How To Become Agentic Commerce Ready
On our 50th episode of AI moment podcast, we had Geoff Gibbins join us to discuss agentic commerce and how he has built a dedicated tool, Reconnix and building for agentic offerings for companies, including Coca Cola.
Short on time, skip to key sections:
00:00 Introduction to the AI Moment Podcast
00:06 Meet Geoff Gibbins: Author of 'When AI Shops'
00:41 Defining Agentic Commerce
01:30 The Shift Towards AI in Commerce
02:57 Real-World Examples of Agentic AI
04:49 Marketing in the Age of AI
07:36 Understanding Agent Psychology
11:30 Building AI-Friendly Business Models
24:13 Future of AI and Robotics in Commerce
29:13 Geoff's Advice for Businesses
36:34 Closing Thoughts and Takeaways
The Key 6 Points Discussed
Defining Agentic Commerce: Geoff defines agentic as any commercial transaction where AI is involved in the research, decision-making, or execution phases for either the buyer or the seller.
AI as a New Marketing Channel: AEO is not SEO; we are entering the world of AEO (Agent Engine Optimisation) is a whole new channel to SEO. Marketing to agents is a “greenfield” opportunity that brands like Coca-Cola are already exploring.
The “Agent-ology” Shift: Agents don’t share our “FOMO.” While a human sees “only 3 left” and buys, an AI sees “only 3 left” and avoids the recommendation for fear of transaction failure.
Positional Prejudice: It was a “gasp” moment in the interview when we discussed how different LLMs have “spatial” biases—ChatGPT leans towards products on the left, Gemini to the right, and Claude to the centre.
The Trust Deficit and Accountability: Who is liable when an AI makes a bad purchase? Currently, it’s a “hot potato” issue with no clear legal framework, making trust the ultimate currency.
2030 and Robotics: By 2030, we won’t say “agentic commerce”—it will just be “shopping.” We’ll likely see personal shopping robots in-store that sync with our wearable health data to make real-time suggestions.
Free Resources
Geoff’s LLM assessment tool - Reconnix (get a score and recommended actions of how ready you are for agentic commerce)
Connect with Geoff
On LinkedIn or on Human Machines his business site
Check out his latest book When AI Shops
Subscribe to the AI Moment pod
If podcasts aren’t your thing, here is the full transcript
AI Moment Interview With Geoff Gibbins From Human Machines
[00:00:00] Jonathan Wagstaffe: You are listening to the AI moment with me, Jonathan Wagstaffe.
[00:00:03] Danny Denhard: And me Danny Denhard.
[00:00:06] Jonathan Wagstaffe: Today we are delighted to be joined by Geoff Gibbins, who is the author of a recently published book called When AI Shops, and Geoff's focus is very much the human AI commerce interface.
[00:00:20] Jonathan Wagstaffe: So he is really focused on how we're going to use AI to buy and sell things as we go through the coming years.
[00:00:27] Jonathan Wagstaffe: Geoff, thank you so much for joining us on the podcast. Where do we find you today?
[00:00:32] Geoff Gibbins: Well, thanks. Great to be here. Uh, I'm in New York today. It's actually a very cold, sunny day in New York.
[00:00:37] Jonathan Wagstaffe: It so cold and sunny. Sounds good to me. Sounds better than we've got in the UK at the moment.
[00:00:41] Defining Agentic Commerce
[00:00:41] Jonathan Wagstaffe: Let's start with agen commerce. Uh, define agentic commerce for the audience. Just a short definition so we know where we are with it.
[00:00:49] Geoff Gibbins: Yeah, so I think agent commerce, I would define it simply as the situation when there's buying and selling, taking place between, one party buying or another selling [00:01:00] at least.
[00:01:01] Geoff Gibbins: AI is involved somehow in that process. So that could be that it's helping to research for the buyer or the seller. It could be that it's actually making decisions for the buyer or the seller, or it could even be that it's actually executing transactions on behalf of the buyer or the seller. So it's, it's really about AI playing different levels of involvement.
[00:01:21] Geoff Gibbins: So agentic commerce is. Still emerging and it's lots of different things that it could be in the future.
[00:01:28] Jonathan Wagstaffe: And you described that well in the book.
[00:01:30] The Emergence of AI in Marketing
[00:01:30] Jonathan Wagstaffe: What, what, when was the moment when you realized that the shift was inevitable?
[00:01:35] Geoff Gibbins: I think for me, you know, about three odd years ago, we all started talking about chatgpt and it was amazing the task that it could complete and the automation that it could do and the potential of that.
[00:01:48] Geoff Gibbins: I think it became clear to me, probably about a year and a half ago I what I, I actually remember reading a, an article in the New York Times that was very early on in this journey of understanding [00:02:00] how do you change what an LLM does and says, and they had an article where they talked about how some companies were starting to figure out, well, how could you get recommended more?
[00:02:10] Geoff Gibbins: As a result of what you are publishing and what you're doing.
[00:02:14] Geoff Gibbins: And it wasn't always, it wasn't just talking about shopping, it was talking about, you know, all sorts of ways you can, you know, burn your personal brand and get AI to talk about you in a different way. But it was, it really opened my eyes to the idea that. AI was gonna be a new marketing channel. And then in the, in the next month or two after that, I started talking with clients at the Coca-Cola company at Suntory, and it became immediately clear that this is an entirely new marketing channel of marketing to agents, which was just a whole greenfield opportunity and a challenge for them.
[00:02:45] Jonathan Wagstaffe: Yeah. And I think one of the things that struck me listening to you speak and, and also seeing the book is. I think this stuff is further down the road than a lot of people realize there, there are literally millions and billions of dollars being done agenticly already.
[00:02:57] Practical Examples of Agentic Commerce
[00:03:09] Jonathan Wagstaffe: And I think lots of people are the mindset that, yeah, we know [00:03:00] this might be coming, but we didn't realize how far it's already, it's already got, if you like, c can you just give an example of a, of, you know, how that would work in practice, how an agentic ai commercial process would work in practice?
[00:03:11] Geoff Gibbins: Yeah, sure. So I think as you, as you say, it's already happening in practice and so what that could look like is that. Already there are, there's a big shift in terms of people not using a typical search engine box to, or search results to find and research answers to a question.
[00:03:29] Geoff Gibbins: They are either using ChatGPT, or Claude to get it, or they're actually going into Google and they're getting like an AI generated response from that, and as a result, they get told, here's, an energy drink you might be interested in that meets your needs, or here's the type of retirement plan that you might be looking for.
[00:03:48] Geoff Gibbins: And the AI is actually making that decision easier for them. It's helping to navigate based on what it knows about them and their history and their previous conversations, what's actually right for them.[00:04:00]
[00:04:00] Geoff Gibbins: It's engaging in a dialogue with them about it and saying, Hey Geoff, let me figure out which is the right retirement planning option for you based on five different questions I'm about to ask you.
[00:04:10] Geoff Gibbins: And so that's already happening. Yeah. Today that's happening. And a lot of companies are seeing a big uptick in the proportion of their sales that are coming from ai. Uh, and some companies are just kind of finding this happens without any particular effort on their part. And sometimes they're finding, sometimes companies are finding that they're really struggling and they're left behind and they're, they're trying to figure out where do I stack up in this new world?
[00:04:35] Geoff Gibbins: And, and what do I do?
[00:04:38] Jonathan Wagstaffe: Yeah. And I think there is a lot of, um, there's a lot of confusion out there at the moment. I think there's lots of people there's, there are, there are people selling snake oil, and there is, and there's lots of people trying to figure out how to get around this whole process.
[00:04:49] Danny Denhard: When we talked, uh, originally we were talking around, you know, the forms of ag agentic commerce and, and how we're going to basically going on a journey or with AG agentic, do [00:05:00] you, with a lot of the work that you are doing and some of the, the work, um, and projects that you're undertaken with, with these brands, are people asking the same sort of questions or is it actually people are at far different ends of the scale because. My experience is some people are right at the start and other people are away in the future, and actually ahead of where we're at. Can you give us a, a sort of a sense check and give us a guide please, Geoff?
[00:05:25] Geoff Gibbins: Yeah, sure.
[00:05:26] Marketing Strategies for AI and Humans
[00:05:26] Geoff Gibbins: So I think, um, so I think there's a lot of different reactions as you say, and part of it depends about how far ahead you are, but also what kind of role you play within agent commerce, right?
[00:05:35] Geoff Gibbins: So, so you have a lot of people who are marketers and they are looking at this as, how do I market in this world? Right? And so they might be, you know, they might be a company, I've worked with companies where they started thinking about this a year and a half ago, and they are probably, they were further ahead of the curve.
[00:05:55] Geoff Gibbins: But now they're trying to think about, well, how do I actually balance marketing to AI and [00:06:00] humans at the same time? So there's new challenges that come up you.
[00:06:03] Geoff Gibbins: I think a lot of companies that I speak to today when it comes to marketing, they have a general sense that there's this new thing, and it's called AEO or AIO or whatever it might be.
[00:06:15] Geoff Gibbins: Something like that, that they have to learn about and that it's different to SEO, but it's kind of like SEO. So they have a sense, they have a reference point for that. But they don't necessarily understand what to do.
[00:06:25] Geoff Gibbins: The other types of companies though, are companies that actually play a pivotal role in the mediation of that transaction, right?
[00:06:33] Geoff Gibbins: So retailers, they are now looking at not just how do they market to ai, but how do they actually, uh. Rebuild their experiences in order to be AI shoppable, and how do they actually play a role within the purchase journey.
[00:06:47] Geoff Gibbins: You also have companies that are working in payments and they're looking at how do they actually enable trust within these new types of payments.
[00:06:55] Geoff Gibbins: What's the way that you actually, change your business model around how you manage [00:07:00] transactions? So there's lots of different entry points depending on who you are.
[00:07:05] Jonathan Wagstaffe: Yeah, I saw you speak last week, I think it was Geoff, and one of the things that came out really clearly, there was a, that a EO is not necessarily just like SEO v2, there's a very different way, but you go about that whole piece.
[00:07:17] Jonathan Wagstaffe: But B, as as you talked about it really, marketers, if they're clever, they've got to present their information for AI and for humans. So you're almost talking to two discreet audiences now. The clever marketers will talk to those two discreet audiences in the way they set their website up, the way they set their other digital assets up online, et cetera, et cetera.
[00:07:35] Geoff Gibbins: Absolutely. So there's this kind of like dual psychology challenge that we have to solve for.
[00:07:40] Understanding AI Biases and Agent Psychology
[00:07:41] Geoff Gibbins: Uh, so if you think about it. Marketers have been trying to understand consumer psychology for like over a century where there's still a lot, we don't know, frankly, like, but it, you know, it's pretty good as an understanding, but there's this whole new realm of agent psychology, or "agentology" as I call it in the book, that you have to start to understand.
[00:07:58] Geoff Gibbins: And the thing that's [00:08:00] really interesting about that is that agents are not actually. Completely rational. They have biases, uh, partly because they're actually trained on human language, right? So they take on some of the biases of humans, but they're also strange and weird quirks and biases in how they analyze information that actually are unique to ai.
[00:08:19] Geoff Gibbins: So for example, if I was to look at a website and it says there's only three of a product left As a human, I think, wow, I've gotta buy it. Now I have that fomo.
[00:08:29] Geoff Gibbins: Um, an AI agent looks at it differently. They actually are optimized to make a recommendation that sticks and that can be delivered on.
[00:08:36] Geoff Gibbins: So they look at that, not with fomo. They look at that with like a fear of making failure, right? They might not be able to follow through on that recommendation, so they actually dramatically underweight that kind of choice.
[00:08:50] Geoff Gibbins: And so the same piece of language is looked at in different ways by humans and agents.
[00:08:54] Geoff Gibbins: And so some of what I've been starting to do is helping companies figure out how do you [00:09:00] become, human readable and likable, and how do you become machine readable and likable, and how do you do those two things together?
[00:09:08] Jonathan Wagstaffe: Yeah, and I think one of the things, one of the facts that, um, you talked about last week that I've since repeated and is along in one case, caused somebody to gasp was the, I can't remember the detail now, but it was wherever the, the product is on the page, depending on the browser, depending on the ai, LLM, certain LMS prefer it to be on the left hand side of the page.
[00:09:28] Jonathan Wagstaffe: So on the right. Tell us a little bit more about that.
[00:09:30] Geoff Gibbins: Yeah, sure. So, so it's really interesting. Over the last six months there have been a range of different academic studies conducted on how AI agents make purchasing decisions and how weird they are. And so the one you just mentioned is, is called positional prejudice.
[00:09:46] Geoff Gibbins: So this is actually, um, when reading a product page. ChatGBT actually consistently leans towards products that are on the left of the page, uh, Gemini, the right, and then claude, uh, the center. And so you have [00:10:00] to think about, as a marketer, how is my content gonna be presented to these different LLMs? And then the other thing that's interesting is that as these model updates happen, the it changes too, right?
[00:10:12] Geoff Gibbins: So you don't just have to solve for different LLMs, you have to solve for. You know, up upgrades from 5.1 to 5.2 to 5.5 to whatever it might be.
[00:10:22] Geoff Gibbins: And so there's this entirely new discipline of monitoring how machine likable you are, which is emerging, that companies are gonna have to understand that as a first point and then actually have an ongoing monitoring capability to understand what's the latest changes in how models look at you in the policies of chatGBT.
[00:10:46] Geoff Gibbins: Gemini or whatever it might be. And so that's one of the things I've been working on, is actually building a, a tool and an infrastructure for monitoring that with, for companies, uh, because it's gonna be hard to keep up with all of these things. Like so many people [00:11:00] feel left behind already and it's like it keeps changing every month anyway.
[00:11:05] Geoff Gibbins: And so that's gonna be a real challenge for, uh, a lot of companies.
[00:11:09] Jonathan Wagstaffe: Yeah, people are really interested in, you know, what is it that makes the agent confident to make the recommendation or make the purchase? And, is it price? Is it reliability? Is it structured data? Is it social proof?
[00:11:18] Jonathan Wagstaffe: Is it, I think people are scrambling for that information. You know, which of those things are the ones that really move the needle in terms of AI recommendation?
[00:11:25] Jonathan Wagstaffe: You've got any thoughts on, I mean, just talk about the fact it's changing a lot. Any thoughts on. Which are the headways.
[00:11:30] Geoff Gibbins: Totally. Yeah. So, so this is what I've been working on over the last few months, a lot.
[00:11:33] Geoff Gibbins: So it is actually building a, a data-driven model for how you actually analyze this and analyze an individual business and identify its recommend ability effectively.
[00:11:43] Geoff Gibbins: So, so the first piece is discoverability. So this is, can AI actually find all of your content? So that's, you know, that gets to the point around things like, do you actually have structured data?
[00:11:56] Geoff Gibbins: Do you have server side rendering so you can actually [00:12:00] have all of your content readable, not in JavaScript, for example.
[00:12:03] Geoff Gibbins: Are you accessible to AI crawlers? There's like a whole realm of that and, and that that's sort of about, you know, about a third of what matters, I would say. Yeah.
[00:12:13] Geoff Gibbins: Uh, the next section in the next sort of piece is external authority.
[00:12:17] Geoff Gibbins: Does AI actually trust you based on what the rest of the internet says about you? So is there a Wikipedia page on you? Is there a crunch base page on you? What do people say about you on Reddit?
[00:12:28] Geoff Gibbins: Do you have a lot of earned media that's positive about you? Uh, do you have a lot of third party reviews? Like all of those things? And are they positive too? All of that is, is important too. So it's not just about are you visible, but what's out there that is, is visible,
[00:12:43] Jonathan Wagstaffe: right?
[00:12:43] Geoff Gibbins: Uh, and then the last two pieces, the. The third one is around. And agent psychology. So do they act? Do they actually like your content?
[00:12:52] Geoff Gibbins: Do AI agents feel like you have really clear attributes that it can pick out?
[00:12:57] Geoff Gibbins: Do you have that social proof on your [00:13:00] site?
[00:13:00] Geoff Gibbins: Do you have all of the kind of trust signals? Um, that are clear around how others trust you. You're making that clear and super readable.
[00:13:07] Geoff Gibbins: And then the last piece is, can AI even interact and buy from you and do you have commerce readiness?
[00:13:13] Geoff Gibbins: So for companies that are looking to actually sell products or services on their website.
[00:13:19] Geoff Gibbins: Are they set up with the latest feeds for chatGBT for Google's UCP? Do they have payment integration? Do they have product identifiers? All of these things. And so you have to keep up with all of these changes because the model changes all the time.
[00:13:32] Geoff Gibbins: And so that's what I'm building is the, is the monitoring infrastructure to understand where are you and what do you need to do next to stay ahead, uh, and even get ahead.
[00:13:42] Danny Denhard: It sounds to me like, and obviously we're, we're all in this, we're all sort of "AI pilled", but it sounds to me if I was an outsider who was confident that AI is gonna be the future, but it seems very expensive and a huge investment to make as a business, whether you're small or large. [00:14:00] So, you know, like ChatGPT saying, they're gonna charge 4% fees on (agentic) transactions.
[00:14:06] Danny Denhard: To some people that's, that's a game breaker that, that that's, that's a showstopper them, it just means that they won't be able to do it. Do you have a sense, or do you give a guide to people on what approach that you'd you suggest they take? Or is it a case of. Case by case and very much a week, you, if you can afford it, do it.
[00:14:24] Danny Denhard: If you are gonna make a long-term decision, this has to be part of your, your long-term plan, do you have a, a leaning in or, or a strong recommendation for people?
[00:14:35] Geoff Gibbins: Yeah, that's a great question. So I'd say, I'd say the first thing is really understand where you're at today. And so, so the, the tool I've been building recon, uh, recon ai, I've, you know, put the link in the description.
[00:14:50] Geoff Gibbins: So that is really designed to help you figure out in five or 10 minutes, where do I stand today? Right? What's actually my current state of play? Am I doing well? Am I not? How do I compare to my competitors? [00:15:00] But then you raise a really good point. The question next is, actually, what do I want to do? So as you, as you've seen over the last few months.
[00:15:08] Geoff Gibbins: I think a year ago everyone was saying, well, I have to be visible. I have to have the a EO. Right? And, and that's, that might be the answer for some companies, but what we're also seeing is a lot of companies are now starting to block ai. From accessing their website.
[00:15:21] Geoff Gibbins: So Etsy for example, they have a privileged partnership with ChatGBT.
[00:15:26] Geoff Gibbins: And so, uh, one of the things I found actually in building the reconnix.ai tool is that they block bots from accessing their website, right?
[00:15:33] Geoff Gibbins: They don't want to be access, they don't want AI bots accessing their website all the time because it's actually quite expensive, and they want to control what's the message that is sent to chatGBT and to Gemini.
[00:15:46] Geoff Gibbins: So that's why like the AEO tools are great if you, if you want that, but you have to decide if that's what you want. Um, and so it's a question of like, what's your strategy overall and that, and that might change over time for companies too. [00:16:00]
[00:16:01] Jonathan Wagstaffe: So what would you do if you are A A CMO or A CEO? Geoff, what are the first three moves in the next 90 days?
[00:16:06] Jonathan Wagstaffe: Then, I think you've already started to say one of 'em is to understand where you are using the tool, but what, what are the three things you wanna do in the next 90 days?
[00:16:13] Geoff Gibbins: Yeah, so I think the first thing would be, yeah, understand sort of what's your current state assessment of, of where you are, what, how, how well set up are you across those four different factors.
[00:16:23] Geoff Gibbins: I think the second piece would be to understand. How are your customers actually interacting with ai? Because that looks very different for different businesses.
[00:16:32] Geoff Gibbins: So what types of tools they're using for what specific types of purposes and how does this shift the mental model of how they actually need to interact with your business and what your experiences need to be?
[00:16:44] Geoff Gibbins: And then the third thing is to use that information to figure out not just what kind of infrastructure changes I need to make. But actually, what does that mean in terms of what we should be selling differently? Right. So part of what I think [00:17:00] I've discovered with working with companies is not about how you market, it's also about what's the type of thing that you sell?
[00:17:06] Geoff Gibbins: Do you need new ways for people to actually engage with your products?
[00:17:09] Geoff Gibbins: Do you need new services?
[00:17:11] Geoff Gibbins: Are there things you could be doing that are just fundamentally different that you haven't done before that are better suited to an AI mediated experience?
[00:17:19] Geoff Gibbins: So I've been working with, um, a friend of mine is a she's the chief marketing officer at a medical device company, and what they've actually developed is a new onboarding service for that is specifically marketed to AI agents that is an entirely new product that they've offered because it was very trustworthy and very appealing to AI agents and it also works a lot better for humans too.
[00:17:44] Geoff Gibbins: And so that's the kind of work companies need to do is not just how do you market to ai, but what do you actually market them to?
[00:17:51] Jonathan Wagstaffe: Wow. So they've built a new revenue stream there purely on the back of the AI piece. On top of their existing
[00:17:56] Geoff Gibbins: Exactly.
[00:17:56] Jonathan Wagstaffe: Revenue pieces. Nice. Like it.
[00:17:58] Jonathan Wagstaffe: Is there anything, anything that's [00:18:00] emerged where you can say, uh, this is something that always needs to be true about a brand for an AI agent to reliably choose it?
[00:18:06] Jonathan Wagstaffe: Are there any real North Stars yet?
[00:18:09] Geoff Gibbins: I mean, there are some super quick fixes, right?
[00:18:11] Geoff Gibbins: So, so this is some of the stuff that I, I see in talking to people, so. You know, if every company pretty much has a website, you, one of the things you can do is just make sure that. AI can read your website easily. That doesn't take huge infrastructure changes or org redesign.
[00:18:30] Geoff Gibbins: It's just let's make sure that there's structured data that reflects the accurate content that we want on the site.
[00:18:37] Geoff Gibbins: And then simple things like I, I worked with a company recently. They had a a lot of amazing reviews of their products on their website, but they were all stored as images. You just store 'em as text, like simple tweaks like that make a big difference.
[00:18:49] Geoff Gibbins: And so there's some very quick things that anyone can do that really don't cost much money or time at all, uh, but can have a huge impact.
[00:18:58] Danny Denhard: We were talking and I've [00:19:00] made a load of notes, which is a, is always a good sign.
[00:19:02] Danny Denhard: But one of the things you said, and I think it's vital, like a, just a, like a recap so far, is it AI in the way that you explain it, is another channel. So I think that's a whole new way for businesses to think. And I also think with agentic commerce. You have to be not only AI ready, but you have to really understand your reputation and your offering.
[00:19:24] Danny Denhard: And a lot of people who I deal with and potentially who Jonathan deals with as well, they're seeing it as one whole piece as opposed to potentially a new stream that they have to adapt to. And where most people fail in this like workshop and hackathon approach.
[00:19:42] Geoff Gibbins: Yeah.
[00:19:43] Danny Denhard: And trying to take it along their own way is.
[00:19:45] Danny Denhard: It's almost trying to take it like BAU and, and just trying to drive something forward that looks and feels the same and they're just gonna try and apply the, apply the same rules. Do you think with agentic and a lot [00:20:00] of other things that you've said, do you think with agentic there are signals where people can apply lessons and learnings from?
[00:20:06] Danny Denhard: From other tectonic shifts, or do you think this is real people have to get it into their, into their work ethic? That it is brand new, it is net new activity. It is such a dedicated channel that you have to prepare and, and bring people on, um, to drive, to drive this real change?
[00:20:24] Geoff Gibbins: Yeah. Yeah. Great question.
[00:20:26] Geoff Gibbins: So I would say, I think the thing that can be learned from other tectonic shifts and technologies is that ultimately the real value and the work that needs to be done eventually is not the technology adoption. It's actually how do you reinvent the way that you work? Right.
[00:20:44] Geoff Gibbins: And so that, that is, that has been true of e-commerce, of social, of cloud Metaverse.
[00:20:51] Geoff Gibbins: Name your technology, right? Like, well, not Metaverse 'cause no one did it. But, but you know what I mean? Like that that would've been the way to do it with Metaverse.
[00:20:58] Geoff Gibbins: But so that, [00:21:00] that's the real lesson that we can learn from other technologies. Um, the thing that, and so that's, that's, you know, that rhymes that continues.
[00:21:08] Geoff Gibbins: The thing that's different about AI though is that it's not a simple deterministic technology. It is weird. And so there is this whole journey that we're going on as, I guess a species of learning about this kind of strange, similar, but different form of intelligence. And one of the things that I found really interesting is that I've actually learned more about what it means to be human by interacting and learning with this new type of intelligence.
[00:21:36] Geoff Gibbins: And, and, and actually the, the fact that it's not one kind of intelligence, it's a bunch of different types of intelligence that are all kind of weird in their own ways.
[00:21:44] Geoff Gibbins: And so it's not like cloud, it's not like the internet in that. In that way, it's actually this whole new thing that we have to understand that operates kind of alongside us, but in a very different way.
[00:21:58] Jonathan Wagstaffe: That's, tell me more about that, Geoff. Like what [00:22:00] you found out more about what it means to be human. What are the, what are the learnings you can share with us there? That's fascinating comment.
[00:22:06] Geoff Gibbins: So I think, I think if you think about how, so this, maybe the first thing I would say is people often will criticize AI for not being very good at things, right.
[00:22:18] Geoff Gibbins: I remember there was a really funny quote that I saw like a year ago or something, and it said, I can't believe that AI makes mistakes. Whenever I read 60 million books, I never make an error. Right?
[00:22:28] Geoff Gibbins: It's like, it's kind of funny when you compare, like the, you know, people say, well, ai, you can't actually explain how it made that decision.
[00:22:35] Geoff Gibbins: We can't really actually understand how humans make decisions still, we've been studying that, like humans have been studying that for thousands of years, and we're still trying to figure it out.
[00:22:44] Geoff Gibbins: The way that, um, you know, the way that you, you know, the way that humans sort of confabulate memories is kind of the same to the way that ai, you know, AI hallucinates.
[00:22:54] Geoff Gibbins: So those, those are some of the things. But I'd say in interacting with ai, I think the real [00:23:00] power of collaborating well with AI comes with thinking about how do I as an individual learn best?
[00:23:05] Geoff Gibbins: How do I think best,
[00:23:06] Geoff Gibbins: and how do I deliberately work with this partner that can help me to to think and work in new ways that are right for me.
[00:23:14] Geoff Gibbins: And so I think it kind of shines a mirror on who you are as a person.
[00:23:18] Jonathan Wagstaffe: Wow. Okay.
[00:23:19] Danny Denhard: There's a, there's something, when you're talking something, when you're talking around like behavioral science, we've. We've started to learn and understand how different people interact, how we have our own biases. Like you said earlier, it'd be really interesting to see how behavioral science shifts and changes once we understand as "human intelligence". And then there's this "new intelligence". And I think it'd be really interesting, uh, the, the sort of knock on and the ripples that happen, say in the next few years and, and understanding that in the work environment would be critical.
[00:23:51] Jonathan Wagstaffe: I think linked to that is, uh, the points you've already talked about, Geoff, which is there is an assumption that machines are kind of infallible and they're consistent. And as you've already said, [00:24:00] AI exhibits biases and those biases change maybe daily. So, so it's not even, there's not even a consistent, uh, bias there. It's a, it's a much more random process maybe than we might Yeah,
[00:24:11] Geoff Gibbins: totally.
[00:24:13] Danny Denhard: Just to round off with Agentic, is we're slowly starting to see in the Western world.
[00:24:20] Danny Denhard: WeChat and, Qwen in China have, it's kind of been under the radar. It's happening more and more out there. Do you, do you think we'll be learning a lot from the China, China and the Chinese way of agentic, or do you think the way that these LLMs and really big institutions are attacking it, do you think it'd be different or do you think there will be some sort of convergence that would be very similar?
[00:24:44] Geoff Gibbins: Yeah, great question. I mean, I, I'm not an expert on China, so I should, I should caveat that, but I, what I do think is that a lot of what will, a lot of what will happen and not happen will come down to human factors of how do people actually build [00:25:00] trust in working with AI and in in people as groups working with AI too, like right now, I think so much of what we talk about is like one person and one AI system.
[00:25:10] Geoff Gibbins: The dynamics of communities, of humans and AI working together, I think that's gonna be. That will really determine how the future evolves, I would say, and that's very culturally specific.
[00:25:21] Geoff Gibbins: And so I think what doesn't work is companies sort of assuming that like a, you know, sort of Silicon Valley, New York City approach to life and community and what that means is gonna apply everywhere in the world.
[00:25:32] Geoff Gibbins: That usually doesn't work out. Uh, when people try that.
[00:25:35] Danny Denhard: Yeah, I think the removal, when, when people start thinking around actually really building for agentic, I think the tribalization of, how we are as humans and we look to be safe in tribes. I think there're gonna be so many geos and so many different approaches when it comes to, comes to this.
[00:25:52] Danny Denhard: And interestingly, the CTO of Mozilla said recently, he thinks it's gonna be all open source, smaller models that will be, you know, leading the [00:26:00] charge in say, six months.
[00:26:01] Geoff Gibbins: Yeah.
[00:26:01] Danny Denhard: Community based.
[00:26:03] Geoff Gibbins: Yeah.
[00:26:03] Danny Denhard: So I do think I, my, my current way of thinking is I think there will be community versions of, of LLMs that come out and not necessarily be localized, but I do think this multiplayer as opposed to single player LLM is gonna be a, gonna be such a big shift for so many people to get ahead around.
[00:26:21] Geoff Gibbins: Yeah, totally. And I think that might be a big part of how people start to build more trust too.
[00:26:28] Danny Denhard: A hundred percent.
[00:26:30] Jonathan Wagstaffe: So if we've got a, if we've got my agent that's doing my buying and somebody's got their agent doing their selling, Geoff, and there's a bad decision or a mistake in there, how do we manage the accountability there?
[00:26:40] Jonathan Wagstaffe: Who's accountable if those things start to go wrong?
[00:26:44] Geoff Gibbins: Yeah. So that's really interesting. So, so right now. The LLMs actually don't want to have that accountability. It's sort of like a bit of like a hot potato question in the industry as I understand it.
[00:26:57] Geoff Gibbins: Um, I think they may, they [00:27:00] may do the do so in the future.
[00:27:01] Geoff Gibbins: 'cause I think we're, we're still so early in this journey. I think it's really important to understand
[00:27:05] Jonathan Wagstaffe: Yeah.
[00:27:05] Geoff Gibbins: The, the actual implementations of AI shopping agents, they're not really actually looking at like completely autonomous shopping decisions in, in most contexts.
[00:27:15] Geoff Gibbins: So right now it is, I, you know, usually it's kind of the merchant or the, um, payment system that might change.
[00:27:23] Geoff Gibbins: I think what's the problem is that it becomes a lot more murky when you don't know who's actually involved and how a decision was made.
[00:27:30] Geoff Gibbins: So I think that's gonna be a huge legal and economic issue. 'cause then it comes down to who actually pays for when things go wrong. Um, I know this is a topic that a lot of the payment networks are looking at actively, um, in terms of how do you define that accountability and who should that be, but I, I don't think we've seen who that will be yet.
[00:27:49] Jonathan Wagstaffe: Yeah. And I think you even get, even outside of the world, ai, you know, if I'm, if I'm in the UK buying from a US vendor on a commerce server, a server that's in Singapore [00:28:00] with a payment system that's in Switzerland whose law applies even if you agree that there's a liability that whose law are we suing each other under? There's lots of things like that. I think on top of the AI question that still have to play out. Where, from your point of view, where, where's the red line?
[00:28:15] Jonathan Wagstaffe: Where in this new world, where should humans stay involved? No matter how good the agents get? Are there certain places that there must still be human intervention?
[00:28:26] Geoff Gibbins: I think so, and I think the complexity is that that will be different for every person, right? So I think as an individual I might say, I absolutely want to choose every toilet paper purchase, but you might say, that's ridiculous, Geoff.
[00:28:41] Geoff Gibbins: Why do you wanna do that? Right? And so people's preferences around what they're willing to delegate and not will change and be, uh, change over time for them and then be individual and that will look different when you are. Buying in different contexts for your family versus for yourself. And so it will be very individualized and [00:29:00] contextual.
[00:29:00] Geoff Gibbins: And so that's a whole realm that needs to be unpacked. Um, as, as well, I don't, I don't think we have the answer, but I think it, it won't be one answer
[00:29:12] Jonathan Wagstaffe: if we look forward. It's go to 2030, tell me what's the likely scenario and maybe speculate on a wild card that people are ignoring at the moment that you think might be more influential than they realize. Question.
[00:29:25] Geoff Gibbins: Okay. So I think 2030, I think we, we won't be talking about agentic commerce, right? 'cause it will just be buying and selling stuff. Um, that that's, you know, like no one talks about did you use a website to sell this? Right? It's like, well, obviously, yes. Um, so I think that's the first thing.
[00:29:42] Geoff Gibbins: Uh, I would imagine that my prediction is that we're gonna see a lot of growth in very AI driven purchases, right? Which is kind of what we think of today in that it will be [00:30:00] actively researching buying, but I think we're gonna see new dimensions emerge. So the first is that the realm of recommendations, I think, will look quite different. So right now a recommendation from a sort of machine learning type AI system might be buy this product.
[00:30:17] Geoff Gibbins: I think that might evolve to a much more complex and interesting space. So instead of buy this product, it could be sign up for this entire plan and we will manage your groceries for the entire year.
[00:30:28] Jonathan Wagstaffe: Okay.
[00:30:28] Geoff Gibbins: Right. And you're buying into a system of recommendation and it's gonna evolve and adapt and learn from what you have in the house and not, and all that kind of stuff.
[00:30:37] Geoff Gibbins: So it won't be about like individ individual product recommendations.
[00:30:40] Geoff Gibbins: It'll be about how do you orchestrate buying and selling for an individual, for a company. Um, I also think probably a lot of the bigger changes we're gonna see are actually gonna be within companies in terms of how they actually manage procurement as well, not just consumers and then I [00:31:00] guess the wild card in all of this is that everyone's talking about this in the realm of chatGPT and Gemini, like a chat bot that you're interacting with on a computer. Uh, I think what we're gonna start to see is the, is the interplay of that.
[00:31:15] Geoff Gibbins: Humans and then robotics. So in any kind of space where there is stuff sold like a store, for example, right?
[00:31:23] Geoff Gibbins: There will be that you already see robots in some stores. Um, there are gonna be, there's gonna be much more robotics in those stores though monitoring the shelves, monitoring the people, interacting with the people, interacting with the shelves. Shelves that move around and restock, or whatever it might be.
[00:31:39] Geoff Gibbins: I think there's gonna be a lot more that happens in the integration of robotics.
[00:31:42] Geoff Gibbins: That'll probably be quite early in 2030, but I think that's where the conversation will be rather than chatbots, uh, in a few years.
[00:31:50] Jonathan Wagstaffe: You, you just move one step from that to your point.
[00:31:53] Jonathan Wagstaffe: Okay. So you get the AI to manage your shopping, your food shopping. It's only one more step. Then say, and I'm [00:32:00] actually gonna get the AI to talk to my health data.
[00:32:03] Jonathan Wagstaffe: And so the AI's going, you've put a couple of kilos on this week, you need to just eat more salad. And so you can start to see back to something that Dan and I were discussed on a recent pod.
[00:32:12] Jonathan Wagstaffe: We were talking about devices, wearables, and the way that they become. Life assistance almost.
[00:32:16] Exploring the Role of AI in Daily Life
[00:32:16] Jonathan Wagstaffe: They, they're advising in all areas of your life. 'cause they understand you and there's implications around privacy there. But that, you know, that's one possible route forward.
[00:32:23] The Future of Retail: Robotics and Personal Shoppers
[00:32:23] Danny Denhard: the one area that a lot of people just aren't talking around is, is robotics and robots. And my take is that, like you said, if we go in a store, you, you might actually have like a personal shopper with you that is that robot and it might, you might swipe a QR code or you might put a thumbprint or something and it then might interact with your, your own "small language model" and it'll go and help you and it says, oh, you haven't eaten for six hours. Do you want me to rustle up a rotisserie chicken meal for you? In that moment, I can imagine that there's gonna be all these, remember the am Amazon-fication of the internet. I can [00:33:00] imagine that's gonna actually happen and, and revitalize a lot of retail stores and not, and high streets or, you know, blocks in in us.
[00:33:07] Danny Denhard: I, I do think there's there. And one other thing that you said, which was really interesting is.
[00:33:13] The Importance of Loyalty Programs in Retail AI
[00:33:13] Danny Denhard: if I was in one of the big super supermarkets or the superstores sales Tesco's or, uh, Kruger's in America, for instance, one thing that I would be really interested in looking at is that their play like the clubcards, their store card and the importance of that, uh, I upset someone once when I said, when I was speaking at a conference, why I don't understand why Tesco's or Clubcard don't ping me and say your milk is running out.
[00:33:39] Danny Denhard: You should order it now or get, you know, a rapid delivery or quick delivery service.
[00:33:44] Danny Denhard: I can imagine it might actually be the clubcard that is more important than Tesco's or Krugers because it, it knows so much contextualized information about you and all you'll need to do is, is scan and interact a little bit more, and then it's gonna have so much more about you.
[00:33:58] Danny Denhard: So if I [00:34:00] was in that space, I was like nectar from Sainsbury's, for instance. You could, you could see it accelerating so much further and faster than maybe the supermarkets themselves. So that's one area that I've been sort of talking to different individuals. And the way you explained it, Geoff. Sort of rang that bell in my head to, uh, to bring it up.
[00:34:19] Danny Denhard: So I definitely think there is that, um, to it. Yeah,
[00:34:22] Geoff Gibbins: totally.
[00:34:22] Identity Tokens and Privacy Concerns
[00:34:22] Geoff Gibbins: I think you could think of them as like different sort of like identity tokens that you have to take around the world with you. And depending on how much they know about you, they can contextualize the information that's given to an AI agent without actually sharing all your information.
[00:34:39] Geoff Gibbins: Right. It can just share the information it needs to make that recommendation to you. And it is a protector of your, you know, more personal information, whether that's your health or how much milk you have in your fridge, that, that can be important too. Um, and I think then the question becomes, who do I trust to be that identity token for me?
[00:34:58] Geoff Gibbins: Is that nectar [00:35:00] or is it clubcard or is it my phone or is it my bank? Like, who, who's actually gonna be that?
[00:35:07] Final Thoughts and Key Takeaways[00:35:07] Closing Remarks and Future Plans
[00:35:07] Jonathan Wagstaffe: So Geoff, where should people follow your work?
[00:35:10] Geoff Gibbins: Yeah, so people can follow my work on, uh, my company website, human-machines.com. Uh, also on LinkedIn. And uh, yeah, I, I have a newsletter which people can follow along as well, usually with a couple of posts a week.
[00:35:25] Jonathan Wagstaffe: And of course there was recently the book.
[00:35:27] Geoff Gibbins: Yes, so I, uh, recently published a book called When AI Shops, it's all about the revolution in how we buy and sell and what companies and individuals can do to understand what AI is as a channel, what it takes to market, to humans, and to agents and, and how to get started on that journey.
[00:35:44] Jonathan Wagstaffe: And we can highly recommend that Dan and I are both fans of Geoff's work and, um, it's really worth reading for those of you that are, uh, looking at this kind of space. Geoff, if there's one takeaway, if there's one thing that listeners should remember from what we've just got talked about today, what should that be?[00:36:00]
[00:36:01] Geoff Gibbins: Yeah, I'd say, uh, the first thing is that AI is a channel, not just an automation tool. It is a way of buying and selling products and services that every company needs to start to learn about and it's super weird, uh, and fascinating. And to get started, there are some simple things that any company that can do, no matter who you are.
[00:36:26] Jonathan Wagstaffe: Brilliant. Geoff, it is been an absolute pleasure to speak to you today. Thank you for sparring us the time and good luck with everything you do in this spring.
[00:36:33] Geoff Gibbins: Great. Thank you.
[00:36:34] Danny & Jonathan Interview Recap
[00:36:34] Jonathan Wagstaffe: Dan, I really enjoyed that interview.
[00:36:36] Danny Denhard: That, that was great. I, I loved geeking out and I loved just hearing someone who's in the middle of it, just who wants to tackle and help businesses just progressing in AI and in a human and tech future.
[00:36:51] Danny Denhard: I think it, I think it was brilliant and so much to learn from it.
[00:36:55] Jonathan Wagstaffe: And I like, you know, there was some very practical advice, wasn't there? There was, you know, we were able to go into the middle [00:37:00] distance and say, what might it look like? Geoff was able to go into some of the deep technical stuff, but at the same time I thought there was some really good practical advice about the way we have to behave this year.
[00:37:09] Jonathan Wagstaffe: Businesses, things that business have to think about this year.
[00:37:12] Danny Denhard: Yeah, exactly. I, I, let me, let me share a few of my scribbled notes down. Um, and I think that these are the ones that business leaders and, and department leaders really should sort of take away, we'll share it on the newsletter as usual. But, the most important takeaway for me was.
[00:37:27] Danny Denhard: Geoff really does understand AI as a channel, and I think we, everyone has to start thinking about that. It's not just an extension, but it's a whole dedicated channel.
[00:37:36] Danny Denhard: We really do have to think about it as an ecosystem because it's gonna keep evolving and there's gonna be layers and layers built on top and connected layers to what we're doing now.
[00:37:46] Danny Denhard: I think as we, we alluded to, it's, we have to think of it as tech and human. We can't think of it as, as one or the other is gonna be the tech side, which might be agentic and, and [00:38:00] other ways. And there's gonna be human side of humans make decisions and buy and purchase in different ways. And then I really liked the future framing from Geoff, which was the future is LLM and ai.
[00:38:11] Danny Denhard: It's agentic. There'd definitely be a wearable play. And then there will all be robotics as we've sort of hinted at before. Is there anything you would add to that?
[00:38:21] Jonathan Wagstaffe: I think the, the short term one is around, um, you, you need to be thinking about people and AI as you are doing the work you're doing in the next year or two.
[00:38:30] Jonathan Wagstaffe: Uh, the, the things that AI looks for are not necessarily the things that people look for. And his, his point about if you, if you play a scarcity card by saying there's only three left, that will trigger humans to act.
[00:38:40] Jonathan Wagstaffe: It will trigger AI not to recommend you. So you've gotta be playing for both audiences in, at least in the short term.
[00:38:47] Jonathan Wagstaffe: So a really fascinating interview. Really enjoyed that and hopefully, you know, we can, we can talk to Geoff again in a few months when things have evolved. It will be really interesting to get Geoff's takers this year evolves.
[00:38:57] Danny Denhard: Definitely, and if you need to re-listen, [00:39:00] or you want a break or a full breakdown, definitely subscribe to the newsletter where we'll break down the 90 day plan that Geoff, hinted at and how Jonathan and I are really thinking and, and pushing on this subject as well.
[00:39:12] Danny Denhard: So thanks so much for listening to everyone and, we'll have many other interviews like this in the future.
[00:39:18] Jonathan Wagstaffe: See you next time.
[00:39:19] Jonathan Wagstaffe: Danny.
