Dunnhumby ventures Invests in Azoma | Max Sinclair on The Agent Economy

Dunnhumby ventures Invests in Azoma | Max Sinclair on The Agent Economy Speaker 1: Hey there, guys. Welcome to The Agent Economy, Episode 15. We're back in real life. We're back in person, myself and Stephen. Today, we're joined by Max Sinclair, CEO and co-founder of Azoma. It's great to have you. Thanks for coming. To start, do you want to give us a bit of an introduction to Azoma and what you've been up to? Speaker 2: It's such a pleasure to be here, guys. I know you did an episode where you didn't mention Azoma, and I got a little bit upset, especially as Stephen's a customer. Get my whole own episode now. I think it's fair. But yeah, I'm Max. I'm the founder of Azoma. We are an agentic commerce optimization tool. We're working with 18 of the largest 50 CPGs, Mars, Unilever, L'Oreal, etc. And then we also have a bunch of really exciting challenger brands. And then we also have some agencies as well that we that we work with as well. Speaker 3: Awesome. Very cool. And I think there's maybe some big news that's dropping from Azoma because you were telling us before the pod that Some exciting things happening on the investment side. So can you share more there? Speaker 2: So we have some breaking news. First time we've discussed it, which is we have just closed some investment from Dunnhumby Ventures, which we're really excited about. And Chris, you know Dunnhumby well, right? Speaker 1: I do indeed. Tesco, an old client of mine. And yeah, we were moving into the retail media space with Dunnhumby. So yeah, very exciting. It's a huge bit of news. Speaker 2: Yeah. And I think the key thing about this is, as you guys know, you talk about GEO and AEO the whole time. The actual measurement is a hard thing. What is a share visibility? A, share visibility, in my mind, is a bogus metric. But B, even if you take it as a good metric, what does that actually mean into revenue? And this partnership and investment from Dunnhumby will allow us to do that and basically track eventually what increasing a brand share visibility will mean in ChatGPT and Gemini to sales in Tesco's and in You know any of the other retailers that they partner with so it's been a lot easier for us on the Amazon Rufus. A Walmart Sparky site to attribute our work to revenue gains. It's been very, very hard to do that in ChatGPT and Gemini. And I think this is going to be a big step forward to enable us to do that. And I think that's just, it's really exciting. Speaker 3: That is super interesting. Like there's a few things to unpack there, but totally with you that like on platform tools like Sparky and Rufus or Alexa for shopping, like having Influence there, like it's a lot easier to maybe see that definitely like flow through into the sales uplift on the platform. Yeah, but what you're talking about here is when you're influencing the AI models and improving or increasing, let's say, the recall or the visibility for brands in those models, so more people see them. Is there then a brand uplift for them in the whole market across all the retailers? And what's super interesting about that is I know from time working at places like Unilever, Mars, when we were doing Amazon advertising, And we were doing more mid-upper funnel advertising rather than just the pure lower funnel stuff. The question was always like, what's the impact off Amazon? And they would do brand uplift studies. And those brand uplift studies were the unlock to more investment in the channel. So that's a great unlock for you, Azoma, to be able to bring out the influence of the tool. Speaker 2: Exactly. And we want to build the best, like, we're the only vertical solution. So there's so many horizontal players, and you've discussed all of them on your previous podcast, and I'm not going to go into it. Focus on consumer brands, yet only. And therefore we want to build, you know, we don't want to be the best for SAS. We don't want to be the best for FinTech or finance. We want to be the best for consumer brands. And I think the work we do on Amazon Rufus and Walmart Sparky already It means that we're better than all these horizontal competitors who only look at kind of half the agentic commerce world from their perspective if you miss out where most of the revenue is today. But then this is going to enable us to also have a better offering than these horizontal competitors in a chat between the Gemini. So it's very exciting and yeah, I look forward to get back to building now. Speaker 1: How did you land on that particular ICP, that particular vertical? Was it your background? Speaker 2: Yes, I'm ex-Amazon. So I spent six years at Amazon. I worked in Amazon search. So I was kind of responsible for making sure that when customers are browsing the catalog that the relevant products came up. You know, fixing the right browse structure and that kind of stuff. Speaker 1: Okay. Speaker 2: And with the deterministic AI, we'd always have this issue, you type in like yellow t-shirt onto Amazon, and you'd get a blue and a pink and a purple one because the AI wasn't smart enough to read color as well as just a keyword. Speaker 1: Sure. Speaker 2: And then AlexNet launched. I was working a bit on that. So AlexNet is the first AI model that could kind of like Tell the difference between a dog and a muffin if you know what I'm talking about. So we're doing that kind of stuff. And then when the LLM, ChatGPT launched, you know, me and my co-founder like, wow, this is going to change search. All of these retailers are going to adopt this because it's just obviously such a better experience. And yeah, I'd admit at that point, we didn't see Gemini and ChatGPT themselves becoming this search engine. But we definitely believe that Amazon and Walmart and others would adopt it. And that's when we started the business. And then, yeah, three and a half years later, we're still serving that customer that we had in mind back then. But obviously, now we do Gemini and we do ChatGPT. We do, you know, we do all of the LLMs as well as the retail LLMs. So Sparky, Rufus, Clementine on Instacart, Tesco is about to launch one. I think every retailer will have one soon. And yeah, we'll be. Speaker 3: Supporting super interesting that you have such good coverage. And, you know, in my experience, you know, working with a number of different platforms, like a lot of them just do ignore those retail LLMs, I guess, quite challenging maybe to track. But also, if you've come from the B2B space, then maybe it's just not even interesting for you. Speaker 2: I think they just don't understand these customers. I mean, fundamentally, I don't think it is particularly that hard to track, and I don't want to advertise what we do to everyone else to do it, but we're not the first people to do this, right? You have all of the existing digital shelf players who've been tracking these retailers for years. The Profiteer is a commerce IQ is a stack lines. So it's not like this is an impossible technology to solve, especially with AI. Yeah, it's harder, but I think it's more that these guys just don't understand this customer segment whatsoever. They are Almost all of them focusing on marketing teams, and they're not focused on e-commerce teams. In these companies, typically, we do work with marketing teams, we work with PR teams, we work across the board. But typically, the revenue leader is sitting in the, as you all know, when we're working together on Mars, is an e-commerce leader who's in charge of growing that brand revenue online. And that is kind of the ultimate stakeholder who's got a dollar tied to their name. And that is normally the person that we are, you know, who we're building towards and enabling that person to work across their organization with marketing, with PR, with legal, with finance, all in our, all in our tool. Speaker 3: Interesting on that, because I like totally with you. That maybe some of these other organizations just aren't thinking about it in the same way. And then, you know, I was an e-commerce director at Mars, at Lego. So it's like banging, you know, like what would be the remit for that type of role. And what I think is really interesting that's happening at large organizations now, Like AI search is a hot topic. Yeah, like everyone's talking about it. Speaker 2: Everyone wants a piece. Speaker 3: Everyone wants a piece. And it's also in theory, like a career accelerant, right? If you're on the next new hot thing, and it's going to allow you to be promoted, and whatever comes next. So do you see any of those challenges with the stakeholder management at large organizations like You're dealing with one person, lots of other people want to get involved. I guess you have customer success managers that are helping you. Speaker 2: Yeah, well, I think it's broadly a good thing. And we want to empower our, you know, part of one of our cultural principles is to make our customers look like superstars. And we want to like, you're totally right. Like we see our customers as like using Azoma as a career accelerant, and they can become this visionary leader. We're leading a 100 year old organization through this agentic commerce transformation. And that's exactly what we want to support them with. So yeah, we are very biased towards like helping our customer and equipping them with the right data, the right frameworks, we can talk about some point, like we've seen this play out now across multiple organizations. And we know how to like, ultimately, we've got to drive revenue tying back to the dunnhumby investment. That is gonna the first question is, oh, like, people probably using ChatGPT, how am I showing up? The next question will be, okay, I'm not showing us up as much as I want to, like, what can I do about it? And then you'll do some stuff. And then they'll be like, Oh, so like, like, how can we measure this improvement? And then the next question once they've presented that internally will be, okay, but like, what about revenue? What happened? Did this drive anything? And like, so we've seen that story play out. And obviously, we've, you know, got global contracts of Mars now from like, our early pilots on like a small number of brands. So we've seen like how this goes, and hopefully made like a lot of customers like superstars on the way that submission and, and yeah, like, accelerated some careers. Speaker 1: For sure. So we obviously see this quite a lot, the pace of innovation just in the market with the AI models, new innovations coming out to market. Have you seen there's been a disconnect from that to how quickly the enterprises can actually move? We're starting to see that there's really, there's challenges there, like so much coming out in the market. It's quite difficult for these large organizations to adapt. Speaker 2: I think it is, I think it can be overwhelming. And I think that's part of our role and why we still need humans is, you know, our custom success people will start most of their conversations about like what's just happened. Speaker 1: Yeah. Speaker 2: So like Metamuse just launched last week, Instagram, Instacart, Clementine just launched last week. And that like, making sure that our customers are like up to date on what's happening on the cutting edge of the industry, because they don't have the time, like they've got busy jobs and lives. So you know, they have a 45 minute call with Azoma, and like, which, you know, showing them some great results, hopefully tying it to revenue. Plus, like, you need to be thinking about Metamuse, you need to be thinking about Instacart, Clementine. So I think that's all part of I'm sure you guys as agencies, that's what you do. Speaker 3: Yeah, for sure. And I think sometimes I'm trying to actually just make it a little bit more simple for everybody because there is a lot of noise out there. There are a lot of platforms you can track and maybe, certainly thinking about in Europe, if you're a consumer brand and you're selling online, you probably want to be paying attention to Amazon, Rufus and Alexa. Probably you want to be paying attention to ChatGPT. Probably you want to be paying attention to Google's ecosystem and Gemini. But you probably don't need to track perplexity as much detail, you know, Claude, okay, getting some traction, but again, mass consumers, enterprise, exactly. Yeah, so like, actually helping them to narrow it down sometimes is helpful, but they need to know about everything because I've been in a meeting where we're talking about influence into ChatGPT and of course like Reddit comes up, YouTube comes up maybe a little bit more for Gemini and someone's seen probably my post or your post or one of your posts saying your Reddit citations have absolutely tanked in the last two weeks. So obviously they need to know a bit about that. Speaker 2: They need to know what's going on. Speaker 3: So they don't look stupid, yeah, in front of other stakeholders who also might have seen that content coming out. Speaker 2: They need to listen to your podcast. Speaker 1: There we go. There's an economy. Speaker 3: And follow us all on LinkedIn. But I think there's like a balancing act between like, hey guys, here's the latest info that you need to be aware of, but here's actually where you need to focus all your time. Speaker 1: Making it digestible and focus. Yeah, I think there's so many organizations out there, tech companies, that have led with product-led growth. But it's important to have that service layer to like interpret the changes that are happening in the market. Speaker 2: It's a different customer. I mean, ultimately, it's a different customer. Like we are, we know our customers are these like global You know, consumer electronics brands, and then like the challenger brands. Speaker 3: And that's your sweet spot. Those two. Speaker 2: Yeah, exactly. And that's who we work with. So we work with some of the most exciting challenger brands as well, like ancient and brave. And Perfect Ted and these kind of really like exciting guys who will probably be acquired for a billion, two billion dollars by Unilever, Denon, Nestle, one of our existing customers. Speaker 3: I had Perfect Ted for breakfast. I drank it this morning. Speaker 2: It's good. Speaker 3: Yeah, it's a good product. I like it. Speaker 2: Hopefully you found it through ChatGPT or Rufus. Speaker 3: Which is the best match as promoted by Azoma on the podcast. No, I think that's super interesting. And then in the global nature, because obviously you're dealing with some of these really large organizations working with them around the globe. So, you know, I talked a little bit about what the focus could look like for a brand if they're in Europe, but like, obviously, that's different if you're in the US. And then, like, are you doing anything in Asia at the moment? Speaker 2: Yeah. Customers care about Lazy Chat, Deep Seek. We have, you know, we have customers in Japan, India. Yeah, we're a global company in Outlook. Speaker 3: Awesome. And in terms of the platforms that you're tracking, are you looking to continue to expand those? Speaker 2: Yeah, I mean, we basically, our technology is set up so we can do whatever a customer pays us for. So the question is, is a customer ready to pay us for it? Do they care enough? But ultimately, like, we're agnostic. We can, you know, we've got, I won't go into details, but yeah, we could track any of these retailers, shopping systems as when they crop up. Probably take us like two or three weeks to set it up properly. Yeah, it's the same tech that we're using. So yeah, we track the Asian ones, obviously, like we're more US and European focused, given that we don't have like Any Japanese speakers in the company, right? So we do have actually Panasonic in Japan for Amazon Rufus, but like how much support we can give them is pretty limited. And it's for an agency partner as well. So they I think do that side of it for us. Speaker 1: What about the markets? Who's really like pushing the boundaries on the client side? Who's like really innovative, wanting to track loads of stuff? Would you say Asian market, Europe, US? Speaker 2: I think these, I think the people in these companies are really smart, honestly. And obviously, Stephen, you are one of these people, right? Like, they're smart people, they're switched on, you don't get to work at a Mars or Unilever or L'Oreal unless you're like, you know, a sharp person, right? So I think they see the shift that's coming. And, and, and yeah, like a lot, like they're a lot more on it than I think People would assume. They're really on it. They recognize agentic commerce is the next frontier. They recognize it's also a career accelerant for them. So I think people are pretty on it. Speaker 3: I think there's a couple of other interesting things there. So I think, you know, these large organizations, yeah, they're clued on. You've got some smart people there who are realizing this is something that they need to do something about. I think sometimes at the large organizations, just like the general process of getting things done takes longer, right? So when you think about a large organization, you're like, They're going to be a bit behind the curve because, you know, it takes a long time for a big machine to move or talk about it like a big ship taking ages to turn versus a challenger brand who could maybe say tomorrow, let's go. Here's the check. Speaker 2: But as soon as we get into one place, like when we started with Mars, we were in like, I can't remember what it was like m&m's like you get one brand of one marketplace as soon as you get that then wildfire right and the other people in the organization looking foods looking yet pets looking you know different countries looking although like Mars is more organized by brand rather than country but. They were quite unique in how they're structured. But fundamentally, you get a good use case, you prove ROI quickly, and yeah, it spreads. Speaker 3: Nice. And then we talked already a little bit about, potentially a bit complicated, all of the information that's out there at the moment, maybe trying to simplify things for clients. I know you've got a framework, the 5Cs. Speaker 2: We do. Speaker 3: Tell us a bit about how you use that, how you implement it, and actually, what do the 5Cs stand for? Speaker 2: Yeah, so we worked with the Digital Shelf Institute to set up the 5Cs for agentic commerce. So that is completeness, citations, correctness, context, and customer acquisition. I'm glad I got them all live on the job. Speaker 3: With our notes. Speaker 2: Exactly. So completeness is about having your data. Speaker 3: Yeah, which is super important, isn't it? And if you think about that one, I remember there was an example I gave in a presentation. I used to work for Pocket Tea back when it was owned by Unilever. Speaker 2: We've got them as well. That's another one of our customers. Speaker 3: Great brand. Drank their tea again this morning after the match that happened. And I remember, I mean, I was probably responsible for this, so I feel like I'm allowed to say it, but there was one of their products before you were working with them. So this must have been a couple of years ago. I had a look. And it was green tea. I see green tea's naturally caffeinated, it's got caffeine in, and on the attributes it said not caffeinated, right? And it was probably me that did the upload form and spanned that, so I have to potentially take some responsibility for it. But, you know, two or three years ago it didn't really matter. Like, those attributes weren't necessarily surfaced super high on the PDP, so humans are not necessarily going to read them. Speaker 2: Only if you're clicking on the refinement on the side of Amazon. Speaker 3: Exactly. Speaker 2: Nobody searched like that. Speaker 3: No, exactly. So it didn't really make that much difference. Now, obviously, critical. Because the machines in the background are going to look at it and say, This consumer is looking for a green tea. Obviously green tea has caffeine. This says it doesn't. That's an error. Not going to show this product. It's critical that brands get that right. Speaker 2: And I would add that the completeness means different things to different agents. So for Gemini, we're talking about GMC, Google Merchant Center. Speaker 3: Which they've actually expanded out all of the attributes as well now. Speaker 2: Exactly. It's basically like an Amazon vendor. You know, Walmart vendor. So like the different agents will have different places that data needs to be complete. And I also like to say like SEO is part of completeness. So like keywords and all that stuff. We're not saying throw this away. We're also not saying, I kind of don't like it when people say to me, oh, we've got to fix the SEO and then we're going to do the AEO. It's like, no, it's one holistic thing. And like, it's part of SEO, it's part of completeness, foundational part of it, you know, as are your backing attributes and all the boring stuff. Next one is context. So basically, and this is something we help our customers to do, identify what questions customers are asking the shopping agents. And then generate that context in those answers. So the context will change based on trends, based on time of year, and basically having product listings or blogs or whatever it will be with the right context answers questions that the consumers are asking the shopping agents. Speaker 3: For sure, makes sense. So that's like me going and asking Rufus about what's the best Matt shared to help me wake up in the morning. I want to feel energized and then figuring out, okay, what's the copy on the PDP that's actually going to speak to the type of question I'm asking. Speaker 1: And how are you getting some of that insight into what the consumer is typing and looking for ultimately? Speaker 2: Well, I mean, on Amazon and Walmart, it's easy because you get the data from them and Gemini. On ChatGPT, it's completely impossible, black box. Nobody knows apart from OpenAI. And I personally think that all of these bogus, like we never use panel data. I think it's complete garbage that some of these other AI tools will provide. But again, it comes back to Dunnhumby and like we are, you know, leveraging their plan to be leveraging their data in a way to look at that as well. That's, you know, we understanding what prompts to track and then tracking the revenue of the prompts and stores as part of that partnership that we've been discussing with them. Speaker 1: That'd be huge. That'd be a unique position for you to have access to that first party data. Speaker 3: And on that, what prompts are people asking? Do you think that, I mean, I have a view that at a certain point in time that data is going to be made available because they're going to want to monetize it. Speaker 2: A million percent. And it's already available in Gemini, as you know. So we basically will use what you can get from Gemini, Amazon, and Walmart onto ChatGPT. Speaker 3: Gotcha. Speaker 2: That's an approximation. That's a good approximation for now. And then leveraging Dunnhumby as well with that. And I think that's much better than like just hallucinating some garbage and claiming that you've got this panel data of like what people search for, which is. Speaker 3: And one more question, then we'll come back to the next C. So you kind of mentioned, I guess, across different platforms, there'll be maybe different views about volumes of prompts, maybe even answers of prompts. I think, you know, for people who are watching who may be working at brands and they're thinking about using different platforms and they see different results, maybe what would you say to them about that? Because there is some volatility, isn't there? Speaker 2: Yeah, I mean, this is why I think share voice is a bogus metric. And I would encourage our customers and brands watching to say like, share voice to who? We share voice, right? Because obviously, these are super personalized. So what share voice to the API, you're going to go into peak and you're going to look at some API data. And you're going to say, Oh, we've increased your share voice from 5% to 10%, according to what we've collected off ChatGPT API, totally different from what consumers see. Speaker 3: Sure. Speaker 2: And we've we've actually launched this digital twin technology. I don't know, maybe you looked at this and maybe not. But this was one of the early things we did of simulating How customers are speaking to these shopping agents. Customer gives us their persona. We simulate how that customer is speaking using the data we have from Amazon or Walmart or Gemini or Google Analytics to speak to and therefore we can have like specific, you know, share our voice. So I think the generic stuff that you'll see in some of these tools Totally garbage, literally useless data to be tracking anything. Speaker 3: Because you should be giving a persona to say the same thing. Speaker 2: Share a voice of what? Exactly. The API's share a voice, which is also different from the front end. I mean, it's mad. There's a lot of fake. Speaker 3: I totally get that point. And maybe on the persona bit, and we'll definitely come back to the 5Cs after this, but on the persona point, Like I've heard some challenges, which is like, hey, well, everyone's got different chat history. Everyone's slightly different. Everyone's going to be asking questions in a different way. But what is also true is as humans, we're all quite similar. We all have similar themes, you know, not necessarily everyone's collectively together, but you can, I think, quite easily define collections of people into personas. Speaker 2: I mean, what we use, our customers have got personas who they've been marketing to for years and years and years. We look at the big brands, they'll be like, hey, L'Oreal will have, you know, premium beauty, this, this age, this, obviously, maybe women style. Speaker 3: Exactly. Speaker 2: And then you have a persona. Speaker 3: Yeah. Speaker 2: I think one of the few companies that, you know, the food companies tend not to, so if you're... Speaker 3: Because they're more mass. Speaker 2: Exactly. If you're selling snacking or, you know, like chocolate or pasta or whatever, like, you don't have a persona. Speaker 1: Yeah. Speaker 2: And then fair enough, you can use a just API is a proxy API. Speaker 3: Exactly. Speaker 2: But really, like any other brand, which is doing you will they'll have it the marketing will have a persona. Yeah, a million percent. Right. And maybe the eCommerce team doesn't even know that there's a persona that marketing have that that happens sometimes. Speaker 3: For sure. For sure. Speaker 2: Yeah. Speaker 3: Okay. So, context. Speaker 2: We've done completeness, we've done context and citations. I mean, probably don't have to explain it to you guys, but citations is about getting your brand in the sources that the AI agents are going to for their answers. Speaker 3: For sure. But one interesting point on that, I think, is On Amazon now, in the US, you might correct me if I say anything that's incorrect here, but I think on .com, you pretty much search anything on Alexa for shopping and it tells you the citation sources that it's pulling information from. Not seeing that in Europe yet. Speaker 2: We don't have it yet. Speaker 3: No, no, no, no. Speaker 2: But we will. Speaker 3: Yeah, yeah. Speaker 2: Breaking news. I can break that news. We will. We will have Alexa for shopping. I would predict that comes before our next big shopping event is part of that. So we'll have it in at least in the UK and Germany, I'd guess. Speaker 3: And then what's super interesting about that is if you think about the optimization game on Amazon, for all of Amazon's history has been about the content that you have on Amazon, the images, the titles, the descriptions. All of a sudden that's getting blown wide open with, hey, we're also pulling information from these other sites across the internet and using that to inform the answers. And so for anyone who's been thinking about optimization on Amazon, that game now becomes optimization on Amazon and off Amazon as well. Interesting. Speaker 1: Is that a recent change Amazon made? Speaker 2: It's like a year. Speaker 1: That's been around a year. Speaker 3: Yeah, in the US, right? Speaker 2: Yes. Speaker 1: Yeah. Speaker 3: Yeah. So I mean, it's just the same as walking. Yeah, it's in the same thing as all of it. And just in the chat, like if you go to Amazon.com, ask anything at the bottom, we'll just tell you where the sources of information are. Speaker 2: And they have, they've got licensing deals with specific publications, same as OpenAI, same as Gemini. I think Amazon is like Condé Nast and a few others. So they will bias, you know, all the shopping agents will bias towards where they have a licensing deal to get access. Yeah, and you can see where that is. So like a free tip is just go and like look where Gemini, OpenAI or Amazon have got those licensing deals and just Target your PR teams towards those ones. Speaker 3: Also get a deal with those people. Speaker 1: Is it generally the large publications that they're obviously cited sources? Speaker 2: Yeah, when we started tracking Amazon, we should actually do like a little LinkedIn study on this. Now you mention it, but like off the top of my head, we've tracked tens of millions of prompts and maybe we'll follow up with some data. It started being the Amazon affiliate sites. So I don't know if you remember, like back, you know, it started being like these random blogs that would just rank products. Speaker 1: Sure. Speaker 3: Once you're making money from kicking people to Amazon. Speaker 2: That's who that they would literally use those rankings. And then obviously, these affiliate sites had no idea that this was influencing resource. And we would be contacting them and adding people in on behalf of the customers that shifted a little bit to the more We've covered standard big publications, a bit like an OpenAI or a Google Gemini experience. Speaker 1: Sure. Speaker 3: Makes sense. Speaker 1: Interesting. Yeah. I mean, we covered the Reddit stuff last episode and helping brands navigate the changes in citations, like one day it's this, the other day it's this. Speaker 2: It'll be back, yeah. Speaker 1: Yeah, I'm sure. I mean, it's still using for the training data and it's still using it, but it's not necessarily a citation. But that's another challenge we face with brands that we're working with. Like week on week things are changing pretty dramatically. You have to kind of adapt the strategy and make sure your clients are on top of things, but it's difficult with the innovation we've got in the market. Speaker 2: So it makes it fun, right? The number four is correctness. Speaker 3: Correctness. Speaker 2: So this is identifying hallucinations or false information. Speaker 3: For sure. Speaker 2: And correcting it. Talking about Lipton, we've got a great example of Lipton, as you may know, with their bleach tea bags. Oh, yeah. There's a lot of social media talk about bleaching teabags is unhealthy. Speaker 3: Not true at all. Speaker 2: It's actually perfectly safe and like makes your tea have better flavor. That's why obviously Lipton do it, right? But there's, you know, there's kind of these influencers and like Reddit posts and YouTube about, oh, unbleached tea bags. It's like a non thing, right? But if you, you know, then if you start asking Gemini and ChatGPT about like safety of tea, this is kind of leaking into the models. So we would, we put out a lot of content, press releases, blog content about You don't have to even name a specific part of content. You just have to kind of address the topics where you're monitoring. And, you know, you turn the narrative pretty quickly on these AI agents because they prefer going to the brand website over the Reddit and the YouTube. Speaker 3: That is interesting. Do you think that, I mean, do you think that any tea brands, other competitive tea brands are trying to push that narrative because they don't bleach their tea bags? Speaker 2: Well, I don't know. I mean, we don't work with any other tea brands. Speaker 3: No, no, no. But I wonder whether there's, like, smaller, challenging tea brands. I don't know. Yeah, maybe. Speaker 2: I don't know. Speaker 3: They're like, hey, we can do that. Speaker 1: Fly in the ointment, basically pushing the negative sentiment. I mean, could be. Speaker 3: I don't know. I'd be surprised. Speaker 2: I think most do bleach the tea, right? Speaker 3: Yeah, most do. Speaker 1: So your team's handle, it's almost like a PR, right? So you're managing that negative sentiment in the market. Speaker 2: We're not PR. Everything is done by AI and Azoma, but we will have a customer success person who will talk to their PR team and help them to show them the findings in the tool and why it's important. It's pretty self-explanatory once but you Yeah, we you would need to sit down with their team and explain it and because I think we talked about this as well when you're saying you know, you really as I was a tool, you're selling a tool software. Yeah. Speaker 3: Yeah, not necessarily service. You have a protocol and you have You have customer success managers who are helping to make sure that the tool gets used as well as it can be in the organization. But it's up to the organization themselves to actually act on that information and data that you pass to them. Speaker 2: And we, you know, we have a head of agency starting next week who I'm going to introduce you to. So we're going to, we're like, the goal is to try and train our agency partners. We have a few of them now, some like some very big ones and some more like independent ones of like how to Do this service partners like obviously like our team. I think it's pretty good at doing it. But then we yeah, we've been less good at supporting the agency partners on what to do. So we're going to have Britain starting. So yeah, the goal is to also create a lot of content. So the I see some of the other tools do this very well. We've been so focused on like delivering for customers, we've been less good at it to like create content on how, you know, what everything means. Speaker 3: Yeah, yeah. Speaker 1: That's nice. Maybe a nice segue into like product development, how you're using AI. Speaker 2: Let's talk about the last C. Speaker 1: Yeah. Sorry. Yeah. Speaker 2: The most important C. Speaker 1: There we go. Speaker 2: Because all of this has to drive to revenue. Speaker 3: Yeah. Speaker 2: Right. And this is a C that is missing, I think, for many of like our competitor set of AEO, How does this actually, what is this doing? So on Amazon and Walmart, it's very easy to do. As I said, with with ChatGPT and Gemini, it's harder to do, but we are now working with Dunnhumby to get there. So that's. Speaker 3: And that would be in theory for a customer being say, hey, we've made all these changes, like trying to remove all the noise that exists in sales around promotions and app stocks and everything else. If we take all that out, we can show that the activity over here had an uplift. Speaker 2: We want to drive 10 to 100 times ROI for the customer. If, you know, we're charging them 100k, we want to be driving a million in, like, not just revenue, but, like, actual, like, margin for the customers. Speaker 3: So that's the goal. I mean, it's funny, at the large organizations I worked at, you know, whenever there was a new thing that needs to be done, like an activity to be invested in, the question was always, like, what's the ROI? And the rule of thumb numbers, it just needs to be at least three. And so, like, every proposal had. Speaker 2: Guess what? Speaker 3: 300. Because that meant it was going to get signed up. Now, obviously, like, if it wasn't going to, then it would get binned and some would definitely be higher. But three was always the number. Speaker 2: There we go. Speaker 1: Do you, on the 5Cs, do you plot it on like a maturity framework and then work through it step by step with your clients? How does that work in terms of, like, operationalizing it? Speaker 2: I'm an Amazonian, so I kind of see this as a flywheel. And I think it's like a continuous thing. So, like, completeness, you know, your attributes always changing, GMC's turning up, Amazon's launching item highlights, browse nodes and attributes are always changing. Obviously, we discussed the citations always evolving, context is always changing with trends and seasons. Speaker 1: So it's not start here, work your way through it. Speaker 2: Completeness, new stuff crops up all the time we need to monitor for. So you need to do all those continuously and then together it drives the customer acquisition in a flywheel. Speaker 1: And then maybe the stakeholders within the organization that you're typically dealing with, we've kind of touched on it, but PR, web dev, e-com, finance, legal. Right, okay. Speaker 3: But e-com is like your... Speaker 1: That's your entry point. Speaker 2: We say we need an agentic commerce leader. Speaker 3: Sure, yeah. Which I think we'll start to see those roles. Speaker 2: We have. Many of these companies have agentic commerce leaders. Speaker 3: Or are hiring into them at the moment. Speaker 2: I wouldn't say that they have to come from e-commerce. It could be the marketing person who's taking on this agentic commerce mantle. Could be an entirely new person. In some customers, it's like a data kind of person. Speaker 3: Interesting. Speaker 2: So someone, they need to have a... Taskforce, we've all got stakeholders of all of these different organizations they catch up with like bi-weekly or weekly and they need to have executive sponsorship to actually get the taskforce to do stuff to drive revenue for the organization through shopping agents. So like that's the structure that we would recommend for a customer. Speaker 1: It's going to be some people listening to this that are licking their lips thinking, right, let's work our way through the organization and become superstars. Speaker 2: Exactly. Speaker 1: So product development, obviously in the era of AI. You can build super fast. You can launch however many new features, functionality. How are you tackling that with AI? Are you shipping lots weekly or what does that look like? Speaker 2: We have bi-weekly sprint cycles. We have a few long-term things that we're doing and that would be like You probably know my style now. I like easy things to think about for the team. So we've got the three A's, right? Agents. We want to have all the agents. We want to have TikTok. We want to have, we're building it, so Clementine launched last week. Speaker 3: So tracking all the agents. Speaker 2: All the agents. So agents will come. We want to track all the agents. Then the next one is actions. We want to be doing more and more actions as we can for the customers. We've got into Reddit. We want to go into YouTube. Then we have attribution, which we've talked about. And then also we have advertising, which sponsored prompts and and these kind of things which you want to start looking at. So we have that's like the long term vision, sure, of where we're going. But we also like the feature set, like, Should this widget sit here or there on the tool? Is there this dashboard or that dashboard we need to build? We do on a two-week sprint cycle. Basically, every two weeks, we have a call of all the CS. We ask them, what are customers saying? What do they want to see? Why do they want to see it? Then me and my co-founder will sit and listen for an hour. Then we will then have a separate prioritization call, just me and him, where we'll discuss I talk to the senior leaders at all of these companies frequently as well. I want the team to tell us. I don't want to just tell them. Then we will plan on two weeks what to build. We're going to build this new dashboard. We're going to change this. What gets built is decided every two weeks and done every two weeks. Speaker 3: How did you come up with that approach? Have you seen that somewhere before? Speaker 2: I'll tell you what I have seen before. When I was working at Amazon, we used to do 10-year cycles. So one of my jobs at Amazon was I led Amazon grocery in Europe. And we would be planning what supermarkets to place. So this is where we are three-piece. So we'd have three-piece supermarkets selling through Amazon. Speaker 3: Like the Morrisons. Speaker 2: Like Morrisons. Iceland, I think, is there now. Monoprix, Dia. You know, we're across Europe. Tegut in Germany. So this was my job, right? And we would be planning in 10 years time what cities and what supermarkets are we going to launch on this program based on where we think density is going to be. And you know what happened in 10 years? COVID happened, AI happened, and then now Amazon Grocery is shutting all their fresh stores. So the whole thing is a complete waste of time. You cannot predict what's going to happen. And like COVID kind of accelerated everything. So we redid everything to accelerate. And then all the leadership were like, Oh, this is just everyone loves shopping, you know, online, and this is just going to continue. It didn't. So then you had to redo everything to because it kind of fell back a bit again. So it's like it's a waste of everyone's time. So I think two weeks is a lot better to kind of stay on the cutting edge of like actually what's happening, you know, Metamuse launched. Speaker 3: Yeah, for sure. Speaker 2: Exactly. And like that's part of our vision. Agents, we want agents. So if we've got a new agent, we've got to put that on the prioritization list and build it. Right. Speaker 3: And there's been a lot I think said at the moment for like what's a moat with a software company and Velocity could be that moat right now because like it's Relatively easy for people to build, but if you're building at that speed and you have all the insight and knowledge and data, then you should stay ahead of everyone else. Speaker 2: Data is a mode, right? And we have great data based on like, we've been taking action on behalf of customers for three and a half years and tracking the impact of that. So our data is, what is the impact on like revenue on Amazon or Walmart? Or like views to a website on ChatGPT, Gemini, based on an action we've taken. So that's our data, which tunes our actions to be better. And then obviously now we've got Dunnhumby as well, which is 30 years of shopping data. So like data is a real mode because your AI is only as good as your data. But yeah, speed is the only other one, right? Because Everything else I think collapses. Collapses, yeah. Speaker 1: What does the team look like within Azoma? Do you have like big software development teams? Speaker 2: Yeah, we're about 50-50. We're about 40 people. It's about 50-50 split in terms of devs and kind of a commercial team. So we have, you know, of the 20 or so in my commercial team, probably half of that is in customer success because we have Very big enterprise customers. And honestly, in my view, you can only have each customer success manager could probably take two or three maximum. And today they will to do well to do to do a good job with these customers. And bearing in mind, all these customers could be We've got over 200 brands in 200 geographies, right? I mean, arguably, if you look at the VMLs and DOVPs and dentists of the world, they'll have like 10 people per customer. But anyway, I think two or three is probably the maximum and today they probably have five or six each. So I think we could be giving a better customer experience. We're hiring. So if you're listening and you want to work in customer success at an AI native company, Talk to me. We're hiring. Speaker 3: You know, you got the funding. Speaker 2: I had a coffee this before this of someone who so like we would, you know, hopefully by the time we've got two new people starting next week, and so we're continuously hiring for this because I do think we need to Give a better experience and then yeah, I've got a couple salespeople. Speaker 1: Market expansion as well, looking at other markets. Speaker 2: We've been global since day one and we obviously have an office in Toronto and some folks based out of there on the commercial side. We have people in the U.S. My co-founder has now moved to Austin and we've got another one of our team on the commercial side who's joining him and building a hub there. We've got a salesperson there as well. So we have three people in the U.S. All the devs are in Canada, apart from my co-founder. All the devs are in Canada and then the London team as well. Speaker 1: Very cool. Speaker 3: You touched a little bit on use of AI or you were talking about expanding your team with or without it. Tell us about day-to-day use in the company. Are you encouraging everyone to use it? What platforms are you using? Speaker 2: We truly are an AI-native company and I think I'm probably the least AI peeled out of everyone, because I like my days are so busy. And you know, like I'm back to back in customer meetings or not. So I don't have the time to sit back and explore. But like, it's everything you know, like, we've got Just signal tracking across every part of our journey, our head of rev ops is set up. So from like the very, an outbound person all the way through to like existing customers, like we like AI is tracking and giving us slight notifications to support in different functions to do stuff. You know we've got our customer success people who will be vibe coding up what a dashboard could look like to give it to an engineer so it's kind of. You know, like it's from the customer success point of view, which is probably the best point of view. I mean, it's everywhere. Like the one thing I hate is the presentation. So everything's on Canva. So we have a design team. We don't do AI presentations because they look like AI presentations. Speaker 3: They do. They look the same. Speaker 2: And we don't use it to write emails, but like across everything else. Yeah, like, in amazingly imaginative ways that I'm not describing well, we're using AI, you know, across everything. Speaker 1: Is there a model you use across like default across the org? Is it Claude? Speaker 2: I think Claude, you know, basically, we, you know, we, we're a startup, like, whatever, if someone wants to use something, I will, they will use it. And they like, I don't mind. And I actually, I quite like them to explore. But I think Claude has become The best one for work at the moment, right? That's what I would put. That's what I use personally. Speaker 1: We cover this a lot. And obviously, we speak to a lot of like founders and people in the market, like, how do you become AI native? Like, what do you need to do? And we've seen with like Brain Labs, for example, one of the agencies that are using Claude across the organization, implementing skills across different functions. So there's different ways to approach it. We are obviously big on Claude. Speaker 3: For sure. Yeah. And I think it's gonna be really interesting, like the companies like yours and ours have been set up in this moment in time where you could use AI from day one. We're probably going to look very different from the organizations that didn't do that and then implementing it afterwards. Speaker 1: Retrofitting, yeah. Speaker 3: Retrofitting into the organization. Speaker 2: Well, the big problem going back to the enterprises is they're using co-pilot. Speaker 1: Yeah, we've seen that a lot. Speaker 2: These guys will have AI in their personal lives, like you probably with the instinct, you know, like cooking your dinner or whatever, but you can't do that at work because of the permissions for whatever reason. And I think, honestly, some of the reason that some of our competitors have got lots of traction is they're basically enabling these guys to use good AI, just basically like a Gemini or Claude, rather than like the stodgy copilot rubbish that they've got. But actually, they're not providing much value apart from what a base model is, but in a way that an enterprise can. Speaker 3: Tell us, Max, where do you think all of this will be in a couple of years' time? If you go forward in time to maybe Black Friday 2020, Black Friday 2027, what's changed? Speaker 2: I think it moves at a rate that is hard to comprehend for a human mind, like the actual exponential curve. So I do a lot of talks and podcasts, like agent and agent commerce, yeah, 2027. I have Christian who's our salesperson in the US, downloaded Metamuse and bought a toothbrush on Metamuse. Metamuse went on Amazon, purchased through Amazon via Stripe. Delivered to his house. Speaker 1: Wow. Speaker 3: That's what they're trying to stop from happening with. They had a whole lawsuit around that. Speaker 2: Exactly. They lost the lawsuit, right? Speaker 3: Yeah, yeah, yeah. I mean, they're trying to block it. Speaker 2: They're trying to block it. I mean, you can go on LinkedIn and have a look. He kind of documented on LinkedIn. So, like, agent to agent commerce is here now, today. Like, the future is not evenly distributed and we don't have it yet in the UK, but we do have. Speaker 3: Instinct. Speaker 1: Instinct. Big plug for Instinct. Speaker 3: Unblocked or what do I do? Cancelled a subscription this morning with Instinct. Speaker 1: You cancelled? Speaker 3: I cancelled a subscription this morning. Speaker 1: It's my restaurant reservation for me. So it's working end to end. Speaker 2: So I mean, it's going at a rate that is hard to comprehend, right? Like, we didn't think sitting here six months ago, we'd be at agent to agent commerce. We like like AI shopping assistants like Rufus or Walmart Sparky are very widely adopted now and you just listen to the earnings calls of these guys I think. Walmart said that 50% of app users have used Sparky, and when they use it, they have 40% higher conversion rate. That's from last year. Andy Jassy says that 350 million people use Amazon Rufus, and when they use it, they also have 40% higher conversion rates. So this is like, you just listen to an earnings call from Amazon or Walmart, you'll hear this agentic commerce stuff. And agent-to-agent commerce will be with us by the end of the year. I'm going to try Instinct out now you've invited me. Speaker 3: Yeah, please do. Speaker 2: Now I'm in the club. Speaker 1: What do you think websites, like going forward, are they going to become smaller, like less important for brands? What are your thoughts on that? Speaker 2: My view is that there was a, some of our people looking at this space around the same time we were looking at it three and a half years ago, had this theory like, oh, we're going to build a website for AI and a website for humans. And that was one theory that we didn't subscribe to. But some of our competitors, I mean, Matt Scrunch in particular, did and a few others in this space who did. That transpired not to work because like the AI want to, you know, the open AIs and the Geminis want to use a website built for humans in their answers. And they're just not like they're ignoring the LLM.txt robot. They're ignoring all that stuff. Speaker 1: Yeah. Speaker 2: So I think that was a I mean, nobody knew that would be the case. Three years ago, it just happened that that's happened. So I think websites still play a role. Both for humans and agents. Obviously, there's more traffic on the internet now from bots than there are from humans. So they still play a role. Speaker 3: Awesome. And we've covered so many interesting topics today. We're coming to the end of the podcast. So just want to say massive thanks to having Max here today and giving us all the insights about Azoma and the 5Cs and we're going to be back with another in-person one in a couple months time and definitely a Christmas edition as well and we'll see what's happened and changed by then. Hopefully agents will be buying us beers and bringing them into the podcast studio for us. Speaker 1: Tesla robots. Cheers Max. Speaker 3: Thanks very much.