In this presentation, the speaker contrasts two strategic vectors for commerce: the "solve" vector, which focuses on removing friction from existing customer journeys, and the "create" vector, which aims to generate new demand that previously did not exist. While friction reduction can be achieved with or without AI, the speaker emphasizes that emerging AI-driven discovery and the rise of autonomous agents are reshaping how consumers find products. Trust dynamics between agents and their principals (the entities that own the relationship) are highlighted as a critical factor. The talk critiques the traditional sales‑led fashion model—buying large inventories months in advance and discounting unsold stock—and contrasts it with demand‑led approaches exemplified by Zara and SHEIN, which produce based on real‑time demand signals. Building on this, the speaker proposes a "Demand Profile" that captures both a consumer’s upcoming needs (replenishment, seasonal, personal events) and their nuanced preferences, arguing that conventional personas are inadequate. By storing these profiles in dynamic, agentic operating systems, demand can be treated as a tradable asset. The concept of "liquid demand" is introduced: demand packaged with clear commitment levels, allowing consumer agents to negotiate with retailer agents and, collectively, to pool demand on a "Demand Exchange" platform. This flips the current marketplace model—shifting from aggregated supply to aggregated demand—potentially reducing logistics costs, improving personalization, and creating new market mechanisms for buying and selling consumer intent.
Chapters
Vector 1 is you solve for something, and vector 2 is you create something that wasn't there. Now, honestly, most of us are in vector 1, whether we thought about it or not, but we tend to be in vector 1. Vector 2 means creating demand where there wasn't any demand. And that's, of course, tricky, but technology can be an enabler for that. All of a sudden, technology arrives, and that happens once in a while. Some may say the iPhone was such a moment. So every now and then, some technology lands, and all of a sudden, things become possible that weren't possible before. Your bucket, aka your business, is full of holes. It's the entire value chain is leaky. Everything. I mean, literally, there is no single part of your value chain that's not leaking. Whether it's the way you generate traffic, your paid media, whether it's the way your shop works, your app works, web conversion engagement rates, average order value, however you want to call this. Yeah, so keep, keep that friction. That's helpful. But please try to, try to identify the rest of the not so helpful friction, and then get rid of it. So understand what mission your customer's on, what jobs to be done are your customers trying to achieve, and then remove friction out of that process. Has nothing to do with AI necessarily. Maybe it does, maybe it doesn't. Who cares? The endgame is to remove friction on your customer's job to be done mission. Identify, solve. The second vector I find a bit more intriguing, to be honest, and that's what I call the create vector. So that's trying to step into a territory where demand somehow is hidden, but we didn't think about it because, for example, it was technically simply not possible. And of course, What changes from the technology standpoint needs to be reflected in order to become, let's say, appreciative of the options at hand. And things that happen and that simply weren't there. Number one is discovery goes AI. And for those that still look at that zero-something referral rate, happy to talk afterwards why this is a very, very dangerous interpretation of the reality. So there is already a fundamental shift in the discovery, things moving to AI, but at the same time, we now have agents starting to go mainstream. The other thing that I think we need to appreciate is that whenever we discuss agents and principals, and the counterpart to the agent is the principal, so who owns actually who owns the relationship? Yeah? So you're my agent, am I the principal or who? So that discussion is super important because trust obviously comes with the question of who's the principal. So if I want to use your agent but your principal is, I don't know, a tech vendor, is an advertising business, that becomes a bit fishy, doesn't it? So you have vendor agents, And at the moment, everything is more or less in that box. So all the agents that we see are somehow related to a vendor of any kind, of some kind. And of course, you know, people may not bother. Let's see. But were they to think about it, they would realize it is a bit fishy. I mean, the broader the assortment, the broader the offering, the less fishy it gets. But ultimately, you know, you're talking to Nike and asking, you know, what are the best running shoes? Well, what would the principal Nike tell you? So that Nike agent most likely is not the most neutral source. Maybe. So when we talk about what to create, we as a business, for how long we can possibly think about our business, we've been in the business of pushing inventory. It's maybe a very harsh way of summarizing it, but we have been in what I call sales-led commerce. So we buy 12 months in advance when you look at fashion, or even more, and then we get it wrong, and then we push it to the market. When it doesn't sell, we put discount on top of it. And that's basically how we roll, right? Zara, SHEIN— I know it's under discussion, but 2 very famous concepts that start at the opposite end of it. And I call this the demand-led commerce, where they try to understand what demand is in the market and they try to fulfill that demand short notice. So production cycles limited from months in advance to weeks in advance. So they listen and learn, produce, test, and the replenishment is the bigger order than the initial order. In normal business, it's the other way around. Normal fashion business, Big order is initial, small order is replenishment. They turn it by 180 degrees. Now I would say we're in the create phase, so bear with me, don't throw problems at it. But I'm just saying, now let's add one dimension to it. Let's not just work on demand-led sales, let's trade demand. Imagine someone would know when you will need something even before you start to search. Imagine, imagine someone would know, imagine someone would know what you're planning or intending to buy before you even type in a search query. Just imagine. And now imagine that same someone would even know your preference, your taste. The first part of that magic is what I call the Demand Profile. So you guys that run technology, you solve this, yeah? I'm just describing what we need. So we need a demand profile. We all are, you know, we sit on different pockets of demand. Some stuff we just need to replenish: toilet paper, paper tissues, you name it. It's a long list of stuff we just need in a replenishment mode. No emotions, no nothing, you know, cheap, cheerful. Delivery time more important than, you know, who's the retailer. The brand of the product mostly, mainly matters more than, you know, the retailer that sells it to you. And then we have things that are easy to anticipate. You have kids, well, there's back to school. You know, a new sports season, well, there you go. So you have certain things that are not difficult to guess. There's Christmas, Mother's Day, Father's Day, Easter, blah blah. That's easy to anticipate. And in most cases, that somehow is a moment Where commerce happens. And then you have things that are even plannable. So it's your wedding, it's your grandmother's birthday, it's whatever it is in your personal life that's plannable. And then the second layer to this is our preference. And of course, bad news is we've been talking about personas for the last 100 years, right? Which is, if you think about it, a bit funny, huh? It's a bit funny. Look at yourself Just for a moment, and you will have to confess, and I call it a distribution of daily contradictions. We have to confess or appreciate that we're nothing like a persona. We're completely schizophrenic, multi-personal. And just within a day, it does make a big difference whether you search for something 8 o'clock on the bus. You're very more into, I need it quick and I just want a solution. My sock has a hole, I need a new sock, and I want it in an hour. I don't care about the price, I want it in an hour, and I want it black. So just solve that for me. I don't need inspirational photography, I don't want an Insta story, I don't want a video. No, no, no, I just need the bloody sock, but it needs to be quick. Whilst when I'm at the same day, I don't know, in the evening, and I'm on my sofa, and I think about cashmere socks, It's a completely different ballgame. I want the whole monty. I want Kashmir flowers flying through my window. I want, you know, smell and sound, and I want the butterflies, and I want the whole Kashmir. I want the cow— oh no, it's a goat, sorry, a different animal. So I want that animal to look at me, right? Smile at me. And I want that goat to give me my socks. It's the same person a couple of hours later with a glass of red wine, and everything changes. And one minute you want the cheapest stuff, and the next minute you want the best quality possible. And it's all us, it's all us in the same day. And now you come along with your marketing folks and you have a persona. Well, good luck with that. Good luck with that. So imagine that we would capture your personal, our personal preference as well. So demand and preference. Just imagine we would be able to capture that. And then we put it somewhere where it's not an Excel file, where it's not a Klaviyo post-purchase campaign, where it's not, I don't know, a Salesforce file. Now just imagine we think more in agentic operating systems. So we think more in filing systems that are dynamic, that allow intelligence layers to do something. Just think in that direction. And you can capture demand and you can capture taste and you put it somewhere where intelligence can happen, not static, and then you can package it up. And I call this demand as an asset. So all of a sudden, demand becomes an asset. Can you imagine? Demand packaged up becomes an asset. And it's an asset you can start to trade. You can start to trade the individual asset of that one consumer, but you can even start to pool it and you can start to trade pooled demand. Let that sink in. So you have different levels of commitment you can trade: hardcore committed, You can then, once you understood the commitment, start to negotiate with merchant agents, with retailers. You can start to negotiate— your agent, your consumer agent can start to negotiate the demand it knows on a commitment level that's reliable, can start to negotiate that, and then can pool the demand and negotiate even further. I call it liquid demand because for the first time ever, we would package up demand with commitment levels that are clear, Based upon a demand profile and based upon a preference profile that is true, not static, and we can package it up and therefore it's liquid demand that we can now take to market. And I built a quick mock-up. This is how that could work. It's animated. Can you see it? So on the left, Juno, that's our agent. So Juno now gets the commitment levels and it's different levels You see firm, firm, flexible date. So imagine that you're flexible on the date because you don't care. School is in 3 months, I just need the new rucksack in 2 months-ish. So the retailer that provides you with that rucksack is very flexible in choosing logistics solutions. So all of a sudden they're not forced into next-day delivery all the time because they start to understand you don't need it. You don't need it tomorrow. So why pay that extra logistics? No, you need it in a month. Well, fine, let's agree. So it needs a decent commitment, but let's agree you get it in a month. The retailer has a saving on the logistics, can pass on the saving to you, for example. And of course, there is a next level to liquid demand even. Now imagine that we bring more of those consumer agents together. Just imagine that they would gang up, they would team up. So you have an army of demand packaged liquid demand agents, and they would all of a sudden meet up in what I now call— not a trademark— the Demand Exchange. So demand agents meet supply agents, and then they can trade based upon committed demand, based upon your taste profiles. This is nothing short of the exact 180-degree pivot to what we call marketplaces today. So today it's an aggregated supply base. Marketplaces, the marketplaces today is a supply gathering. And what I'm saying is once you think through how that demand playbook could actually work, you turn it upside down and it becomes an aggregated demand destination.
