Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Roger Dunn - EP014
38 Plays12 days ago
Roger Dunn on AI, Agentic Commerce, and the Future of Retail Media
Retail media expert Roger Dunn joins hosts Nathan Guerra and Emmalee Crellin to discuss the evolving landscape of AI and retail. With more than two decades of experience, Dunn is a leader in the field and an IAB Retail Media Council member who authors the AI Commerce Brief on LinkedIn.
Four Futures for AI Shopping: Drawing on a BCG framework, Dunn outlines four scenarios:
- Open Bazaar — Consumer AI agents freely discover and buy across retailers.
- Super App / Data Fortress — Major tech platforms become all-in-one ecosystems for shopping, entertainment, and finance.
- Creator-Led Revival — Trusted creators and communities drive discovery.
- Brand Resurgence — Brands regain control through owned platforms and direct data relationships.
The "Alphabet Soup" of AI Protocols: Three protocols enabling AI to interact with commerce:
- MCP (Model Context Protocol) — Open-sourced by Anthropic, MCP standardises how AI models connect to external tools and data — a "universal USB" for product catalogues, inventory, and more.
- ACP (Agentic Commerce Protocol) — Associated with Stripe and OpenAI, focused on secure payments and checkout within AI environments.
- UCP (Universal Commerce Protocol) — Google-developed, connecting merchants to AI agents via infrastructure like Google Merchant Center.
Winning the "AI Shelf": Traditional SEO is like Battleship — you pick two or three keywords and hope you hit. Generative Engine Optimisation (GEO) is more like spear fishing. By building structured, machine-readable content — FAQs, specific use cases, detailed product information — brands can meet the 25-to-60-word conversational prompts consumers now use with AI tools. More detail and tighter parameters mean more precise matches, turning discovery from a guessing game into a homing missile.
Quote of the Week:
Dunn's adviceI: “Start with the building blocks. Learn the technical fundamentals, get hands-on and build something yourself, then think strategically about where AI can drive efficiency and innovation.” Without that foundation, the strategic thinking doesn't land — technical literacy and direct experience are the prerequisites, not optional extras.
Mentioned Links & Resources
- The AI Commerce Brief — Roger Dunn's newsletter on LinkedIn
- BCG Framework — The matrix discussing four potential scenarios for the future of AI shopping
- Roger Dunn on LinkedIn — Follow for his "Friday Memes" and retail media insights
- Bon App!: 100+ French classics, made easy by Isabelle Dunn (Amazon)
Mind the Model: Got questions? Email the show at mindthemodelpod@gmail.com
Recommended
Transcript
Nathan's Holiday & AI Tools Discussion
00:00:12
Speaker
Hi and welcome to Mind the Model. I'm Nathan Garrett and as always, I'm joined by my co-host Emily Krillin. Em, how's it going? Hello, hello, Nate. I'm well, how are you? Fresh back from holiday, I see a bit of a tan on your face. Indeed. I was in New Caledonia using lots of Google Translate and Google Lens to AI menus and things like that. So yeah, it was it was good fun.
00:00:33
Speaker
What have you been up to? Not much, actually. just ah I know i told our listeners a few episodes ago that I was using AI to understand contracts and building the pest reports. I've since bought a home and sold my home, which is very exciting. And i have been using AI for shopping purposes. so really researching all of the things that I need to get for new house. I'm moving in from a two bedroom unit to a three bedroom home with ah quite a large space to fill, understanding what bar stools, counter stools I should buy based on table heights. So yes, Claude and Gemini have been quite helpful helpful for me in
Introducing Roger Dunn
00:01:13
Speaker
that regard.
00:01:13
Speaker
I know we'd like to riff a little bit, but that feels like a perfect segue into our guest because yes and he is a master of all things commerce and shopping. He is really the master. i would like to describe him as more of the oracle of retail media for Australia. So our guest today is Roger Dunn. He is a retail media leader with over 20 years of experience, including senior roles at Diageo. Citrus Ad and Group M. He is a member of the IAB Retail Media Council and writes the AI Commerce Brief newsletter on LinkedIn.
00:01:44
Speaker
He is highly decorated in the industry. Raj has served as jury president for the Drums Commerce Media Awards, been honored as a RETHINK, all caps, RETAIL Top Expert for three years running, and was named Retail Media of the Year at both Remade and P2PI Awards. Roger Dunn, welcome to Mind the Model.
00:02:02
Speaker
Hey, Nathan, good to be here. um And yeah, looking forward to
Roger's Cooking & Ideas Sparring
00:02:05
Speaker
this chat. Well, Raj, we'd like to start things off with few quick fire questions. um So I've got one for you. Your wife recently published a French cookbook, if I'm correct. What's your favorite recipe to cook?
00:02:17
Speaker
For me to cook, well, I actually focus on um Sunday roasts. Okay. I do chili con carne and I make curries from scratch. so So no French food there at all? No, exactly. we We divide and conquer there. So I go for the stodgy stuff. and so And what's your favorite recipe to eat then from the cookbook?
00:02:36
Speaker
Well, in terms of French recipes, like a good French onion soup is always nice. Again, thank you. Back to some AI questions. What is your favorite AI model and what's the last thing that you used it for? My default model is Claude now. So that's the one I've been using for quite a while, actually. um i just find it's, yeah, it's the best one in terms of like some of that writing and like challenging your ideas. And then obviously that's a space that I i work quite a bit in. I think the last few questions have been um Probably about some of the topics that I delve into. So I've kind of built out a few different projects in Claude and like things around the AOC commerce brief, which is my newsletter and what topics always on the go there.
00:03:17
Speaker
um And then I guess other spaces in
Raj's AI Memes & Image Tools
00:03:21
Speaker
the industry as well. So thinking about like what's coming up ah in terms of the posts that we're going to do next around retail media. So yeah, it's just just a good sparring partner I find in terms of some of those really interesting topics. And then you mentioned your newsletter, Raj. You're also quite well known for your Friday memes.
00:03:38
Speaker
um Now, i don't know if you take submissions, but I've just shared one with you that I think um should be maybe in consideration. so um ah go ahead and open that email up and let me know what you think.
00:03:53
Speaker
Okay. Well, that's good. on Yeah, nice. No, that's good. And obviously on theme with the AI theme, and that is something I'm kind of focusing a bit more on as well, just seems to be the sort of topic of the moment. So there's lots of material, let's say, to work on. That one can be a guest post if you want. No pressure, feel free to- Just make sure to credit Mind the Model. Yeah, of course. We're always here for some meme ideas. And I see a little, is it a Gemini nano banana thing in the corner?
Roger's AI Methodology & Project Evolution
00:04:23
Speaker
So, yeah. Absolutely. Yeah. I'm still using Gemini for my image creation. I find it to be pretty, pretty good stuff.
00:04:30
Speaker
Raj, back to Claude. I know you said that you were using as more of a sparring partner. do you have a specific prompting methodology or how you like to kind of play in that realm? I don't have a methodology as such, but i i did used to have a folder of super prompts, which was like my my favorite prompts, but I've kind of evolved that now and built out the projects. So they kind of live within those projects now. So obviously there's a ton of like context and like basically just good stuff I'm reading or reports or good content that I feel is capturing what
00:05:03
Speaker
you know, what I think is important because you can't overdo it as well. Like a project it kind of has a limit in terms of like how much you want to put in there as well. And and then the instructions are now those super prompts. So it's like they're they're built out in terms of like different ways of analyzing data, different ways to look at the problem, thinking about contrarian views and like, what's the opposite side of this argument. So there's lots of ways that you can obviously build out that structure to have like all different angles on a particular topic.
00:05:30
Speaker
And then it's coming down to something practical. So is it, what's the end use case for that? So if it's going to be a newsletter article, like how do we, How do you want that output to look? And therefore, what different ways do you want to get to that that point and really sort of hone your thinking in them on the topic? And from a project perspective, I'm kind of interested. Do you have... So would you have a project around a very specific, tight topic and then you throw stuff into it? Or how do you structure things like that?
00:05:56
Speaker
I think it's more around really high-level topics. So like AI commerce or energetic shopping might be one where you've got a lot of that material in there. But for me, I wouldn't want to...
00:06:07
Speaker
necessarily mix up with say retail media bucket of information. Obviously there is some crossover there, which we're going to as a conversation, but um I think it's nice to keep it quite defined.
00:06:19
Speaker
um And then it becomes kind of a very um you know, vertical use case of of that type of content. So that's what I found, but you know, I'm open to ideas. I mean, the thing I've not used as much is skills in Claude. So if you guys have got any good tips on that, but I'd be open to to a few there as well.
00:06:40
Speaker
Not yet. um Definitely something I'm looking to build out into is building out the skills to, you know, more of a clean output every single time rather than having to reinvent the wheel. And then also need to, this is a reminder, I need to start dabbling in co-work again. and Nate and I have some grand plans for creating that little like mini AI assistant agent.
00:07:02
Speaker
think we want to nickname her May or Maybel. Nate, if that's right. So whether that's through um ah Claude Cowork, we've discussed perhaps, you know, going rogue and doing open claw. So yeah, should be some fun things to start playing around and we can share the learnings with our listeners.
00:07:21
Speaker
So Raj, you mentioned um already retail
Commerce Media Strategies & Retail Media Shift
00:07:24
Speaker
media. um You know, there's a ton of great stuff around. It's all at Curie Masters, just posted a ah whole thing on rebranding or thing from retail media to commerce media. Like what's your take on the shift? Is it something that's destined to happen? Is there a reason that when it hasn't happened? the way i mean The way I think about it, retail media is a subset of commerce media. So retail media obviously needs to have a retailer involved. So that's the retailer's own inventory on the app or website, in store, or using their data offsite to target.
00:07:56
Speaker
And I guess commerce media is more, I guess, other verticals coming to that party. So it's like travel companies or finance companies, you know, have equally valuable data. Perhaps it's slightly different angle. You know, if you think about someone like Visa or Mastercard, they have multiple merchants there. and So a really broad view on what a profile or a shopper is doing, but perhaps they don't go quite as deep and go down to a skew level like a retailer would. so It's very complimentary from a an advertiser perspective and it obviously depending on what vertical you're and like where that value is. um
00:08:31
Speaker
But yeah, there is this move towards commerce media as a sort of this broader t term and ah it makes for a broader industry as well and you know a bit more diversity in terms of what opportunities are out there. So I think it's all all positive. um If we think about retail media itself, I guess that is also expanding, I would say a bit more offsite and depending on the market in store as well. It's kind of traditionally in many markets been a digital thing where it's been sponsored products and banners.
00:09:02
Speaker
on site, but that off site piece where we take the shopper data and use it in different other media channels, um that obviously has you know almost unlimited potential in terms of wheret where that could turn up. So that's where things are being built out from a sort of pure retail media perspective.
00:09:19
Speaker
And is retail media big enough to, you know, your retail media the leader of the year, you know, should that is that still a big enough space or is it going to converse? You know, is it should you be the commerce media of the year next year instead? It's definitely a big space, but and it's only getting bigger. Right. So, you know, despite some areas maybe becoming more mature in the sense.
00:09:39
Speaker
perhaps the double digit growth maybe is is slowing down a little bit in some specific formats or areas. I think the growth is still there. And, you know, as an advertiser a marketer who doesn't want accountable, um really at attributable kind of marketing. So you want to target a particular shopper or someone who's interested in a particular category or or product.
00:10:02
Speaker
and then you can track all the up all the way through to a sale. I think retail media and commerce media both offer that. So very effective and you know efficient way of of marketing and and delivering sales.
AI's Impact on Retail Media Efficiency
00:10:12
Speaker
So that's that's kind of the perfect combination. And then if you then take that and put it into those offsite channels where it's potentially ah you know more of a branding or consideration environment, but you can still track it through to a sale, then that's almost bringing kind of the best of both worlds.
00:10:29
Speaker
And you've spent decades really shaping retail media strategy in Australia. how have you seen ai more recently change the way that brands approach retail media networks today? I guess it starts more as a feature on existing platforms. If you think about where it's showing up, it's kind of the ability to identify audiences. Maybe it's the ability to optimize platforms and media um to to an an outcome as opposed to kind of manually doing that or working to other metrics. um We're seeing it turn up in
00:11:01
Speaker
um the creative production side of things as well. So you might input all of your brand guidelines and specifics around what you want to see, but then you can make multiple versions of an ad and make you know really personalized ads.
00:11:15
Speaker
At scale. There's definitely those use cases, which I guess if you think about in AI, we talk about the efficiency side of it, where it's making existing processes more efficient. And then what does it unlock in terms of brand new spaces and brand new environments? And perhaps that's a bit less us you know explored today, but we're definitely seeing a lot of the efficiency side of AI kind of bringing you know smart ways to to optimize, to create um and to target and different shoppers. i constantly amazed at how deep you can go on every different platform and
Agentic Shopping Scenarios by BCG
00:11:47
Speaker
and consultancy, etc.
00:11:49
Speaker
i You recently i wrote something around, I think it was was it BCG who talked about um the Open Bazaar, Data Fortress, Super App, and the Embraced Creator revival. Was that BCG? I think.
00:12:03
Speaker
ah you know Can you break down maybe why it's important to do all four of those? Yeah, well, I guess it's not necessarily you as an advertiser or individual is doing one of those things, but i think it's just, it was quite an interesting way to split up the different ways Genetic Shopping might show up in the future. so I think the way they broke it down is,
00:12:23
Speaker
you know where is there a human involved versus an AI involved? So when it's agentic or AI first, it might be more like this, you know, the the dreamland would be this open ecosystem where your agent, you know, it with your interest in mind is able to roam freely and buy and find products across any of the different retailers.
00:12:45
Speaker
um yeah Kind of that that open internet dream in terms of like what we can find and how we can you know purchase, you know discover, find and buy. um I think the other world in that in terms of like a bit more of a concentrated market, market maybe a bit more what we've got now with big big tech is this sort of super app an environment where some of the big players, let's say like ChatGPT or Claude, become this all-in-one app where there's a lot of different services in their ecosystem. So shopping being one of them, but maybe entertainment as well, maybe finance and payments.
00:13:20
Speaker
and that becomes an environment or an ecosystem that you need to then navigate, but it's owned you know the rails are owned by someone else. um I guess there's other ones would be, you know, where it's a little bit more human first.
00:13:34
Speaker
It might be like a creator led sort of this authentic revival, they called it around a human led journey. So creators and communities shaping discovery and trust, you know, people are then falling back to this kind of human first and tastemaker type environment, which I thought was interesting. um And there's certainly an element of that, as you see, you know,
00:13:56
Speaker
the so-called AI slop come out there or different different ways that AI shows up, perhaps in a more negative way, people then rely more on the people they know and trust to to find products.
00:14:07
Speaker
and And then the final one was more of a brand resurgence. So brands really owning their own platforms, owning their data kind of fortresses as they described it and and being able to, you know consumers preferring different brand driven experiences. So and again, it's probably not gonna be any one of these four that really shows up as the yeah only way forward, but it's interesting to think how, you know where that might sit and like what opportunities would open up in the future. Yeah, the thing I appreciated your commentary on was like BCG explaining this is not a solution. This is literally, you know, potential ah scenarios that we may encompass. And it's almost like if you read the breakdown on your newsletter, each one of the four, you're like, oh, yeah, okay, I can see a world where that would make sense.
00:14:54
Speaker
But it only seems to make sense for like a few select companies. So the creator revival, like key for TikTok, TikTok shop, the super app, like we know OpenAI is is going towards that.
00:15:05
Speaker
But Yeah, it's it's interesting because I think it was very um humble for BCG to come forward and say, we actually don't know what the future holds, but these are just some potentials of what we could and you know see down the line.
00:15:19
Speaker
And the other important thing is that even if any of these kind of future scenarios do come true, there are some basics in there that you need to get right in terms of owning your own data and being ready and available, but you know, using MCPs to put your product information and have your products found in whichever of these futures and platforms becomes the dominant one. um So, yeah, there's definitely some sort of best practice, no regret type things that you can focus on um to to be successful and Yeah, things like the new disciplines around ah ao and GEO and the sort of commerce versions of that in terms of because obviously there's a few unique things with with shopping that in stocks and versions and colors of different clothing and that kind of thing, which are quite specific to to commerce and ri retail.
00:16:06
Speaker
Does AI commerce make brand more important or less important?
AI Commerce & Brand Importance Debate
00:16:10
Speaker
I'm very confident it makes it more important. like in in a few different ways as well, if you think about it, because I think strong brands and popular brands are in themselves going to have better reviews, better comments on Reddit, more surface area on different stores and different you know publishers and and sort of sites across the internet. So in terms of just providing the natural language that LLMs love to feed off and like create their models off, and they're going to be showing up a lot better and in a much more positive way anyway. That's a good problem. Not a strong brand necessarily, right? Not necessarily, but I think it's...
00:16:50
Speaker
the smarter they get, the more they'll go after real, genuine, authentic, good opinions. And yeah, it's not quite the same as branding necessarily, but I think it's it's part of that positive kind of input that they'll be looking for. But that, I mean, that aside, I think,
00:17:08
Speaker
Branding at this stage anyway, it seems like there's this short list economy where LLMs will get you down to it like a handful of brands quite quickly because that's that's the hard work that it does, right? It does the grunt work in terms of like casting it in the net across the all the different options and brings back this defined list that you've you've prompted or that you've set the rules around.
00:17:31
Speaker
And so when you get to that list, I think within that, you're going to be familiar with some of those products and perhaps less with others. And that will be your your shorthand as it is in any environment.
00:17:42
Speaker
know I don't know. See, my point of view is um I'm old school in the fact of like, I still think there is a marketing funnel, even if it's shrinking or changing shape, whatever. Like there still needs to be... brand awareness and the importance of branding to then go down through the funnel to performance and conversion lab activity. But then we start to think about more of an agentic world where we are no longer the decision makers of choosing what brand or what product to buy. And it's on behalf of our agent to just make that decision. And I think we're at the crux of this position where
00:18:19
Speaker
There's not huge uptake on it. It's very slow to kind of come through. We don't really have many learnings from other um areas across the globe, but there will come a point in time where, you know, is branding as important? Because or if our agent just thinks that product is the best for us, you know, just because I saw a really great bus ad or, you know, was at the movies and saw a pre-roll, like, does that even matter anymore? You're right. like it's It is changing the way certain categories are shop. And I think looking at it by category is in an important lens as well, because those really high-end luxury purchases, you can use AI a ton to find
Winning in AI-Driven Shopping Environments
00:18:57
Speaker
what's about it. And you're going be obsessive in like researching it And there's some products that take years for you to buy because they're so expensive. And AI will totally play a role, but you're not going to outsource the final decision of buying it.
00:19:09
Speaker
to an AI. Like it's it's just a tool and a means to an end. um And you could argue the ones where you would outsource it and like have a little bit more automation around it, was brand even that important in those categories anyway? Like do you Do you really care about the brand of pepper or salt that you, I mean, Master Foods might say they have- They ruffle the feathers of suppliers. Yeah. And that's ah that's a funny thing, and right? And I think you're absolutely right to focus on category because I think category makes a major difference here.
00:19:42
Speaker
You know, if you're buying Prada shoes, you're looking to buy that brand. Is your point if you're buying salt and pepper? yeah I mean, I'm very price driven when it comes to my spices on the aisle. um But speaking of the aisle, Rog, like how, how can brands win the shelf when the shelf in that category is just an AI's recommendation?
New Roles in Retail Landscape
00:20:07
Speaker
I think it's this discipline, obviously that it's an evolution and it's not the same as, but it's built on the foundations of SEO where you actually have
00:20:16
Speaker
data that can be machine readable and found by, in this case now, LLMs as opposed to Google or search engines. um So it's having that structured data being available and found, the schema, the markups. I guess where it becomes a little more LLM focused is when it becomes AEO and GEO, sort of generative engine optimization. and that it's about context. So it's like having those FAQs, it's having context. It's like, what does this product go well with? um And what does this what's the use case for this product? Like, where would I use it? And you're just translating it back into those questions that people are putting in as prompts in the first place. It's like, how can I have a very specific answer to this?
00:20:57
Speaker
And the way i I guess I think about it is, if you think about SEO, it's often been a game of like, going after two or three key keywords. It's like battleships. You've got two coordinates and then you're trying to hit this in the middle and you might miss a few times. And it becomes very concentrated around those few keywords. It's really competitive, right? Trying to get to the top of the page. Yep.
00:21:19
Speaker
But on the LLM side, you've got these prompts that are like at least 25 words, maybe 60 words, maybe a multiple prompt conversation. And by the stage, you've got all of that context.
00:21:30
Speaker
You're like spear phishing or using a homing missile to hit something. you're actually being very specific, you've defined all the different parameters, you might have given a price, and then you're going in and finding this specific product. And if you can be that product, then that's great. great And it's it's just opening up a lot of opportunity, I guess,
00:21:49
Speaker
at the long tail, but not necessarily just for small brands, it could be big brands as well, as long as you're the answer for that that particular um you know definition. Mondelez, just highlighted this as well, they just hired a a global lead of emerging commerce platforms.
00:22:06
Speaker
um and What does that person do dayto day to day? Are they worried about SEO, SEM and GEO? mean, is that what they're focused on? like Well, you have to ask them. I don't know. But yeah, I did i did see that job opening and they're looking for it And I just find that interesting in itself that it's become an um you know important role in and of itself. And I think that's the right way to go because there is a lot to know here in terms of, I think about it, like there's a lot of tech, some technical knowledge that you're going to have to read up on, understand, like if it's the protocols and that kind of thing.
00:22:37
Speaker
Then you've got kind of like the builder mentality of like, how do we put this together and like which bits need to connect with which. Then you've got to be somewhat strategic and think about, okay, well, where is this industry going? Where's where's shopping going to be in the next few years? And like, what tools are people using?
00:22:53
Speaker
um And then you've got to plug it all into like what your brand can offer and like what products you you have and where you can address some of those needs and and remain kind of being being found. So yeah, it's like that, that in in its you can see there's like there's an end to end thing where actually someone having a think, a thought and like working through all of that um would be quite important in terms of pulling it all together. And you're obviously going to plug in with existing departments. you're going to speak to the commerce team. You're going to need to know marketing. You're going to need to look at the product and digital shelf and all of that SEO work that's probably already happening. But I think having that role to coordinate things is definitely a an interesting space to be. And you um you mentioned something pretty interesting in that um answer before about protocols. And Nate and I have coined it the alphabet soup, MCP, UCP, ACP.
Significance of MCP, UCP, and ACP Protocols
00:23:43
Speaker
What does it mean? If you can just explain it to our listeners at like a really clear level, um what are the differences? Why does it matter really? Who needs to care? Yeah. So the protocols obviously are part of the technical makeup now in this space. So yeah,
00:23:59
Speaker
I think that if you think about where LLMs find their information, they can either go directly to a site and plug directly in or they're doing a web search, which is somewhat inefficient for them in terms of like trying to find the information you know that's out there.
00:24:12
Speaker
And if you think about the most basic foundational layer is MCP. So it's like model contact protocol, which is people have described it as this universal USB that allows one LLM to plug into you know another platform. And it's just about unlocking I guess in the commerce space, all the attributes from your product catalog, it's like checking inventory, it's retrieving prices, understanding product attributes. So all of those things become much more easily available to an LLM, whether that LLM is an external one like ChatGPT or you know the retailer's own LLM on their site.
00:24:48
Speaker
and But that's the that's the way to plug in, I guess, in terms of finding the the the information. And then the other ones are more, they're sort of third, when that one was founded by Anthropic, so originally, but is now open source. guess the other ones are the ones that have been developed by some of the big players in this space. So we've got ACP, which is a genetic commerce protocol, which was co-developed by Stripe and OpenAI. It's much more focused on the transaction. So it's enabling that secure payment process, whether it's like, you know, potentially the confirmation, the refunds, the post purchase flows. More relevant to instant checkout and like things that are happening within
00:25:29
Speaker
That's right, yeah. So it was originally launched with Instant Checkout, and which was that click to buy within the LLM. Obviously, they've changed that approach a little bit, and now it probably lives more in the ChatGPT apps in their world, and but still does the same same role in terms of a transaction.
00:25:46
Speaker
i think the comparison here is with the UCP, which is the Universal Commerce Protocol that Google developed, and they launched at NRF back in January. And this one's a bit more broad reaching. And obviously they're coming out from a different perspective in the sense they've already got things like Merchant Center, which has a lot of product information already in there.
00:26:05
Speaker
um And they've enhanced that and have different elements coming in there. But it's about you know being that one to many kind of connection between merchants to sort of many different agents.
00:26:17
Speaker
And it's building out yeah the the availability of your products in all of those different platforms. Right. So what I'm hearing from you is if a brand wants to, specifically a retailer or a brand, wants to ensure visibility of their products and schematics and ah more of the details that end users of LLMs would be searching for, it's important for them to consider MCP, UCP, or ACP to connect directly to that technology. Is that what you're saying? Yeah.
00:26:43
Speaker
Well, MCP is much more broad in terms of that's the almost the the the method of unlocking it. And then the others are much ah more, I guess, ecosystem specific in the sense they'll open up different areas for you. So again, it's probably not a ah one or the other conversation. It's probably a both. And yeah, depending on kind of where you you weigh where it makes sense for your products to be available. Yeah, I'm i'm anticipating most parties will...
00:27:08
Speaker
Support both ACP and UCP. At this point in time, it feels like you should probably should be experimenting in both and and seeing where where the opportunities are. Retailers of significant size are going to want to play with both Google and Microsoft, ChatGPT, Claude, and providing supporting both is just is the way to do that.
00:27:29
Speaker
Yes, I think so for for for today. yeah And there may be more coming. like what what We haven't seen exactly what Meta's play is in this space, and which will be huge, no doubt. Claude, again, hasn't really focused on shopping and commerce just yet. Raj, what have we not asked you that we should have?
Retailers Adopting LLM Experiences
00:27:49
Speaker
I don't know.
00:27:52
Speaker
We haven't talked about like LLMs, but you know, retail media and what retailers should be doing with their own sites and like the opportunities. Give us your pitch for what retailers should be doing with LLMs.
00:28:06
Speaker
From what where I see it the... people are being trained to use LLMs one way or another. So we're all using ChatGPT or Claude and you know more people every day. It's like this huge directory of new customers. So there's this change in behavior of people looking for products using natural language and longer prompts. So I think as people get to retailer sites,
00:28:28
Speaker
they're going to really expect that experience. Like I think it will get quite old quite quickly to have a huge drop down menu with like a load of taxonomy to kind of flow through and find a specific category or, you know, even like a search bar, if you're just putting in keywords and getting like,
00:28:44
Speaker
relevant results, but just nowhere near as relevant as a sort of natural language prompt. So I think there isn't a world where retailers shouldn't be having LLM type experience on their own sites.
00:28:57
Speaker
And then where does that go next? to that That becomes quite a valuable part of the site. And if we think about things like retail media, then there probably are monetization opportunities there, as long as it's helpful, as long as it's native and it's going to and provide, you know, outcomes both for the shopper but also in the advertiser. I think that'll become a really valuable space for retail media. So I'm excited to see what happens there and like where the where the opportunities kind of build. It sounds like the the speed and the pace at which AI is really booming
00:29:30
Speaker
retailers and anyone essentially who has a website should really take stock of their current user experience and start to adapt for that change in consumer behavior to ensure they're getting the most out of the website.
00:29:41
Speaker
And even if you think about publishers as well, you you go to the front page and you see all the headlines and the stories, but they've got such depth of information and knowledge on their sites from all the historical stories they've got. Why not make that surfaceable and more much more available? And equally for shop for shopping as well and marketplaces, the whole problem is But the whole beauty of them is you've got this huge assortment, loads of selection, loads of choice.
00:30:06
Speaker
It's about surfacing it. And this is a tool to surface the right products. is And I guess, how do you think about... that being on site versus LLM. And in part, like I know Microsoft has brand agents where you can set up a basically, you know, I wanna talk to Nike, I wanna talk to Woolworths, and you can engage with them within the context of an LLM.
00:30:30
Speaker
who um i I guess, are you looking at a world where both of these things are enabled and they sit side by side and with the same kind of underlying data set underneath it? Is that the idea?
00:30:41
Speaker
In order to surface and probably be become available and native into the LLMs, you probably need to do the work yourself anyway. And if you've done the work in terms of getting your data ship shape, creating an LLM type experience to feed into ChatGPT, you may as well have that same experience on your own site.
00:30:58
Speaker
And i think that the I think the connection between those two is perhaps what was missing with things like that original instant checkout. is I think Walmart have said, you know if you put something in your basket here, it doesn't show up on your basket when you get to the Walmart site. And you actually want that to be connected up. And that's why the yeah YAP approach now is a little bit more...
00:31:19
Speaker
more seamless And that's ah just a very specific one example of how the space is evolving. And like, you'll see these developments where it it gets, gets smoother, it gets better and it gets a lot more kind of native. And the interesting part is then when it starts,
00:31:36
Speaker
knowing you a little better and being a little bit more proactive in terms of what it can offer you in terms of shopping. Well, that is a great place, I think, for us
Balancing AI Benefits & Threats
00:31:44
Speaker
to start. always like to end at the future. we do like to wrap things up with a couple of quickfire questions.
00:31:51
Speaker
Raj, do you consider yourself to be an ai optimist, realist, or pessimist? I think I have to say realist in the middle there. Like, it's... it's Obviously, there are... There's so many...
00:32:03
Speaker
huge potential for it in many areas and the ability to do so many things and like the, the you know, the functionality it's brought to what what I do in my day-to-day life is is great, but there are those bigger exist existential kind of threats that we're all we're all thinking about as well. So I think you have to be a realist and i think that's healthy to have, you know, the both sides of that be, you know, be part of your consideration set, the yin and yang.
00:32:29
Speaker
What is your top tip for marketers who are listening who want to use AI better in their jobs? We were saying earlier, it's it's it's getting all the different building blocks. So it's knowing a little bit about the technical side of things, doing some of the building yourself and understanding how to put these building blocks together and like what works and what doesn't.
00:32:47
Speaker
And that will change over time as well. And then it's thinking about it strategically and thinking, okay, well, where where can I actually apply this and where does it work? you know, create an efficiency game in terms of like making things a bit quicker, but also where's an innovation game in terms of like, what can i now do that I could never do before? And I think you need the first two things in order to get there.
00:33:08
Speaker
So yeah, I think it is a, you know, so there's a lot to learn, but equally, if you can just boil it down in your space and and that's what I try and do with with my thinking is I'll look at it from a retail and there was sort media and then sort of AI commerce perspective. and that's all I expect to kind of really be the the the expert on.
00:33:28
Speaker
And Raj, answer this one, fill in the blank. Agentic commerce will be blank in three years time. Agentic commerce will be, that's good question.
00:33:45
Speaker
We'll be a lot further ahead in three years time. Safe bet. Excellent. Roger Dunn, thank you so much for joining us on Mind the Model. Thanks, Raj.
00:33:57
Speaker
Good to be here.
00:34:01
Speaker
That was fantastic. Nate, what did you think of our chat with Roger Dunn? What a i mean you know it's um i feel like we've been very lucky to have like serious retail media experts on our podcast, you know between Kiri, between Raj and the Commerce Media guys. um yeah like serious heavyweight people. And I thought Raj just ah was so clear and so considered about things like where to spend your time and energy if you're in this. Yeah. And it's funny because listening to hear him talk, like I know there's a lot of chatter in about the world of AI and the efficiencies and automation and, you know, coding and software engineering and all that jazz.
00:34:45
Speaker
Our niche, ah like Raj was saying, just play to your niche and you know, upskill on what you know. Our niche is AI and marketing on Mind the Model. And it seems like a ah really... um clear intersection of those two points is retail media and commerce media. So it may seem a bit silly to others listening of why we are going deep on this specific topic, but I think there's just so much value to be unlocked, um especially with agentic commerce coming up, ah you know, chat GPT doing instant checkout one day and the other day they bring it back.
00:35:18
Speaker
um But yeah, I think Raj is a ah wealth of knowledge. I really appreciated his point on discoverability and how you know retailers should use that shift in consumer behavior to you know adapt their catalog or change their website and no make things a bit more fit for purpose for how people are currently buying.
00:35:42
Speaker
Well, yes, but I also thought he was quite considered in terms of like you know talking about that this is still only a fraction of a fraction of your total media, a total retail spend.
00:35:52
Speaker
And you know while it's really important to us, while we talk about it all the time, it is you know something that you know it's always worth paying attention to the future, but like focus on the now and what needs to be done and and know that this is coming, but it isn't going to change your world today.
00:36:08
Speaker
think that's a good point. Like for coding and software engineering, they're having a completely different trajectory and kind of timeline of how their worlds are changing. ah For us, you know, people but will so still go in store to buy items, um you know, still sometimes call in and place orders over the phone. Um, so it's definitely a slower moving beast in the retail and commerce space. Um, but yeah, great to have Dodge on, you know, we've been trying to get him on for months now. So, um, it was always great to hear his expertise and we'll go ahead and add to the show notes, the, um, The BCG matrix that we were referencing will link to Roger's incredible newsletter. Definitely a must read for anyone in the AI and retail media and marketing space. um
00:36:57
Speaker
But yeah, thank you all to our listeners. ah Be sure to like, share and subscribe. Any questions you have or future guests you want to see, go ahead and email us at mindthemodelpod at gmail.com. And remember, the intelligence might be artificial, but the wins are real.
00:37:12
Speaker
See next time.
00:37:18
Speaker
This episode was brought to you by Testchimp. Testchimp is a web developer's best friend. Testchimp uses deep runtime context to automate your testing workflows, generate scenarios, write scripts, and fix bugs all in one place.
00:37:33
Speaker
Stop wasting time and start shipping quality code faster. Bugs suck, but finding bugs doesn't have to with Testchimp.