Zenline AI and the Retail Margin War: Why Assortment Velocity Wins

European retail is in a margin war. It didn't choose. Assortments move faster than teams can analyze data quality collapses on the scale and global platforms like Amazon and Temu operate on a level of speed that exposes every structural weakness. Abba SETI built Zenline AI to close that gap.

Agents that act not observe systems that detect margin leaks, resolve clutter and surface opportunities humans never see. Today we examine why autonomous decision agents are becoming to operate the system of modern retail. And what happens when a retailer shifts from dashboard analyzers to product level action in seconds. By the end, you'll understand why the next competitive edge isn't scale its velocity.

Welcome to StartUpLead IO, your podcast and YouTube blog covering the German startup scene with news interviews and live events. Hello and welcome everybody to Startup Rio. Today my guest is Arba Seti, the CEO and co founder of Zenline AI from Switzerland, Goodsy Goody, a company that is redefining how European retailers decide what to sell, what to move and where the real margin sits. With experience at BCG, Procter Gamble, ETH Zurich and Research work at Cambridge, he brings both the strategic clarity and the technical depth required to rebuild assortment logic.

From first principles. Zenline agents clean messy catalog data, detect cannibalization, map substitutes and reveal trends before they become mainstream. They already process more than a million SKUs and deliver recommendations with measurable financial impact enable retailers to operate with the speed of global platforms without the overhead. Aber is here because his work signals a structural change.

Decisions that took once weeks now happen in minutes and the retailers who adapt will redefine competitiveness in Europe's Euro trillion market. That was an intro, don't you think? Welcome to the show and let us get right into it. You said the real power center of retail isn't price or promotion, but but the assortment itself.

When did that insight crystallize for you? In my previous job with bcg, we were advising retailers with assortment optimization. And what becomes very obvious when you work with retailers on the assortment is that the assortment touches the entire organization. So it's not only the top line or the bottom line as is for pricing and for promotion, but certainly it also touches the top line.

So if a customer, for example, doesn't find a product that it wants in the stor stores or on the shelves, that's lost revenue. However. Moreover, it really defines your brand. So the brand is typically not only shaped by the marketing, but it's mostly shaped by the products you offer.

So you can imagine when a consumer has a demand, what do they first Think of what's the brand, what's the retailer that they think of to go and buy the product from? And that's so important to be top of mind for the consumers. And, and that only works with an assortment. That consistency delivers what the consumers actually want.

And then the last one is the assortment. And the depth of the, and the breadth of the assortment influences the complexity of your operations, the logistics, the supply chain, the inventory management. So the broader the assortment, the higher the complexity and the more difficult it is to maintain lean operations. When you talk about assortment, you have something in mind.

For example, when I was thinking about this interview, what I had in mind when I was reading this, at first I was thinking, okay, option one is I look for the basics. That is a retailer with a yellow logo. Then I look for some more fancy stuff. Then that comes in a retailer with the orange logo or a red logo.

Is that something that you're thinking about here? Yes, exactly, that's exactly it. What comes to your mind when you want to buy a product, what is it? Is it someone who's super, super trendy and always knows what the consumers want and can offer it, offers it fast, or is it typically 6 to 12 months behind the schedule?

Your early discovery work exposed how little transparency retailers had into their own margin drivers, what revealed the debt of that blind spot. So the importance of that blind spot comes basically what you said in your intro. It's a super, super, super competitive market with Temu and Shein entering the European market not only in fashion, which a lot of people think, but it's across all categories from electronics to fashion to pet supplies. And retailers are too slow because of the analysis, just take too long, which should be done automatically and systematically with AI.

So basically we're entering this competitive, basically war in retail coming from different countries from different angles. And if retailers don't move fast enough, don't filter out products that have bad margins and focus on the ones, focus on the entire assortment. Instead of managing just 10 to 20%, European retailers will not be able to cope with the speed and with the velocity of these big platform players. So that you're mostly targeting this towards retailers that would compete with Amazon, Temu, Shine, Alibaba Express and so on and so forth.

Yes, we're trying to basically help them to have the same data, science and AI capabilities as these platforms have, who have actually their own internal assortment teams to make these decisions. And you're now entering a market that is basically shaped by Amazon level speed. You order one day, you get it, the Next day or a little bit later. What convinced you that Europe's response had to be agentic and not incremental?

So what's happening right now is a perfect storm. We're seeing years of inflation, declining consumer confidence, increased saving rates and what's making it now very hard for retailers to stay ahead the curve and stay profitable. And this does not only apply to retail, right? We're speaking about all industries that are basically undergoing this wave of the industrial revolution or let's say revolution led by general purpose technologies.

And in these times of general purpose technology is changing not only one but multitude of industries, the answer has never been incremental innovation. And US and China will move ahead and they will basically equip millions of white collar workers with agentic AI. And speed will be the differentiating factor, launching new products, understanding the assortment, understanding relationships in the assortments much, much, much deeper, much earlier than all competitors will. And again just to to basically highlight that if we still think we can do the same, the same thing that we did for the past 20 years, which is incrementally increase efficiency, these companies will basically frog leap the European retail industry.

In the entrepreneur's world, we'll be talked about the moment that data forced a decision you didn't expect. Stay tuned and learn more down here in the link. We've been talking category teams already that some of our large competitors have. They focus on top sellers and ignore the long tail.

What breaks inside an organization when the long tail suddenly becomes visible through good data, through AI, real time observations? I would say the first reaction is inconvenience. There's a lot of things that get exposed, like data flaws, wrong images linked to wrong products which then lead to very high returns because consumers got the wrong product which they actually didn't want to order. But the product image on the website was leading to the wrong product attributes that are misleading or misconfigured or inconsistent pricing and all of these errors, or let's say inconsistencies, misconfigurations in the assortment become very obvious when you start focusing not only on the 1015 to 20% of the assortment that is typically managed very well, but then surfaces and this then creates, let's say inconvenience at the beginning.

But organizations then learn to accept the fact that it was very, very hard to manually be able to manage hundreds of thousands, if not millions of SKUs. But now with the power of AI, you can actually do what retail wasn't able to do for the past. Basically for the past decades. You'Re the agents you provide for the companies, they do challenge internal narratives.

Which narrative collapses the fastest once real data pressure was applied? You said first the flaws are exposed and then I think the next step is their internal narratives breaking down. Yeah, very, very, very good question. And I would say the most common narrative is the intern.

The most common internal narrative is that the consumers know exactly what they want to buy. And this consequentially leads to an increased breadth of the assortment. And we as read retailers, we don't have power to because the consumers have all the power and they will either way just buy what they want. If you don't have it, they will not buy.

However, data that very, very early on will show that curated assortments is actually what consumers want. They don't want decision paralysis, but what they want is a curated assortment. Exactly. Curated to their demands and their basically the consumer behavior.

And what you typically see immediately is that revenue and margin goes up significantly. We're speaking 1, 2, 3 percentage points of margin when the retailer makes the decision of what should be in the store, in the stores instead of basically giving up and saying it's always, it's the consumer who's always right and we need to offer everything that they can possibly imagine. Part of your thesis or your main part is speed. What's the cost of operating a full year behind the real markets decision Tempo.

I just mentioned it. We're speaking 1, 2 percentage points in margin improvement and we're speaking 1 to 3% in revenue increase. Very often we're not even one year behind the real market, especially in the private label segment of the businesses. And in Switzerland, I think for the first time, private label revenues surpassed 50% of the market.

So private label is now making more revenue than the branded market for retailers. And in Germany, in the rest of Europe, I think it's close to between 30 to 50%. And there we're speaking very often about 18 to 24 months of let's say distance to real consumer, consumer changes. I do have in mind that basically we talked about here the less fancy products, especially talked about food here and clothing.

There's a big online retailer who offers essentials and, and that's kind of stuff we talked about here, that that's retail brands, that's the brands they come up with themselves and for some larger specialty food retailer, they are meant by, they are produced by household names. So that' so pretty interesting area to watch. But let us go a little bit further and talk about what impact the data had on teams because you've Watched messy data, paralyzed teams. What's the silent limiter they underestimate most?

So I would like to answer slightly different question which is, and I'll get back to yours, which is companies very often think that data overall is the limiter of, for the generation of the value margin, improvement, revenue increases, making faster decisions, and then they very often focus on the 5 to 10% of the data that is imperfect, where the master data is not clean, where the attribute data is not clean. And then organizations often stagnate and move very, very, very slow because they think of, oh, data cleaning being this one incredibly big IT exercise that they need to perform.

However, what should happen is that while doing these use cases, you clean the data as you need them because you will figure out where the data leads to. Wrong suggestions, wrong recommendations, and then it makes sense. And then there's also a business case to clean up the data and make it worth this exercise. And what a lot of companies and retailers have not realized yet is that agentic AI is making data cleaning and insight generation so feasible.

The cost to or the cost and the time to get cleaner data is reduced on a daily basis and retailers can already save millions in margin without having the perfect data. So instead of focusing too much on the limiter, I think companies do that either way all the time. The message should be although data can be messy and too much data can paralyze, the limiter is very often the mindset towards the data. When we talk about agents, many people have absolute neutral AI in mind.

But every agent encodes a judgment, either by the person who decides what to encode or the person who does the encoding. Which judgment was the most difficult to translate into an autonomous logic? Very good question. And I have to say the entire space is very, very, very early and it will change in a year from now.

But what is happening right now is that agents are very, very, very sensitive to a few sentences in the prompt in how they make the final decision. So because of that, as you were saying, rightfully, the design team of the agent, of the system prompt of the context engineering that allows the agent to make a decision of which products to recommend. One very simple example might be that one of our customers doesn't want to be the most trendy firm. So they don't want to just launch products that are, let's say, that are just coming to the market, but they want some validation, some proof points, and to know that this product is actually performing and will be here two, three, four years from now.

And now the question is, what is A trendy product, when does the trend turn into something more sustainable? And in the end, the makers, which we are here with Zenline, need to be very, very, very thoughtful of how do we encode that? And I think the most difficult part of this is the memory layer. So you do one recommendation and then the question is next week, next month, next quarter, how do you encode all the recommendations that you made?

How do you encode, let's say, the feedback from the real world? Did the product launchers succeed or didn't they succeed? How do you bring that in and how do you feedback that into something that the agent can use to then be able to make better recommendations next time? Because this is something that is incredibly natural to us as humans.

We do something, we, we observe what happens and we have an initial gut feeling of whether we should do that again or what we need to adjust. And this is something that's very hard to, to encode into. Into autonomous systems. Yes, there's a lot of stuff to really hard to encode in autonomous systems.

And I think the more systems we'll have, the more it will be of importance. But let us go to the next question, because I was wondering, your launch validator runs on social signals, competitive shifts and emerging demand. When did you trust its signals over the human instinct? So what's important to distinguish is whenever you launch a product, which is what we aim to support with the launch validator is we want to strengthen the human instincts by giving it more data.

However, as every single human being, category managers, for example, in the fashion world, are biased by what they see on their channels, their social media channels, by walking through the city, watching colleagues, friends and so on, what they wear, and then using hundreds of thousands of data points actually allows them to double check your gut feelings and use these data points from social media, from Google shopping, from competitors, bestseller data, to see whether your initial gut feeling was right or whether you were missing something.

And it can be much, much, much more nuanced, especially in the fashion world, where new designs from quarter to quarter, from year to year very often change in subtle ways which are very hard to grasp. So it's not trusting the signals over human instinct, but it is challenging the human instinct, which can be biased by their own world, which is very often not representative of, let's say, the consumers, what they want in certain categories. Always, always when it's working in consulting and always if you do have strategy project at one point, there's always the question of internal cannibalization.

So if the, if I'M adding product A, does it take away from B, does it take away from C? And your consolidation agent exposes internal cannibalization. What's the most counterintuitive callization call that surfaced? There's always revenue cannibalization and there's margin cannibalization.

And the revenue cannibalization is basically where a lower priced product is purchased by the consumer because although there is a duplicate which is maybe higher priced and is the one product that you as a retailer would want the consumer to to purchase. And the issue here is that revenue cannibalization is very often not counterintuitive. It's very intuitive. However, it's never in the data.

I haven't seen a single retailer that has good relationship model across tens of thousands, hundreds of thousands of products to know which products are very similar because you would need a very good relationship mapping of all products, similarity scores, quite good attribute data to figure out which products are closely related to each other and what is the impact of one to the other. However, margin cannibalization is very often counterintuitive. You take for example two examples that we had are drinks where branded products that are higher priced had significantly worse margins than the product than the private label brands that are cheaper and have higher margins.

So in these situations what happens very often is that companies do not understand sufficiently the impact that let's say price differences and then the relative margin of these products has on in the end the purchase decision of the consumer and then incentivize the consumer to purchase the wrong product by for example, reducing prices by watching the market too closely or trying to increase velocity of the product. We'll be back after a short ad break. And after the break we step into the strategic misread that holds retailers back and where the next advantage isn't data but decisiveness.

Hey guys, welcome back to our interview with Arbor from Zenline. Helping retailers to increase revenue and margins. Working with AI here and I was wondering, your weekly decision cycles replace seasonal ones. What operational structure fails first when a retailer attempts that transition?

And also a little bit tongue in cheek question, when are we at daily decision cycles? Very good question. I have to say we have spoken to many retailers and we're working with many retailers where the decision cycles can sometimes take more than 12 months. We were seeing very good examples and we're seeing bad ones.

The operational structure that fails very often is the relationship, especially in the private label business to the suppliers where it can take up to 12 to 18 months to actually supply the product. So from let's say initial ideation to concept development to iteration to receiving the stock that's required to sell the product, sometimes it takes between 12 to 18 months, which in the end makes AI driven quick decisions completely obsolete because you don't need it if your supply chain process in the end can't deliver what you're trying to do organizationally in terms of commercial output, delivering the assortment that consumers want.

However, with many of our retail customers, we're also seeing very good examples where the operational architecture actually is already in place. But the bottleneck was never the supply chain delivery but, but it was the analysis and the inside gathering because it's bottlenecked at the category managers. And if the category managers simply can output more decisions, the let's say supply chain behind it is already supportive enough, is already, let's say established so that they can quickly start to see, see these products being in the shelves within the weeks.

On the question when are we on daily decisions? I think daily decisions will not be the bottleneck, but daily supply chains will be very hard to achieve for retailers in most industries. If we wouldn't be on a time constraint, I would go into deviate a little bit and go into some thoughts, experiments, what we could do with Jones here. But let's take our main line of interview questions here.

Rejected recommendations reveal a lot about the organizational psychology. What patterns do you see in teams that aren't ready for agentic systems? I would say organizations that focus too much on the errors that happen, or let's say the parts of the recommendations that are misleading, that are error prone, that are just simply not useful at the beginning. And organizations that focus too much on what's not working and where the data is not sufficiently good are the ones that will become the laggards of the, of the decade.

So however, on the other side, the teams that are utilizing agentic systems most effectively basically treat them as junior analysts, I would say. So you have a junior analyst. What happens if you receive a lot of input that will be good, some will be exceptionally good and some will be bad simply because it misses context, it doesn't know exactly what has happened in the last 12 months of the organization. Sometimes recommendations are bad because they have been done, they have been tried in the past, maybe 12 months ago, 24 months ago.

And certain products, certain product types or categories just don't work with the retailer. And a junior analyst also wouldn't know them. So what very good, what patterns of very good teams are is the willingness to give feedback, give nuanced, and detailed feedback to advance the memory of the system and give more context into the system, define the strategy, the company strategy, the category strategy, and thus be able then to improve it over time again, as you would do with the junior analyst.

And I think the more organizations start to think of agentic systems of colleagues, of basically just another part of the workforce, the faster they will be the ones that actually gain margin benefits from it. Your pricing agents benchmarks internal logic against the market reality. What's the structural pricing error that retailers report? Okay, so I haven't seen a single retailer yet who has been able to be very strong in utilizing internal relationships in their pricing.

So what happens very often is that many retailers are very much focused on hey, how is this one single product priced compared to the market to other other retailers selling this one product? What every retailer does is they compare their own assortment very much with the assortment of other retailers. And what they neglect is that a lot of the pricing, a lot of the pricing power that they have is not comparing themselves to competitors, but to the internal assortment. Because very often when a consumer is already in the stores, they will make a decision not based on what is the price of the competitor, but they will make a price based on what are the pro, the other products, the alternatives, the substitutes that are already in the shelf.

And especially with regards to decisions that impede the margins, customers just retailers typically don't have a, let's say, sophisticated enough system to understand the relationships. As you mentioned before, the cannibalization impact of not only delisting a product, but also changing the price of one product or what does it do to all the other products in the assortment. And this relationship model is something that retailers should think about. However, they're doing it way too little because it's very complex technically.

So the classical substitute products substitute variant. Pricing as soon as you get to 10,000, let's say 10,000 products. It's very hard to manage a very clean relationship mapping of your products so that for example, when you change one product, you also change either in, let's say proportionally or at the exact same level. The price of variance can also be the same for substitutes to just understand what does the change of one product's price due to the rest of the, let's say, very similar, very related products.

We've been talking about margins a lot and leaders often freeze under margin pressure. What differentiates those who accelerate versus those who stall? I think a lot is in the tools. Some retailers still consider power bi as state of the art and A lot of them still stick with Excels, which is very convenient because they don't have to learn new tools.

But the spreadsheet logic is failing. It's failing as soon as you have hundreds of products, not even mentioning thousands or millions of products. And I think retailers just need to accept the fact that let's say human driven and spreadsheets driven working modes are not optimized to support retailers where decisions are incredibly complex. And yeah, in the end, competition will not wait.

Competition from other markets, new more digital versions of retailers popping up. And the margin winners will be the ones that rewire how fast product decisions can be made because they have the data ready, readily available, and then they have the systems that output recommendations and then review the recommendations. So basically the shift from going I need to do the spreadsheet analysis myself to I only review the output of systems and I can then fine tune for a few hours instead of being the one who has done the analysis for weeks.

I know how this feels. You gotta be creative here. You guys scaled at a pace even few teams attempt which moment demanded your highest level of personal, personal resilience. Honestly, there, there are many.

As a founder, you constantly are on the edge, but when you have a strong team, you, you can go through very, very, very long hours together. And I think that's the happiest and the best parts. But for me it's a combination of several countless sleepless nights and pressure to keep customers happy. Yeah, but I would say whenever the team is willing to go the extra mile, is willing to go beyond, to basically be able to deliver the results that we are promising with the agents, that's the moments where let's say, well, that's our source of energy.

Your source of energy. Okay, I get the feeling that Asian driven operations are here compressing time, which is just talked about quarterly to weekly in the interview. Which legacy workflow becomes obsolete here? First.

White collar tasks that require desk research. All the research work will be automated in the future. And you can see this already in consulting, which they have huge pressures. It's the same for lawyers, it's the same for any kind of research assistants.

Because I don't know if you have used deep research from Gemini and ChatGPT, but it's incredible how much depth and how much data can be processed with these tools. And they can do it significantly better than humans can. So everything related to research and analysis will be something that will be obsolete in the future. And humans will switch from doing the analysis to making the decisions.

Basically exactly what we discussed before with the leaders of margin drivers and compressing months of analysis into days, if not sometimes even hours. Making fast decisions, making counterintuitive decisions makes explainability become something strategic. How do you maintain trust when high impact recommendation contradicts their intuition? I like that question, especially in an In a machine learning driven world that we had for the past, let's say between 2016 to 2022, explainability was really hard because you had a result.

Your machine learning model told you this should be the new price, but there was very little explainability behind it. What is now interesting with generative AI is that the models have reasoning tokens and they can output and they can explain why they came to a decision. And this reasoning is getting closer and closer to the reality of why the model decided to do what it decided to do. And this is actually one of the things that we have realized with customers is they love to see the thinking of the model.

So not only the final output of why you did it, but even the steps, the thinking steps that happened, the tool calls that happened, the web search, which websites did it browse, all of that to make it very very very transparent. So I think in the last 10 years there was a bit of the dogma to hide, a bit of the complexity behind it, to not confuse especially the managers, but also the category managers and also the executive teams. But now with LLMs you can be very very very transparent about why the model did what it did.

And very often it's you can find immediately in the prompt why it did the decisions the way it did and then you can together with the customers define the prompt to make it basically just be more effective for their categories and their company. At the time of recording and admittedly we were recording this in early December, Europe has a 180 day window before global platforms widened the gap. What is the one move retailers must make now? I might be a bit biased but it's the super obvious one.

Skip all this digitalization work. Just everything related to pure IT transformation needs to be done in parallel to all the agentic more business related work that needs to happen immediately. No retailer has the time now to wait for months and months and months if not years to do IT transformation SAP for Hana Transformations. But at the same time they should be already working in the business.

Not specifically with it, but the businesses need to start working on agent based systems and see what's already possible with them. Play around with it, find very very highly valuable use cases that will make an impact on the margin within four to eight weeks. And I would say if retailers, especially in Europe and the US just feel more confident to do these things, they will see how advanced retailers are. And that's one of the things that we hear very often.

Typically the response is after a month. They say, I would have expected AI to be three to four years away from what you showed us. And I think this is where you need to get the retailers to want to work more with it. I've heard over and over that focus determines survivability, especially for you guys.

What, which idea did you kill that unlocked your current trajectory? In the beginning of Zinlin, we were focusing on the. On a very similar problem, on the assortment problem, but for manufacturers which have very complex portfolios, sometimes hundreds of thousands of SKUs. And as soon as we started talking to retailers who are the buyers of those, let's say, complex assortment, complex portfolios, we realized that the interest and the traction that we received was.

It was instantly obvious. And we basically pivoted our entire company within three hours to this new idea because of the traction that we received there. And I think being able to take very, very strong decisions and let's say, reshifting the organization's focus, not only individual focus, but the focus of the entire organization on a new topic, new type of customers, and then very narrow scope of, let's say, features. This is what makes the startup incredibly powerful.

And to be very honest, it's not about killing ideas very often, it's just about delaying them. You know that customers ask for something, you know that customers are willing to pay for something, but you just don't have the capacity to serve them in the quality that you want to show to them. So basically you just, you have a roadmap and you put it off until the features that you're building are brilliant and you can scale them first before adding the new ones. I just want to ask you this again because I'm not sure I heard it correctly.

You pivoted once with a decision within three hours. Yes. I think that's the future of business. That actually ties in into my last official question.

Your product velocity is unusually high. What principle anchors precision without slowing execution? The two things that we do brilliantly, and it's also related to the one before that is very, very, very high focus on the most important features, only the ones that are really meaningful to the customer. And then we don't obsess with technical details.

We don't obsess with the beauty of the code. We obsess with the customer value and we Every single day we question ourselves, is this the most impactful thing we can do that really saves margin for the user? And we have it on our boards, we have it in our notions. The one key metric that we track is did we improve the margin of our customers?

And in the end, if we don't do that, we're building a bad product. So this is what has helped us to keep the product velocity very, very high. I see. That was the official questions and we usually end with our two standard questions.

Are you open to talk to new investors? We are currently well funded, so we are not talking to investors, but we are aiming to speak to investors again beginning summer 2026. Are you currently looking for talented employees? Yes, we are.

We are looking for the best and most ambitious and boldest engineers. We have two, we're thinking in two teams. At Zenline, we have product engineers which are close to the customer, but in the end they're responsible for building features and building the product. And then we have forward deployed engineers, which is a term coined by Palantir, which is engineers who are managing the customer relationship.

So in the end, Zenline is built purely by engineers. However, we then distinguish between the ones that own the product and the ones that own the customer relationship. Albert, you've shown that the future of retail belongs to operators who move faster, see deeper and trust systems built for action rather than observation. You clarified where margin heights, where complexity wastes energy, and why agentic decision redefines competitiveness.

For those who want to follow your work or understand Zenline's impact more directly, Visit try Zenline AI or connect with you on LinkedIn. Thank you for bringing structural clarity to sector that really needs it. Thank you so much. That's all for Find more news, streams, events and interviews@www.

startuprad.IO. remember, sharing is car.