Latest / IT Infrastructure as a Conversation / Syndigo on Why AI Commerce Is Failing Without Better Product Data
Transcript
- 0:00So a huge thanks to Denodo for supporting the
- 0:03Tech Talks Network, helping us produce more than
- 0:0560 interviews a month. And when it comes to trusted
- 0:08data products, it all starts with the right foundation.
- 0:13And trusted data products start with Denodo because
- 0:16they can help you create, manage, and deliver
- 0:19business -ready data products faster. with secure
- 0:23real -time access across all of your data sources.
- 0:27And you can learn more by simply visiting denodo
- 0:29.com. What happens when the product data behind
- 0:34every online purchase becomes as important as
- 0:38the product itself? Well, today I'm joined by
- 0:42the Chief Product Officer at Sindigo. We're going
- 0:46to have a conversation that takes us behind the
- 0:49scenes of modern commerce. and into the infrastructure
- 0:53that most consumers never see. Because my guest
- 0:57leads Sindigo's product organisation and its
- 1:00product experience management business line.
- 1:03And he's someone that works with thousands of
- 1:05brands, retailers and distributors that all depend
- 1:08on accurate product information, so they can
- 1:12sell across physical stores, websites, marketplaces,
- 1:16app and now AI -powered shopping experiences.
- 1:19But what we don't talk about enough is the infrastructure
- 1:22that makes all this possible and why data has
- 1:26moved from being just content on a page to becoming
- 1:29commerce infrastructure. Because if a customer
- 1:33is asking a chat bot for the best running show
- 1:35for knee pain or using AI shopping agents to
- 1:39compare products, the brands with incomplete,
- 1:42inconsistent or outdated product data, they could
- 1:46quickly disappear from the answer. And my guest
- 1:49will explain today why many companies built systems
- 1:52for e -commerce search, SEO, product detail pages,
- 1:56but they're now discovering that AI driven commerce
- 2:00needs far deeper product attributes, stronger
- 2:03governance, richer metadata. and better handoffs
- 2:07between teams. So I want to learn more about
- 2:09the hidden complexity behind a simple product
- 2:12listing, whether it be from ERP and PLM systems
- 2:15to supply chain, marketing, ratings, reviews,
- 2:19global sizing, regional language differences
- 2:22and syndication across retailers, the list is
- 2:25long. But my guest will share how Sindigo's recent
- 2:29acquisition of One World Sync will also help
- 2:32change the conversation around product data standards,
- 2:35ratings reviews and global commerce infrastructure.
- 2:39So as AI changes how people discover and buy
- 2:43products, is your brand ready for a world where
- 2:46the shelf is no longer a web page, but a conversation?
- 2:50And on that note, it's now time for me to officially
- 2:54introduce you to today's guest. So a massive
- 2:58warm welcome to the show. Thank you for joining
- 3:01me today, for everyone listening. Can you tell
- 3:03them a little about who you are and what you
- 3:05do? Sure. So my name is Tarun Chandrasekhar and
- 3:09I lead the product organization and the PXM business
- 3:14line for Cyndigo. PXM stands for product experience
- 3:18management and I'll drive into what that really
- 3:20translates into for customers. But Sindigo is
- 3:25a software company. We work with about 15 ,000
- 3:29brands and 3 ,500 retailers and distributors.
- 3:32And the best way I might describe this is that
- 3:34we provide the infrastructure that ensures that
- 3:37product data gets to the right place. So if you're
- 3:40a brand and you're building products, physical
- 3:42products, you want to make sure that everything
- 3:44about the product is accurate, whether you're
- 3:47selling it in store, your own stores, whether
- 3:49you're selling it in an Amazon, Walmart, Target,
- 3:52on your own dot com website in a third party
- 3:55marketplace because if that information is not
- 3:57accurate either your returns go up or you get
- 4:00sued because your allergy information is off
- 4:02you want to make sure that to reduce the friction
- 4:06in that sale the product information stays accurate
- 4:09that's product experience management at the most
- 4:11basic level and we sell software that allows
- 4:15you to optimize all elements of that and Two
- 4:19variables we tend to help optimize. The way I
- 4:22describe this is speed to market. What is the
- 4:25fastest way I can get a skew to a shelf? Because
- 4:29if nobody is making money until that skew hits
- 4:32the shelf, then it gets to a store. The brands
- 4:34don't make money, the distributors don't make
- 4:36money, retailers don't make money. So how do
- 4:38we optimize that? And time is the primary element
- 4:41there. And the second is once it gets to a store
- 4:44shelf, physical or digital or increasingly agentic,
- 4:48How much more money can I make? How many more
- 4:50eyeballs can I get on that product? How can I
- 4:53build the trust about the product? So those are
- 4:55the different product lines that we have. I love
- 4:58that. And of course, when we're talking about
- 5:00this world, AI is absolutely everywhere at the
- 5:03moment in e -commerce, whether it be recommendations
- 5:07and search to automated product descriptions
- 5:09and even conversational shopping assistance or
- 5:13concierges. How much of this current AI challenge
- 5:17that people in e -commerce are coming across,
- 5:19how much of that actually comes down to poor
- 5:21product data rather than weak AI models? Because
- 5:25I think very often we focus on the AI models,
- 5:27but one phrase I hear a lot at tech conferences,
- 5:30no data, no AI. And it almost reminds me of the
- 5:34age -old saying of garbage in, garbage out. But
- 5:37what are you seeing here and how much of those
- 5:39challenges come down to poor product data? I
- 5:42think the failure mode is almost always the same.
- 5:46Companies invest in the model, but they skip
- 5:48the foundation. You cannot build reliable AI
- 5:52applications on top of messy, inconsistent, incomplete
- 5:54data. The real hardship that customers are facing,
- 5:59companies are facing today is this. Especially
- 6:01in the commerce space, in the last 20 years,
- 6:05we used the page ranking algorithm and SEO, search
- 6:09engine optimization, as a work around to bad
- 6:13data. Because what we said was, hey, I know that
- 6:16I don't know whether this is a single record
- 6:19or multiple records or if this is unique or not.
- 6:21And I don't know about the governance and the
- 6:23provenance of this data. But I do know that these
- 6:2625 attributes really matter to whether you're
- 6:29going to show up on page number one or page number
- 6:31seven on Google. So I will invest and get that
- 6:34right, which was great. when you're surfacing
- 6:37this product on an e -commerce website, when
- 6:39you're on a product detail page, when you're
- 6:41amazon .com, on Google. But 60 % of customers
- 6:46today have started starting their shopping journey
- 6:48within a chatbot. They're like, you pick your
- 6:53poison. And a lot of times, they're starting
- 6:55a journey where they don't know they need a product.
- 6:58They're like, oh, I have knee pain when I run.
- 7:01That will translate into, when was the last time
- 7:04you changed your shoe? You run 1 ,000 miles on
- 7:06these shoes, you need to go swap the shoe out.
- 7:09That leads to a shopper product discovery conversation.
- 7:13In that situation, AI is not looking at, hey,
- 7:18is this like a marathoner who's been running?
- 7:22Which sports star is wearing the shoe does not
- 7:24influence whether it's going to recommend a shoe
- 7:26to you. It's going to use it based on What do
- 7:29I know about Neil, right? What I know about Tarun,
- 7:31right? What pace does he run at? What kind of
- 7:34races does he do? What kind of what's his weight?
- 7:36It'll take all that into consideration and marry
- 7:39that with product data, which is complete. It
- 7:43could go back to what is the cushioning in that
- 7:46insole? Hey, there's the carbon plate. Is it
- 7:49nylon plate? What kind of material is used in
- 7:52the construction of the shoe? What is the toe
- 7:54width that you want the shoe box width that you
- 7:56need for wide toes or not? and all that information
- 8:00is not on your product detail page all the time.
- 8:03It may be sometimes, it may not be. So you get
- 8:06this entire volume of what we're calling AI enabling
- 8:11product data attributes that were all sitting
- 8:15upstream in your enterprise that were being measured
- 8:19for different reasons. You may actually have
- 8:20all that to measure and build the right size
- 8:23of the box in which the shoe needs to be shipped.
- 8:25And so yes, their supply chain is using those
- 8:28attributes, but you're not bringing it up into
- 8:30your product marketing because you never needed
- 8:32to. And now you have to connect all those dots
- 8:35and bring that in. So it has become super crucial.
- 8:40And the companies that actually invested in getting
- 8:45their foundation right are seeing that leg up
- 8:48today where they're like, hey, I get it. I did
- 8:50that work of connecting these dots internal to
- 8:53my organization. The company's two shortcuts
- 8:55are scrambling right now to say, what are the
- 8:57shortcuts can I take to catch up or leapfrog
- 9:00at this point? Yeah, so true. And of course,
- 9:04you work closely with brands trying to operationalize
- 9:08AI at scale. And I'm curious, what are companies
- 9:12discovering when they realize that their existing
- 9:15product information systems were never designed
- 9:17for AI -driven commerce? Are there any trends
- 9:20in the kind of things that they discover and
- 9:22they come to you and ask for help with? A couple
- 9:26of things that happen. I should probably not
- 9:32say this and maybe I'll get in trouble with it.
- 9:34I don't think so. It's a public company. I worked
- 9:36for BP for seven years. In my career, as I mostly
- 9:39have been on the product side, but about seven
- 9:41years, I went to the buyer side and said, I'm
- 9:43going to implement data management systems and
- 9:46understand why people don't use software the
- 9:48way I think they should use. It was a great,
- 9:51phenomenal learning for me. Thing that sticks
- 9:54to my mind is that it took me months to find
- 9:58out categorically How many wells does BP own?
- 10:04Think about it for a second. It's core to the
- 10:06existence of an oil company Oil wells are we
- 10:09drill but to be able to reliably tell you I could
- 10:12give you ten different answers I could not tell
- 10:14you the one answer as to how many so if I go
- 10:17back to a problem coming back to brands and retail
- 10:20the biggest problem is we don't know the state
- 10:25of our data. We know state of silos or slices
- 10:29of the data. Anything to do with ERP and finance,
- 10:33highest importance. You want to make sure you
- 10:36don't screw up your payment processing. You're
- 10:38going to get that right. But then progressively
- 10:40through different functions, they were always
- 10:42assigned different weightage and different reliability.
- 10:45So people don't know the state of the data. And
- 10:48suddenly, they're being forced to say, everything
- 10:51is important. because you don't know what AI
- 10:54is going to need. So surface all of that. So
- 10:57we've been actually working with a lot of our
- 11:00customers. We offer this thing called an agentic
- 11:02readiness assessment, where we're essentially
- 11:04coming back and saying, here's the spec. We think
- 11:07that these are the things that matter for an
- 11:10AI to surface your data right. And then we'll
- 11:12map you to it and come back and say, where do
- 11:14you fall on that spectrum? And we come up with
- 11:16a score. And it's interesting because when we
- 11:19go in, Most customers are like, I'm going to
- 11:22rate between a 60 and 80 on your score of 100
- 11:25as to how AI ready is my data. And we're consistently
- 11:29finding companies, small and large, at about
- 11:3330 to 40 today. And it's because they were all
- 11:37geared towards e -commerce enablement and not
- 11:41towards agentic enablement. And that gap is vast.
- 11:45It's not small. And I think as consumers, when
- 11:49we hear about the digital shelf, it sounds incredibly
- 11:52simple on the surface. But of course, behind
- 11:55every product listing, there are massive infrastructure,
- 11:58governance, and syndication challenges. So for
- 12:01everyone listening, can you help them understand
- 12:03the hidden operational complexity powering these
- 12:06modern retail ecosystems that we all almost take
- 12:09for granted as consumers? Yeah. So let's take
- 12:12maybe an example. Let's pick up. well, what product
- 12:15would we like? Let's do a shirt. Let's do a polo
- 12:16shirt, right? Hey, wear a polo shirt. The journey
- 12:20is going to start. I'm a brand, right? That could
- 12:22be Tommy Hilfiger, Ralph Lauren, whoever it is.
- 12:24I'm a brand. I'm going to start with typically
- 12:27the question. The first question that a brand
- 12:30has to think through is where is that product
- 12:33born? And it's either in the ERP system or it's
- 12:37in the PLM or the product life cycle management
- 12:39system. I am a broadened guy. I think in design
- 12:43terms. So I think in the PLM side, because I'm
- 12:45like, there's a Dean in the eye of David Lauren
- 12:49or whoever somebody's designer who looks at this.
- 12:51This is the kind of photo shoot I want to build.
- 12:53And they may be hand drawing it, but eventually
- 12:56it gets into a PLM system where they start looking
- 12:59and designing what the field will be, what materials
- 13:01it will be made of, what sizes, what the cut,
- 13:03the style, everything. And the information about
- 13:06that design is in the product lifecycle management
- 13:09system. way before the product was ever built
- 13:12or even like the first samples are built. So
- 13:15if you think about the inception of data, it
- 13:17starts there. There is a lot of theoretical design
- 13:21data there. None of it may be come back to life,
- 13:23but eventually somebody goes and builds a product,
- 13:26tests it out, says, yes, this should be born
- 13:29now. And then it moves from them into classically
- 13:33some variation of a product master. an item master,
- 13:37product master. And there are lots of three -letter
- 13:39acronyms in this industry across MDM, PXM, PIM,
- 13:44where all this makes it. But eventually, it makes
- 13:47it into a system of Repler that says, this is
- 13:50my Polo shirt. It comes in these colors, these
- 13:53tiles, these cues, right? A slim fitting, comfortable
- 13:58fit, plus sizes, petite. These seven colors,
- 14:03which will all be You know, I think in primary
- 14:06colors, my kids make fun of me. I can't differentiate.
- 14:09But they'll come up with the colors and they'll
- 14:12come up with the sizing. And now sizing is interesting
- 14:15because sizing is different in UK versus China
- 14:19versus India versus US. But the skew may still
- 14:23be the same. It may still be the same product.
- 14:26And you're saying, hey, this is a large in Asia,
- 14:28but I'm going to sell the same product as a medium
- 14:30in the US. And that's how do you manage that
- 14:34complexity? Most sophisticated product experience
- 14:37management systems will handle those contexts
- 14:40and support it where you don't have multiple
- 14:42products you're creating just because you're
- 14:44selling them in different regions just because
- 14:47you're having to write down descriptions in different
- 14:50languages or you're writing product descriptions
- 14:52to appeal to different personas because you could
- 14:55be selling a shirt that a 16 year old will want
- 14:58and a 48 year old will want and you want to be
- 15:00able to appeal to them with slightly different
- 15:03things that matter to them. You want to manage
- 15:06all of that in a single environment that allows
- 15:09you to handle that complexity. So just think
- 15:12about that. In that world, you're starting to
- 15:15understand what geographies I said. Now, you
- 15:18could be a manufacturer who has multiple brands,
- 15:21and you're selling the same product for multiple
- 15:23brands in multiple geographies, because in the
- 15:26US, you're selling it under this brand name.
- 15:28In Europe, you're selling it in a different brand
- 15:29name. You want to be able to handle all of that
- 15:32attribution as well. So there are sophisticated
- 15:35workflows. And I say sophisticated loosely because
- 15:38there are complex workflows, is what I mean,
- 15:40which allow any manufacturer, any brand to manage
- 15:46that information. And I haven't even gotten to
- 15:48what types of information. And this is just the
- 15:51complexity of the business that they work in,
- 15:54the systems that this has to go through. And
- 15:58now let's talk about attribution. You're talking
- 16:00about Your primary attribution, if you think
- 16:03about it, is your core identity. Is it a G10
- 16:07number? Is it a skew? What is the name of the
- 16:09product? What is the title? The core description.
- 16:16Something that's going to be fairly immutable
- 16:19across wherever you go. That doesn't change much,
- 16:22right? The core identity of the product will
- 16:24stay like that. But then. This could also include
- 16:27variant hierarchies, which is what I said, color,
- 16:29style, skew. This is the way I structure my product,
- 16:33the model number of the skew. But then you get
- 16:35to the marketing attribution. Marketing attribution,
- 16:39long descriptions, feature benefit, bullets,
- 16:41short summary, intended use of the statement,
- 16:45all of the images. Because you want your hero
- 16:48image is the one that shows up with Tiger Woods
- 16:50wearing that polo shirt. You could have alternate
- 16:52angles. You want lifestyle images. video 360,
- 16:57you've been going down and then you work on your
- 16:59SEO optimization. Keywords, structured queries,
- 17:02phrases that want to surface when somebody's
- 17:04searching for it in Google. And all of this is
- 17:08just towards the marketing attribution, right?
- 17:10But you also want to make sure that the product
- 17:12is ready for supply chain. Does this go in cases?
- 17:16Does it go in callix? Does it go each? And what
- 17:19is the size and dimension for that? Because you
- 17:21get that wrong and your entire truck that you're
- 17:24bringing is now you're going to end up with two
- 17:27boxes extra that didn't fit. And that could lead
- 17:29to dollars lost in that journey because you didn't
- 17:33think this through. So small changes, small errors
- 17:37in data have big impact. So we spoke marketing,
- 17:40we spoke supply chain, ERP. The best way I've
- 17:43always described this is you are managing product
- 17:46information for multiple purposes. How much am
- 17:50I making with this? Which is ERP enrichment is
- 17:53basically about everything about this product
- 17:55better be right so I can make my money. Supply
- 17:58chain is everything about the product better
- 18:01be right so I can ship this product accurately
- 18:04and not lose money. And then marketing, everything
- 18:07better be right so that I can actually appeal
- 18:09to right audience and make money on the ship.
- 18:13There are about seven such vectors that we cover,
- 18:16but these three are the three that probably best
- 18:19explain what we're doing. All of this, I just
- 18:22said, is still happening within the brand. We
- 18:27haven't even talked about after that. This has
- 18:29to go somewhere. It could go to Ralph Lauren
- 18:32stores and Ralph Lauren dot com, but it also
- 18:35needs to go to Bloomingdale's and Macy's and
- 18:37H &M. So you start thinking through that. The
- 18:41journey of making sure the product makes it across
- 18:43the physical product. and the information about
- 18:46the product, such that it's also available on
- 18:48Macy's .com. And what shows up on Macy's .com
- 18:51is exactly what shows up on Lariferafluron .com.
- 18:54It should not be different. You shouldn't be
- 18:56like, ooh. that color is fuchsia and not pink
- 18:58and now suddenly you're getting returns because
- 19:00somebody doesn't understand you mentioned some
- 19:03some big names some big brands and retailers
- 19:06there and on the flip side of this smaller brands
- 19:09they're the ones that often struggle to compete
- 19:11with enterprise retailers because they have such
- 19:15huge data and infrastructure resources so i'm
- 19:17curious Is AI helping level the playing field?
- 19:21Or is it widening the gap between companies with
- 19:23mature data operations and those still playing
- 19:25catch -up? Or do smaller businesses have less
- 19:28technical debt and are able to catch up? What
- 19:31are you seeing here? So it's an interesting question.
- 19:34And the answer isn't as linear that small brand
- 19:37versus large. It almost goes back to the verticals
- 19:40they're playing. So let's kind of separate it
- 19:42out. Think of Food and grocery, consumer packaged
- 19:45goods, retail apparel. I'll just take these three
- 19:48examples. Wherever there is complexity of use
- 19:51cases, I have multiple, hey, I sell this product
- 19:55not just by itself, but in a kit or a bundle,
- 19:57right? I need to now manage that. That complexity
- 20:00is where AI is really helping small brands scale
- 20:04up. So it's not like a volume complexity, it
- 20:07is a... breadth of the different types of attributes
- 20:11that you're able to manage with AI. That's where
- 20:14small brands are able to leapfrog large organizations
- 20:17and say, hey, I can now for the first time handle
- 20:20this complexity at scale without having to add
- 20:24more human resources. Let's be very clear here.
- 20:27We've always been able to manage these complexities.
- 20:30We just threw bodies at it. The last 10 years
- 20:33have not been kind to this industry because what
- 20:36used to be a team of 15 people is now a team
- 20:39of two people who have to manage 78 different
- 20:42channels where this product has to be sold. AI
- 20:45has given a lease on life to a lot of these grants
- 20:48to say, hey, I said that we're not going to hire
- 20:52more people and we'll just make do with who we
- 20:54have. I did not have a way forward and now I
- 20:57suddenly do have a way forward that allows me
- 20:59to actually compete and win in this market. So
- 21:04much so that most large retailers, most large
- 21:09brands, enterprises, they are less worried about
- 21:14what the other large brands are doing. They're
- 21:17looking at what the mid -market and smaller brands
- 21:20are doing, and how are they able to innovate
- 21:22with AI, and they're learning from there to become
- 21:25more agile and nimble. And for as long as I can
- 21:29remember, we have often heard people talk about
- 21:32having this utopia of a single source of truth
- 21:35for enterprise data. But in reality, almost every
- 21:39organization is still dealing with fragmented
- 21:41systems, inconsistent product records, disconnected
- 21:45teams. But here we are now. Why has solving this
- 21:48problem remained so difficult? It is true. 50
- 21:52years, we've had the same issue. We were trying
- 21:54to solve this on mainframes and RDBMS system.
- 21:57The technology may have changed. This problem
- 21:59statement has not changed yet. It's difficult
- 22:03because when you run a business, you're not thinking
- 22:08about data as an infrastructure. You're thinking
- 22:12product data is content. It's like, hey, it serves
- 22:15this purpose. I want to get the... unique identifier,
- 22:19right? Because I want to identify this as the
- 22:20same product that when I scan a barcode, I get
- 22:23the right information. That's all I care about.
- 22:25And so there is a body of people inside the company
- 22:28who are only focused on that. There's a body
- 22:30of people who are only focused on getting the
- 22:32description right because they were hired to
- 22:33get the product description right. And that interconnectivity,
- 22:38which here's how I personally feel about it.
- 22:40I think this is not a technology problem or even
- 22:44a process problem. at the heart of it is a people
- 22:48problem. And the people problem is not that people
- 22:51don't want to. They have the best intentions.
- 22:54Right. But like Jeff Bezos says, good intentions
- 22:57don't matter. Mechanisms do. What's the mechanism
- 23:01that allows you as somebody who is only focused
- 23:05on supply chain or only focused on product marketing,
- 23:07only focused on e -commerce to be able to do
- 23:10your job but still deliver? information in a
- 23:15manner that makes everybody else's job easier.
- 23:17Those pipes, those bridges, we talked about it.
- 23:21We never really invest. Those were never considered.
- 23:24No C -suite discussion, no board level discussion
- 23:27talks about that pipe because at the heart of
- 23:31it, data management is janitorial work. It's
- 23:35not sexy. It's not AI. It's not analytics like
- 23:39it was 10 years ago. It is the boring stuff.
- 23:43And the only time you bring it up, the only time
- 23:46you invest in it is when your sewer pipes break.
- 23:51That's when you worry about it. You're not looking
- 23:53at it proactively saying, if I go back. But I'll
- 23:57go back to military strategy, right? People talk
- 24:01about strategy and tactics, and you'll see the
- 24:03best generals talk about it. Wars are won based
- 24:07on logistics. They're based on supply chain.
- 24:10They're based on, hey, am I able to supply my
- 24:12army as I move stage by stage by stage? That's
- 24:15how wars are won and lost. They are not one on
- 24:17strategy and tactics. They're one on logistics.
- 24:19I firmly believe in that. And this is where I
- 24:22put the data strategy. For companies that think
- 24:26through and say, data is my logistical challenge
- 24:30that I want to address. And if they can do that
- 24:32when they're starting out, that's beautiful.
- 24:34Who does everybody who's starting out in the
- 24:37business of trying to scale up, make money quickly.
- 24:40But there is a point in time when you're hitting
- 24:43that scale, you were a mid -market company and
- 24:45you're going to become a large company. That's
- 24:48the time for you to kind of pause, step back
- 24:51and say, do I have the pipes to be able to scale
- 24:55up? Such few companies do that, but the ones
- 24:58who do absolutely kill it in the market. So if
- 25:02you go back and probably do a study of everybody
- 25:05who was able to scale up and succeed in product,
- 25:09in commerce, you will probably find that one
- 25:12point when they said, hey, I'm going to go back
- 25:14and rebuild that. And doesn't I being as issues
- 25:18here, that doesn't mean a digital transformation
- 25:20project. That doesn't mean you just hired somebody
- 25:22and said, hey, you know what, we're going to
- 25:24throw away everything we had. That just goes
- 25:26back into getting on a board and saying, OK,
- 25:29where is my product born? Like, is it? What countries
- 25:33do I sell in? What brands is going to be part
- 25:35of? Is this going to be like a three level hierarchy
- 25:37or a seven level hierarchy? Just map it out on
- 25:40a piece of paper for on your care. And that helps
- 25:43you drive positions and alignment across departments.
- 25:48The challenge exacerbates the challenge that
- 25:51you correctly pointed out. Why are people not
- 25:54able to solve this problem when these departments,
- 25:56these functions are not aligned on handoffs,
- 26:00right? So I'm a triathlete. I do Ironmans. And
- 26:04when I do, I'm not a fast one. I'm super slow.
- 26:07But the saying that goes in triathlons is that
- 26:10the races are not won during the swim or the
- 26:13bike or the rebound. The races are won during
- 26:15the transition when you switch from a sport to
- 26:18another. And it's the same thing for business.
- 26:20You win in business when it comes to the handoffs
- 26:23between these functions and how easy and frictionless
- 26:27is that handoff across the different departments.
- 26:30And that pipe that we were talking about is core
- 26:33to that thesis. And as you said, right at the
- 26:37beginning of our conversation, AI driven discovery
- 26:40is changing how consumers find products, whether
- 26:43it's search engines, marketplaces, voice assistants,
- 26:46or AI shopping agents. I know I, as an individual,
- 26:50my searching methods have completely changed
- 26:52in the last 12 months. So how do brands need
- 26:55to rethink product content and metadata for a
- 26:58world where humans might no longer be going to
- 27:01search as the primary interface? They're going
- 27:03to be using different methods. For those people
- 27:05listening, what should they be doing to prepare
- 27:07for this? I say prepare, it's already out there.
- 27:10So what should they be doing? And there are two
- 27:14elements to AI -based shopping or what we call
- 27:17agentic commerce, right? The easy way or the
- 27:22right now way to think about agentic commerce
- 27:24is we define four types of formats. Physical
- 27:27commerce in stores, right? Digital commerce on
- 27:30websites. Social commerce, you want to buy on
- 27:33TikTok and Instagram directly, and then agentic
- 27:36commerce, where you're now starting your shopper
- 27:38journey inside a chatbot. Sometimes you're finishing
- 27:41the shopper journey inside the chatbot too, because
- 27:44as you're seeing with what OpenAI is doing with
- 27:47Shopify integration, you're able to now start
- 27:50buying certain products directly. Walmart's enabling
- 27:52it directly inside OpenAI. You'll see a lot more
- 27:55of this behavior happen. Having said that, So
- 28:01I'll define two parts to it. Every brand, every
- 28:06distributor, every retailer absolutely has to
- 28:09make sure that their product shows up when somebody
- 28:13is having a conversation with a chatbot. And
- 28:16that's a much harder thing to do because if AI
- 28:21gets confused about that product, if that product
- 28:23data is not accurate, if that product data is
- 28:26not complete, He has making these decisions in
- 28:29microseconds saying hey that one wants to run
- 28:31he has knee pain Should I recommend Nike or that
- 28:33he does? the question is going to be this answered
- 28:36really quick and if the information it finds
- 28:39about Nike is not complete or Conflicting it
- 28:43ignore it and go to Adidas because it's trying
- 28:45to hit that time commitment as well, right so
- 28:48Step one is to make sure that when your product
- 28:51surfaces, it is more complete than it has ever
- 28:54been Whether you have to force it with hand whether
- 28:57you build your pipes like I said doesn't really
- 28:59matter But make sure that data is complete the
- 29:02data is correct across multiple channels Just
- 29:05because it's correct on Nike comm doesn't matter
- 29:08if Macy's has got the wrong information AI will
- 29:11not surface it It's not like it to show up and
- 29:14say hey, I don't know I got completed It just
- 29:16would ignore it and move to the next product.
- 29:18So that's super important to us. The second element
- 29:21is which is where we don't know how this place
- 29:25will pay out because if you saw in October, OpenAI
- 29:29announced the agent e -commerce protocol saying,
- 29:32hey, instant checkout. You can buy directly within
- 29:34OpenAI. But by March, they were like, we're going
- 29:39to enable discovery, but we're not directly going
- 29:42to source and say instant checkout. We're going
- 29:44to have retailers create MCP servers and connect
- 29:47to them, and then that's reasonable. It's been
- 29:49a little bit of a one step forward, one step
- 29:51back, one step forward, one step back. Nobody
- 29:54knows how this will land. But one thing is for
- 29:56sure, product discovery, just the ability for
- 29:59people to find your product has changed. And
- 30:03that is going to require most of our brands to
- 30:05start. So if I go back and say, what should these
- 30:09companies be doing today? They absolutely have
- 30:12to first make sure that they treat the product
- 30:15as infrastructure, not just content, which is
- 30:17a mind -shaft chip, right? Someone, there has
- 30:21to be some accountability for this charter. Look,
- 30:25if you're just going to say, yep, I already managed
- 30:29data being sent to Amazon and Walmart, and I'm
- 30:31just going to do it there as well, it's kind
- 30:34of the same thing, but not really. But you need
- 30:36one more level of reflection to say, there is
- 30:39a certain body of work we have to do to become
- 30:42AI ready, and are we investing in that today?
- 30:44Which is timely because every board is like,
- 30:47what are you going to do to support AI in your
- 30:49world today? There is budget for it. Start using
- 30:53that. I do think that the other thing that we
- 30:57should look at is around speed of using AI tools
- 31:02internally to accelerate. We don't talk enough
- 31:05about this today, but we call this agentic product
- 31:10experience management in our platform. But it
- 31:12has reached a certain maturity where, again,
- 31:16whether it came to discovering bad data, profiling
- 31:20it, fixing it at bulk, right? Riding agents that
- 31:23can connect multiple workflows and accelerate
- 31:26through that. That journey has also improved.
- 31:29So use the right tools, use the right technology.
- 31:34The hard part here is not the availability. Everybody's
- 31:38building AI tools today. It is the inertia. It
- 31:41is the inertia of a business process that is
- 31:45100 years old today, where a brand or a manufacturer,
- 31:48a distributor, a retailer is sitting and saying,
- 31:51this is how I work. I have these seven people.
- 31:54It's almost like a little assembly, like everybody
- 31:56has to do their part. And in that journey, they're
- 31:59like, oh, I have a 17 step process and step number
- 32:0213 and 17, I can use AI. That's going to give
- 32:07you a little bit. But if you just take a step
- 32:09back and say, is there a completely different
- 32:11way to solve this problem using the right tools
- 32:14today? You're going to see that whole 17 -step
- 32:16process may now become a three -step process,
- 32:19but you're now able to solve your bad data problem
- 32:22at scale without having to go invest and ask
- 32:26your board for another $10 million to do XYZ.
- 32:30That problem gets solved. And before you join
- 32:33me on the podcast today, I was reading about
- 32:34Sindigo's acquisition of One World Sync, which
- 32:38felt like a major move in the product data management
- 32:40space. I'm curious, how has this changed the
- 32:43conversation around product experience management,
- 32:46data syndication, and indeed global commerce
- 32:49infrastructure? What does this mean? The best
- 32:51way to describe this, right? So One World Sync,
- 32:54and there are a couple of product lines we acquired
- 32:56from One World Sync I'll talk about. The primary
- 32:58one was GDSN. syndication, right? So if you think
- 33:01of GS1 standards, which is where a G10 number
- 33:04comes from, industry standards for consumer packaged
- 33:08goods, food, grocery, multiple verticals, where
- 33:10you're saying, hey, if you have your product
- 33:13information structured this way, then tens of
- 33:17thousands of brands and retailers can consume
- 33:19it directly without you having to do more. It's
- 33:21a great standard. Makes real sense to do that.
- 33:25One will think was the largest GDSN data pool
- 33:30is the largest GDSN data pool. It's most number
- 33:33of trading partners use that standard to connect
- 33:37to each other. Sindigo was the number two there
- 33:40because we, in addition to all the other products
- 33:42of Microsoft, had a GDSN business as well. For
- 33:45us, we realized what was happening for a lot
- 33:48of our customers were they were buying a license
- 33:50to Sindigo and then they were buying a license
- 33:52to OneWell Sync. and then they were trying to
- 33:54fiddle around and try to make these things work
- 33:56together to connect the dots. So going back to,
- 33:59again, that conversation of reducing friction
- 34:02in a speed -to -market play, we realized there's
- 34:05an opportunity for us to actually connect these
- 34:08two ourselves and make sure that when you think
- 34:10about your product experience journey, when it
- 34:13comes to master data management, when it comes
- 34:16to enriching a product and adhering to a standard,
- 34:19If we take on the cost of making that integration
- 34:23first party, it provides a tremendous benefit
- 34:26to our customers who are now not having an IT
- 34:29project to connect these pipes and rapidly increase
- 34:33their value proposition for our brand, which
- 34:35says, hey, you know what? I don't have to worry
- 34:37about connecting these dots. I can just do this.
- 34:39Switch is huge again in this era of AI. Primary
- 34:45GS1 attributes I would say are split across logistics
- 34:48and marketing. So you marry those two together
- 34:51and you suddenly the shipment of product and
- 34:54product information from a brand to a distributor
- 34:56retailer. Second element that one will think
- 34:59gratis was power reviews. Power reviews is a
- 35:03company that manages ratings and reviews for
- 35:06brands and retailers. So think about it, right?
- 35:09This is the second element of the of what the
- 35:12KPI that we have. which is building trust in
- 35:15a product reviews have become increasingly important
- 35:18you want to make sure as if you're a consumer
- 35:20you're not shopping without looking at reviews
- 35:23today right you're absolutely saying hey i want
- 35:26to understand what's going on with this product
- 35:28how many people have liked it the people who
- 35:30liked it that is biased that i will not disagree
- 35:33with this because what happens is if you think
- 35:35somebody didn't like that product they're first
- 35:37going amazing why did they not like it and you're
- 35:40already mentally justified saying, is that person
- 35:42like me or is that person not like me? OK, that's
- 35:45a use case I'm never going to have. I can ignore
- 35:47it or no way that person has exactly left me.
- 35:50That will truly impact whether I'm going to buy
- 35:52it or not. So if you think about that world,
- 35:55that world also sits downstream of where we are
- 35:59today because we're making the best product information
- 36:02available at every shelf. And then there is a
- 36:06signal that's coming back from a customer that
- 36:07says. Great. Going back to the same Nike shoe
- 36:09example. Hey, you know what? I like this Nike
- 36:13shoe a lot or I don't like this Nike shoe because
- 36:15the fit is too tight or I like it a lot because
- 36:18it's very cushiony. Whatever it be, you can start
- 36:22if you analyze all those reviews at scale with
- 36:25AI, we're able to now extract a very important
- 36:28information that should actually be a part of
- 36:32my marketing copy that could help me sell more
- 36:34if I did it early. But those two worlds are so
- 36:37far apart. In a customer, one part of that brand
- 36:40is focused on getting product out to Amazon Walmart
- 36:43target. One part is just reviewing those things
- 36:46and making sure that nobody returns the products
- 36:49or is just showing higher rankings so that your
- 36:52process is up, but nobody's connecting those
- 36:54two dots. So that was the other reason for us
- 36:56for this acquisition to say, hey, by buying these
- 36:59two, we can actually help brands say. everything
- 37:01that you learn from that review can automatically
- 37:04and agentically become a part of your product
- 37:06record so that you are able to now make more
- 37:08money with that product because it's highlighting
- 37:10things that are clearly valuable. And I think
- 37:13that is a fantastic, thought -provoking moment
- 37:16to end on. But for anyone listening that would
- 37:17like to carry on this conversation with you,
- 37:20we've only scratched the surface of it today.
- 37:22For people wanting to learn more about you, your
- 37:25team, what you're doing, read about some of the
- 37:28things that you're putting out there as well,
- 37:29where would you like me to point everyone listening?
- 37:32So the best way to think about this right now,
- 37:35if you just go to syndicot .com today, we have
- 37:39two or three things. There is directly all the
- 37:43information about what we're doing in agentic
- 37:45commerce. What are we doing with agentic PXM
- 37:48or product experience management? The two big
- 37:50charters we have. We have a huge presence on
- 37:53LinkedIn. Hit me up on LinkedIn. Feel free to
- 37:55ask me any question that you want to directly
- 37:57from there. There is a Q &A post on our website
- 38:02and there's a communities channel from One World
- 38:04Sync and Sindigo, which allows you to just interact
- 38:06with us around. Awesome, well I'll add links
- 38:09to everything that you mentioned there and I
- 38:11think as a consumer that digital shelf that I
- 38:14mentioned at the beginning sounds so simple but
- 38:16I for one, I've just loved learning today about
- 38:19behind every product listing, there is this massive
- 38:22infrastructure governance and range of syndication
- 38:25challenges. And taking a look under the hood
- 38:28at some of the tech that makes that possible
- 38:30has been priceless for me. And I'm sure it will
- 38:33be for everyone listening. So I would urge them
- 38:34to check out all the links in the show notes
- 38:36to everything you mentioned there. But more Nathan,
- 38:38thank you for starting this conversation today.
- 38:41Thanks for having me. This has been absolutely.
- 38:44joyous for me as you can see I'm pretty passionate
- 38:47about this topic and I can keep rambling on.
- 38:49One of the things I loved about this conversation
- 38:52today was how my guess made something many people
- 38:55never think about feel instantly relevant. Every
- 38:58time we buy a shirt, a pair of running shoes
- 39:00or even a grocery item online we now expect the
- 39:04information to be accurate as standard. We assume
- 39:06the size, colour, ingredients, images, shipping
- 39:10details, reviews and availability are all connected.
- 39:14But as my guest explained today, behind the simple
- 39:17product page sits a huge amount of operational
- 39:20work across systems, teams, standards, workflows
- 39:24and data handoffs. And the part that really stood
- 39:27out to me was his argument that product data
- 39:29can no longer be treated as static content. Because
- 39:33in an AI powered shopping journey, the product
- 39:36information becomes the infrastructure. And if
- 39:40that data is incomplete or inconsistent, an AI
- 39:43assistant may simply ignore that product altogether
- 39:46and move to another brand with cleaner, richer
- 39:49and more reliable information. And that creates
- 39:52both an opportunity and a challenge because smaller
- 39:57brands may use AI to handle complexity that once
- 40:00required large teams and larger enterprises may
- 40:03need to rethink old processes that were built
- 40:06for websites, SEO, marketplaces, being number
- 40:10one on the first page of Google. rather than
- 40:13agentic commerce and the movement from design
- 40:17to product record and product record to retailer
- 40:19from reviews back to marketing copy and internal
- 40:23systems into AI ready experiences all these things
- 40:27will collectively increasingly shape who gets
- 40:29discovered and who gets left behind so as consumers
- 40:33start shopping through conversations rather than
- 40:36search boxes what happens to the brands whose
- 40:39product data He's not ready to be understood
- 40:42by AI. Love to hear your thoughts on this one.
- 40:44We did cover a lot and it was an interesting
- 40:45angle. So you can get a hold of me just at techtalksnetwork
- 40:49.com. You'll find eight different podcasts, 4
- 40:52,000 interviews and lots of different ways of
- 40:55contacting me. A big thank you to Nord layer
- 40:59for backing the podcast and supporting the kind
- 41:02of real -world cyber security conversations that
- 41:05we need more of because there's someone that
- 41:07records 65 plus interviews a month I've personally
- 41:11seen a huge increase in browser based attacks
- 41:14over the past year whether that be phishing malicious
- 41:17extensions account takeovers the list is long
- 41:20and it's all happening where people spend most
- 41:24of their time inside the browser So Nordlayer's
- 41:28new business browser that's built to address
- 41:30exactly that. It blocks malicious sites before
- 41:33they load. It limits risky behaviors like uncontrolled
- 41:37downloads or data sharing and gives you visibility
- 41:41into how your team interacts with web apps. And
- 41:44it also helps you stay compliant by controlling
- 41:46access and enforcing policies. without the need
- 41:50to rely on multiple disconnected tools. So for
- 41:53anyone listening that is thinking seriously about
- 41:55reducing risk in SaaS heavy environments, this
- 41:58feels like a smarter and more focused approach.
- 42:02And you can learn more about it by visiting NordLayer
- 42:05.com slash browser. So let me know your thoughts.
- 42:09I'll be back again real soon with another guest,
- 42:11but thank you for listening today and I'll speak
- 42:14with you again soon. Bye for now.