Like, they'll learn more over time, and they're gonna do really well with a fairly well curated inspect job description. You know, the, the metaphor my, the metaphor a friend of mine uses all the time talking about this with training AI. It's like, you know, training AI is a bit like training a dog. Right? Like, you don't wanna train the dog to sit and roll over at the same time. Right? You know, you get sit nailed, you know, sit, treat, sit, treat, sit, treat, sit, pet, and then eventually it's sit, then you move on to roll over or fetch. Right? So just training one thing at a time is a skill set, and I think that the the really differentiated skill for companies going forward won't necessarily be how fast their machines learn. It's really gonna be how well their people can teach. So Yeah.
Bryan Gildenberg [00:27:21]:
And machine teaching seems like a competency rather than machine learning.
Jordan Wilson [00:27:25]:
Yeah. And and that's a great point. And I do have 1 or 2 more questions for you, on that, Bryan. But real quick, I do have to shout out Microsoft here in the WorkLab podcast. So if you don't know, the WorkLab podcast from Microsoft is made for leaders who want to understand how work is changing. So effective leaders, they adapt, they stay ahead of trends, and they embrace any edge they can get. They also know that AI powered organizations will be better at spotting opportunities, creating new products and business models, and maximizing value. So for real world lessons and actionable insights to help you stay ahead, check out the WorkLab podcast.
Jordan Wilson [00:28:03]:
That's worklab, no spaces, available wherever you get your podcasts. Alright. Yeah. You like, y'all gotta go check it out. They just dropped a new episode. So definitely worth, you know, listening if you haven't already. So Bryan, getting back to that, I love that analogy. Right? You can't you can't teach a dog to sit, stay, and roll over, and shake all at the same time.
Jordan Wilson [00:28:23]:
You kinda have to take it piece by piece, step by step. So maybe with that, where should right now like, you know, you've given a lot of great advice already here. But if someone, you know, out there, they're a a retail leader, and they're hearing like, oh, yeah. Yeah. We've been guilty of that. How should organizations maybe get their AI strategy right? Or maybe for those that haven't even built one out yet, where should they be looking?
Bryan Gildenberg [00:28:47]:
Well, I I think that as with many things, I think you have to start with the business opportunity first. Right? Like, what's the what's the biggest gap between where your where your user slash shopper experience is today and where you want it to be? And then what's the biggest gap in the experience? And then clearly, secondarily, what's the commercial benefit of solving that problem? Right? So if you take those two things together, you then get to the the simplest 2 by 2 matrix in the world. Now if I've got high commercial opportunity and a massive gap in user experience, that's an easy one. Right? So I'm gonna deploy resources against that. And then what you need to do is just need to do a good old fashioned value chain exercise of what are the steps that it takes to get here. And then rather than saying, how do I get AI to solve that problem? How do I use AI to solve the steps along the way and the point solutions that go along the way? And then figure out, you know and this is one of the things that Amazon's brilliant at. Right? What Amazon's really good at even in the pre AI world is not trying to solve tech problems en masse. They solve them by what Andy Jassy refers to as primitives.
Bryan Gildenberg [00:29:59]:
Right? So I'm gonna build little tiny building blocks that work, and then I'll figure out how to connect the things that work together. So but I'm going to do really small specific things that fix a specific problem, then the connectivity is my is it will be my secret sauce later. But if I try to do something that's too connected early on, that tends to be geometrically harder. You know, solving 3 problems at once is 8 times harder than solving 1, not 3 times harder. So solve specific problems along the way, fix those things, test, monitor, and get them working. And then the secondary use case is how you tie those things together. So I think that's 1. And then, yeah, I think you're gonna find that there are opportunities that have very high commercial outcomes, but that are kind of invisible to the user.
Bryan Gildenberg [00:30:45]:
One of your users jumped in earlier and said that supply chain is one of the biggest applications for AI. I would agree with, I would agree with Jackie on that front. There's a ton you can do behind the scenes to create tremendous commercial outcomes for your business that may not be entirely visible to the consumer. And then there's gonna be things that make the consumer experience better, but they don't have an obvious commercial outcome or an obvious short term commercial outcome. That's where that's that's where the vision thing comes in. Right? And it's like you just gotta trust that you have an idea of what your shopper experience needs to be, and that I'm gonna spend a little bit of money trying to understand how to make that better with an with the idea that I know that this better experience will manifest itself in better commerce over time. I can't spend all my aim. I I can't spend all of my money on that.
Bryan Gildenberg [00:31:32]:
No company can. But I can and should be spending money right now on what I might call almost sort of pure AI r and d work. Right? Because no one knows where this is gonna end up. Right? So you're the most interesting use case most businesses will discover for AI is one they haven't envisioned yet. So so that keeping an open mind and really understanding where the technology is going, immersing yourself in it to the degree that you can follow the trends, I think is important as well.
Jordan Wilson [00:31:58]:
Alright. So, Bryan, we've covered a ton in today's conversation. Right?
Bryan Gildenberg [00:32:01]:
It's awesome.
Jordan Wilson [00:32:02]:
Yeah. Yeah. How how generative AI has, you know, impacted us from, you know, its its evolution and impact in in retail over the decades, you know, recent personalization, you know, search and content creation. We've tackled it from all over the place, but maybe as we wrap, what is your one biggest takeaway for whether it's it's for, you know, retail leaders out there in the space tuning in, the average consumer wondering how AI is going to impact, you know, their retail journey? What is that one biggest takeaway that you have for our audience?
Bryan Gildenberg [00:32:32]:
I really do think it's that if you think about AI as a technology that's gonna transform your organization, you're gonna be disappointed. I think if you think about AI as a specific way to solve specific problems in a way that you could not solve them before, then a way that's faster, cheaper, more personalized, and better, I think that's going to be a super powerful pathway to figuring out how to harness this technology. In the end, I really do think that a lot of the best principles people are gonna use to manage AI are going to be the principles that you've used to manage and grow people in teams. So good training, good governance, clear accountabilities, clear responsibilities, and a really well thought out plan for what it is it's supposed to do. That's probably more important to me than anything else. My, my good friend, Deb Weinswig, who runs Foresight Research, who probably knows who knows way more about it than I do, always just says something. She goes, look. AI isn't a tech problem.
Bryan Gildenberg [00:33:30]:
It's a business problem. And the more we can start to think about the business problems and then how the tech helps solve them rather than here's the tech, what business problem am I going to have? You know, so, you know, that that'll get us out of the whole phase that you go through with any emerging technology where it's a solution searching for a problem.
Jordan Wilson [00:33:46]:
Yeah. Love love that. So much, so much great, content and and insights, from you today, Bryan. So thank you so much for taking time out of your day to join the Everyday AI Show. We really appreciate it.
Bryan Gildenberg [00:33:59]:
Oh, thank thank you very thank you very much for having me on. I much appreciate it.
Jordan Wilson [00:34:03]:
Alright. And, hey, everyone. If this was helpful, please let us know. Share this with someone who needs to know. We love having industry veterans coming on and sharing their expertise. So if this was helpful, please go to your everydayai.com, sign up for the free daily newsletter. Hey, speaking of that artisanal content, me the human, I'm gonna go relisten to all of this great content that, Bryan just gave us and write a newsletter for you, recapping it all. So make sure you go check that out at your everydayai.com.
Jordan Wilson [00:34:32]:
Thank you for tuning in. We hope to see you back for more everydayai. Thanks y'all.