Latest / Inside Commerce: Ecommerce Strategy, CX and Technology Podcast / EP325: Elastic Path CEO Bryan House On How B2B Ecommerce Businesses Can Leverage AI Product Discovery & Agentic Commerce
Transcript
- 0:00And do you think that AI compresses the funnel from
- 0:02discovery comparison configuration can all happen
- 0:06much, much faster and maybe in one or two steps or as today,
- 0:09those are, you know, much more elongated, right?
- 0:12Because answer engines aren't, you know, putting in a query and
- 0:15then hitting the filter and narrowing the results.
- 0:17And, you know, and all the sort of sort of manual things that we
- 0:19do today, they're processing the information and doing that on
- 0:23the fly. You listen to Inside Commerce, a
- 0:26weekly ecom strategy, technology and CX focus podcast hosted by
- 0:29experienced digital consultants and re platform specialist James
- 0:32Get and Paul Rogers. Get insights into ecom strategy,
- 0:34leading brands, hear from industry thought leaders, and
- 0:37leading tech vendors with practical advice to help you
- 0:39improve your online CX. This podcast is sponsored by
- 0:44World Pay. More about them later.
- 0:47Another really interesting topic today, we're drilling back into
- 0:49B2B hot on the heels of doing an episode a few weeks back around
- 0:54overall like B2B Trends with Rupert Cross from 5874 Commerce,
- 0:58but really happy to have a highly respected thought leader
- 1:03in the industry. Joined us today to talk about
- 1:05how B2B ecommerce business can take advantage of AI product
- 1:09discovering agentic commerce. So getting beyond the hype into
- 1:12real practicalities of of what this means and what and how
- 1:14things are shaped up in B2B. So warm welcome back to Brian
- 1:18House, the CEO of Elastic Path. Awesome, James, thanks for
- 1:22having me today. Yeah, thanks so much.
- 1:25We Brian was on probably it was about two to three years ago we
- 1:28did an episode centric to the tech stack around Elastic Path.
- 1:32Now we're coming to more broader strategy stuff.
- 1:34But before we get into the questions, do you want to give
- 1:37people who don't know you and don't know Elastic Path a bit of
- 1:39a flavour for like you know, your journey to CEO and also
- 1:42where Elastic Path sits in the market?
- 1:46Yeah, sure. I'm happy to.
- 1:48Yeah. So I've been in an elastic path
- 1:49for about 5 years and I came in initially to run the product
- 1:53organization. I've been in the digital
- 1:55experience and the content and commerce space for a long time,
- 1:59probably too long. And and so came in to run
- 2:02product and quickly took over customer success and did some
- 2:06interesting things with by putting customer success
- 2:09directly in our product management organization.
- 2:11What it allowed us to do was shorten that loop between our
- 2:15frontline individuals working with customers and our product
- 2:18teams, making our products better and helping us innovate
- 2:21more quickly as well as delivering great service to
- 2:23customers. And so that actually worked out
- 2:25exceptionally well here. And then about a year ago I took
- 2:29over, a little over a year ago, I took over President and CEO of
- 2:32the company and sort of, you know, really focusing us on how
- 2:35do we grow in our market. You know, we've recently
- 2:40narrowed our focus down to, to B to B commerce use cases.
- 2:45I think that's where the most interesting part of the market
- 2:48is right now. But more more important for us
- 2:50is, you know, our strength is in complex catalog scenarios.
- 2:54We have a really powerful tool product experience manager built
- 2:57into elastic path composable commerce and it suits.
- 3:01It's exceptionally well suited for mid market B to B companies
- 3:06that have complex catalog challenges and sort of looking
- 3:09to automate workflows around us that have many if not thousands
- 3:14of price books, lots of complex product scenarios, bundling
- 3:18things where they need to do configurations and assemblies
- 3:22very easily. Are you know, our products
- 3:24exceptionally well suited for that.
- 3:25So that's really where our focus is these days.
- 3:28Excellent. Yeah, It's really interesting to
- 3:29to see that. And I'm not surprised actually
- 3:31because the growth in B to B commerce is huge at the moment.
- 3:35And also the like the US specifically seems to be a bit
- 3:37more advanced in terms of the way that B to B businesses think
- 3:41about commerce online. Not say that the UK isn't doing
- 3:45anything, but I think we're probably a little bit behind in
- 3:47terms of embracing more modern tech around B to B journeys.
- 3:51Yeah. And I, I, you know, I think
- 3:53COVID interestingly was a big forcing function there.
- 3:56You know, it's funny, when I joined the company five years
- 3:59ago, all of our pipeline was in B2C and we were at the time
- 4:03spending a lot of time on an R&D effort on B2B.
- 4:05And I'm like, wait, the market's not right here yet.
- 4:07But you know, COVID sort of drove a lot of behavioral change
- 4:11in terms of how people want to buy.
- 4:13And so now I think the the market's ready for the
- 4:16technology, increasingly ready. And maybe, maybe the US market's
- 4:20a little bit ahead of the UK market, but I'm sure it's coming
- 4:22on quickly. Excellent, right.
- 4:26To set a bit of context for all of those listening about the the
- 4:29key topics we're going to talk about.
- 4:31So we're going to look at the shift from search led by and to
- 4:33answer LED discovery. This is specific to B to B by
- 4:35the way. We are going to be looking at
- 4:38the critical role structured product data practical playbook
- 4:42for brands preparing for 2026 and then a broader, like Brian's
- 4:46broader take on 2026 commerce predictions in in B to B.
- 4:50So really interesting topics do stay tuned.
- 4:54Let's what I'm going to start with Ryan, is there's we talked
- 4:57about before recording this is there's a lot of noise about
- 5:01agentic commerce, AI discovery in the market.
- 5:04And amongst the really good information there, I think
- 5:07there's certain, certain conflation of what agentic
- 5:09commerce is. So I want to be really clear,
- 5:11I'm going to talk about where I think it is and then hear how
- 5:14you guys position it and how you are defining it from AB to B
- 5:17perspective. So agentic commerce, to me, it's
- 5:21basically it's systems that can take autonomous actions to
- 5:24achieve goals, which could be buying a product, booking a
- 5:27ticket, whatever that action could be.
- 5:30An AI product discovery is linked to this, but isn't
- 5:33necessarily agentic because AI product discovery can just be
- 5:36the use of AI platforms to help answer informations, retrieve
- 5:39product information without taking specific following
- 5:42actions. It can just be an information
- 5:44request. And those AI product discovery
- 5:46tools like Chat GPTS, etcetera, are using LLMS for doing this.
- 5:50So in my head when I'm, because I have this conversation at the
- 5:53time, is that Ansel discovery is a major component of agentic
- 5:55commerce. And the foundation is structured
- 5:58data, which is where my heads are.
- 6:00When you're talking to businesses and talking to the
- 6:04wider market, how will you define an agentic commerce?
- 6:10You know, it's, it's an interesting, interesting
- 6:12question in that, you know, we don't really, that's not really
- 6:18a, a, a term we use and that and, and certainly we market to
- 6:22because I think to your point, if it, it, it's meaning is
- 6:27unclear and it's really, really big, right?
- 6:29And so I think organizations, you know, running a product
- 6:33organization myself, you're always looking at crawl, walk,
- 6:35run and how do I sort of how do I do small, make incremental
- 6:39investments and learn and then sort of build on those rather
- 6:42than sort of The Big Bang. And I think, you know, a lot of
- 6:44people see this, Oh, we're going to have an autonomous shopping
- 6:47piece where I just, you know, I put in my all the like clothing
- 6:50sizes and, you know, the agent just goes, buys clothing for me.
- 6:53I, I don't think that's in use by, you know, by any sort of
- 6:58stretch of the imagination, sort of a majority of the market.
- 7:01And I think brands are and B to B businesses are sort of
- 7:04skeptical of that today is a thing.
- 7:07But what we see a lot is how do we and conversation a lot that
- 7:12we facilitate and have in with with prospects and customers is
- 7:16where the workflow place is that we can automate using AI, right
- 7:20and drive more efficiency. And so, you know, answer LED
- 7:24discovery is a really important piece of that where, you know,
- 7:28organizations are looking to say, how do I make sure my
- 7:32products show up in the right way in the, in the LLMS?
- 7:37So that way, 'cause increasingly customers are searching there,
- 7:40right? And, and, you know, particularly
- 7:42in AB to B context, there's a lot of like, how do I do a thing
- 7:45or, hey, I have this machine and this is broken.
- 7:48What are the parts that I need, right?
- 7:49And and you get a lot more, you know, where keywords were driven
- 7:54by the fewer the words in the query, the better you know, the
- 7:58Genting prompting. You can have much more complex
- 8:01prompts in there to find that directly the information you
- 8:04have. And now the question is, how do
- 8:06as a brand you make sure your things show up there?
- 8:09So that way the answer engine isn't confidently wrong with the
- 8:12wrong information, which is a problem we see all the time.
- 8:15So I think that is more the, the, the angle that we have
- 8:20taken as we talk about sort of the role of AI in commerce and
- 8:25it's well suited to, you know, the capabilities we deliver as
- 8:29part of our commerce platform. But it's also the, the other
- 8:32piece of that is, is workflow automation, right?
- 8:35You know, one of the things you need to be is you need quote,
- 8:37quote generation and you need workflow processes and you know,
- 8:41typically better, for better or worse.
- 8:43I've built plenty of BPM solutions in my career and those
- 8:47are unnecessarily complex and, you know, giant if then tree
- 8:51statements of, you know, how to run a workflow, you know,
- 8:54process and you spend all this time building products to
- 8:57support them. You build all this time building
- 8:59them. And then someone just picks up
- 9:00the phone and asks the question, right.
- 9:02And they go around them and avoid them and they don't really
- 9:04use them. And and so I think there's an
- 9:06interesting piece from an AI automation that works the way
- 9:09people think more along the lines of then these sort of
- 9:12structured sort of if them statement, you know, white board
- 9:15sessions to and how do we go from point A to point B with 27
- 9:18steps that, you know, people avoid at all costs.
- 9:23Ready to grow your business online?
- 9:25With World Pay E-commerce you can deliver fast, secure
- 9:28payments your customers love, backed by global reach and real
- 9:33people powered support. Scale with confidence and
- 9:37enhance your customers experience with World Pay
- 9:39E-commerce. Yes, and if only humans were
- 9:43that predictable in their consistency of behaviour as well
- 9:46when they're trying to answer questions or buy products.
- 9:50So if we look at the, the, the, the shift from as you're talking
- 9:54about from, I guess search LED buy.
- 9:55And whereas I want to buy a product that is X2 which product
- 10:00is best for this specific business use case.
- 10:03And having just worked with a lot of dental supplies company
- 10:06in the US, they have a set of products and different types of
- 10:10dental service organisations buy different versions of products
- 10:13based on their use cases. So you've got all these nuances
- 10:17in in B to B and you've got, as you've already alluded to, the
- 10:20growth in AI agents such as ChatGPT perplexity.
- 10:23And I know you're we'll talk more about your product and how
- 10:26you've evolved in that. They're called platform level
- 10:28later, but they're starting to offer small base.
- 10:31The volumes per day are going up exponentially and that Q&A
- 10:36search is is evolving out of traditional search engines like
- 10:39Google into these LLM based models.
- 10:42So the big question really is like how ready do you think B to
- 10:46B brands are for this shift? What are you seeing out there?
- 10:49Like what status of maturity are people at?
- 10:51How are people thinking about this at the moment?
- 10:56You know, I'd it, I'd say it's still early days in, you know,
- 10:59and I think that creates opportunities for brands that
- 11:02want to be innovative and create sort of new experiences to lock
- 11:05in customer loyalty and sort of build deeper relationships with
- 11:09customers. You know, I think one of the
- 11:11biggest challenges in these areas around B to B is, you
- 11:16know, because of the complexity you just described in that
- 11:19dental products use case. That's why you have sales
- 11:21people, right? Sales people help you navigate
- 11:23that complexity in there. And so does the change mean is
- 11:27this AI sort of answer LED discovery sort of a superpower
- 11:33for your existing sales team? Is it a replacement for your
- 11:35existing sales team? What role does it play in, in
- 11:38the sort of evolution of your, of your organization beyond your
- 11:41digital strategy? And I think, you know, when I've
- 11:44seen sort of B to B commerce initiative, digital commerce
- 11:47initiatives fail, it's usually that organizational resistance
- 11:51to change is the bigger dynamic rather than sort of a technical
- 11:55fees. But I, I think the other thing
- 11:58that we are starting to see, and this really comes out of
- 12:01discussion is there are scenarios where like, oh, well,
- 12:06it's, you know, our product is our IP.
- 12:08So we don't want that out. It's not on search engines and
- 12:11it's not necessarily going to be on answer engines.
- 12:13So we don't necessarily need AI, But then you start to think
- 12:17about where's what's the more a little bit deeper into the pre
- 12:21sale experience and the post sale experience.
- 12:23There's plenty of opportunity for AI automation.
- 12:26And through those discussions, people are like, oh, I hadn't
- 12:28thought of that, right. I'm thinking just as we started
- 12:31with, you know, AI and commerce means agency commerce, you know,
- 12:35an agent goes and buys all the things and that isn't
- 12:38necessarily sort of de facto correlation, right?
- 12:41It it can be, Oh no, actually there's all this information we
- 12:45send with our products when someone buys it today, we
- 12:48manually key all that in, you know, why can't we automate that
- 12:52in the population of that and have an AI check, a quality
- 12:54check around that, right? That simplifies processes and
- 12:57celebrates things dramatically. Yeah, the, the quote process is
- 13:02a really interesting one. I, I'd had a conversation with a
- 13:04couple of B to B consults pre Christmas about this about this
- 13:08is where I think AI is going to add so much value in B to B
- 13:12processes because it can reduce the tying up of salespeople on
- 13:17on questions that can be answered through automation and
- 13:21help them focus on more value add tasks in that quote process
- 13:24to a value add from the customer's point of view, but B
- 13:28value add from the business and the closure point of view.
- 13:30So you have that ability for someone to, to use an agent to
- 13:33say, Oh, actually I need a product that can do AB and C.
- 13:36Which one of you, which one? Which version of this product is
- 13:39right for X and then say, Oh, I'm not quite sure it does this.
- 13:42Does it do that? And that rely on the structured
- 13:45data that sits behind it in the catalog.
- 13:47But using the the LLM and and the AI capability to then
- 13:50generate the right answer in the right format, the value on that
- 13:53to me feels huge. Yeah, I, I couldn't agree more.
- 13:57And you know, and I also think it sort of replaces the OR in
- 14:02some ways sort of offloads the burden of, of knowledge, right.
- 14:06Like a good example is, you know, for years in and if you
- 14:10wanted to be a taxi driver in London, you had to know, you
- 14:13know, you were tested on your knowledge of the streets of the
- 14:15city and London's an enormous city, right?
- 14:17That's a hard thing. Then Uber came along and and
- 14:19GPSI don't need to know that anymore.
- 14:21That's not valuable. And I think the same thing
- 14:23happens from a human understanding of of complex
- 14:27product catalogs in B to B scenarios, right?
- 14:29Like, you know, I don't need you to memorize the catalog per SE
- 14:32and understand all the permutations.
- 14:34What I need you to do is build relationships and help move the
- 14:37transaction along. And so, you know, I think AI
- 14:39assistance in here, particularly in complex product sets is
- 14:44potentially enormous. It really gives selling
- 14:48organizations superpowers to to do things much, much more easily
- 14:52and focus on the things where they add real value into the
- 14:54into, you know, not generating the quote isn't with a valuable
- 14:58piece, it's actually building the relationship and the trust
- 15:01in the branch so that way people can want to continue buying for
- 15:04your organization. Yeah.
- 15:06And I think this is one of one of the things that that some
- 15:08people are worried about is, well, actually if AI is doing.
- 15:11And so what's the point of my sales team as well?
- 15:13Actually, you're right. It's the ability and time to
- 15:17have meaningful connections with people, whether that's phoning
- 15:19up and saying are you OK, can I help you?
- 15:21Go home for a meeting, take them out for a coffee.
- 15:24Value add things like updating them on latest products or
- 15:27information. The things that make a
- 15:29difference to building long term human relationships.
- 15:31If you don't have to do 6 hours of tedious admin every week on
- 15:36updating quotes and answering quote questions, but you can
- 15:39focus on that that that feels like a a real nice value
- 15:43proposition between tech and people.
- 15:46Yeah, I, I, I completely agree. And I think this, this, this is,
- 15:50has the potential to be an incredible sort of tech enabled
- 15:53use case where sort of the transformation in the
- 15:56organization that this represents is not your sort of
- 15:59traditional digital transformation.
- 16:01While digital remains a part of it, but is actually much more
- 16:05about your organizational change and transformation in terms of
- 16:08how you engage with your customers.
- 16:11You know, and look, you know, people are busy, they want to,
- 16:13they want these things to be as frictionless as possible.
- 16:17And so I think the opportunity to remove friction while adding
- 16:20value at the same time through the use of AI and digital
- 16:24commerce for B to B is is enormous.
- 16:28Yeah, and there's a there's a lot being spoken about over the
- 16:31last like 12 months in B2C commerce about where this is
- 16:36headed in terms of, you know, shoppers using ChatGPT to find
- 16:40products, clothing, beauty, whatever it might be.
- 16:43And there's been some data release where people are showing
- 16:46like from their analytics SC and that it's now, you know, from
- 16:49half to 1% of of sessions coming into etcetera.
- 16:53But I've seen less talk than B to B.
- 16:55And I know that that that you're that you're predicting and
- 16:59seeing that actually the, the shift towards this sort of
- 17:02answer LED discovery is going to be even bigger than B to B.
- 17:05Do you want to give people like a bit of a walkthrough of why do
- 17:08you think it's going to be have a significant impact in B to B
- 17:12and how is how is buyer behaviour need to be changing in
- 17:17relation to this? Yeah.
- 17:20So you know, I think, I think what the, you know, look every,
- 17:25every company, whether they be B to B or B to C is looking to
- 17:28identify and land new customers. And I think the I think one of
- 17:35the things that that the opportunities that exists with
- 17:38sort of answer LED discovery here is by structuring your
- 17:41product information so it's easily consumed and with the
- 17:44correct information by LLMS, you actually open yourself up to a
- 17:50set of market that didn't maybe didn't know you existed, but
- 17:53have problems that you are well suited to solve.
- 17:55Right. So which valve that's this
- 17:57pneumatic, you know, system that I have or what, what, you know,
- 18:01skew replaces this disconnected, discontinued part of my
- 18:04assembly. Those are those are the, you
- 18:06know, the types of prompts that are much better articulations of
- 18:10the problem an individual has can introduce you to to identify
- 18:17and find new customers. Like, Hey, I didn't know that
- 18:20brand sold. You know, I never heard of this
- 18:22company that has a solution to my problem, or I didn't know
- 18:25that and built and build a relationship that way.
- 18:28And, you know, I think the other piece is there's a opportunity
- 18:34given sort of the, you know, with repeat customers is how do
- 18:39you, you know, I think there's a real opportunity to accelerate
- 18:42sort of repurchases and sort of ongoing sort of replenishment
- 18:47and elements here, you know, based on sort of standard,
- 18:50standard orders and, you know, sort of applying AI can sort of
- 18:53apply subscription type principles and sort of behaviors
- 18:57into sort of the reorder process.
- 18:58So that way, you know, you're predict, you know, in the same
- 19:01way we, we talk about AI for predictive maintenance, it's
- 19:04also predictive. Hey, you know, the last 10
- 19:06weeks, you've ordered 10 items with X.
- 19:08You know, let's pre populate your cart, your your quote for,
- 19:12you know, with that includes that information.
- 19:14Or has something changed? You know, it's high end into to
- 19:17purchase demand, right? There's suddenly all this
- 19:19information that's available around how you're your business
- 19:25operates that can inform and make effectively recommendations
- 19:29that are useful in AB to B context given sort of your long
- 19:33standing relationship with a customer, right, right.
- 19:36And, and so I think that actually represents a really
- 19:40interesting opportunity to accelerate some of this sort of
- 19:42purchasing that happens in, in B to B workflows in selling
- 19:47processes based on automation and understanding.
- 19:52Because suddenly now all this data is available to make
- 19:56recommendations to you based on sort of the transaction volume
- 19:59you have, the businesses you're making, the, you know, the way
- 20:02your business is performing at any given time.
- 20:05So I, I think those are areas that are, are really interesting
- 20:10sort of applying this predictive intelligence from an account,
- 20:13you know, a recurring. Engagements with a existing
- 20:18customer. They get really interesting at
- 20:20B2B. And the next question I've got,
- 20:24which is the the bread and butter.
- 20:26This is about the critical role of structured product data.
- 20:28And it's always made me laugh that every now and then we need
- 20:32a reminder for people to take data seriously and improve the
- 20:36data integrity and clean out the mess that they've allowed.
- 20:39You know, the debt, the as with technical debt, the product data
- 20:43debt and the structure product data.
- 20:45And you've, you've written about this a lot.
- 20:47You've, you've posted on LinkedIn.
- 20:49I know the nasty path talks about this and it's a core
- 20:51capability of your product. But I want to get into two areas
- 20:53#1 is about the data and what it means to be to be businesses.
- 20:58The second it then come back on to your solution and where
- 21:01you're going technically and what tools you're building now
- 21:03to support in this. So let's start on the data and
- 21:07to give people a contact. You posted on LinkedIn very
- 21:09recently about issues with Amazon's Buy for Me agent
- 21:13running errors with products due to product data issues.
- 21:15Do you want to give it a flavour of what's happened, what the
- 21:18causes are and then what actually B to B businesses need
- 21:22to do in order to be ready for this style of arts of their
- 21:26discovery? Sure.
- 21:28So I think, you know, one of the challenges here is, you know,
- 21:34the nature of the information that you need to deliver to any
- 21:40discovery experience is evolving quickly, right?
- 21:43And, and, and, and today, you know, and what happens is it's
- 21:47so easy for, you know, whether it be a search index or an LLM
- 21:52to sort of see a little bit of information about you and sort
- 21:54of and sort of make a bunch of assumptions about your product.
- 21:57And this is kind of what happened with the Amazon
- 21:58situation is they've been complete product data about the
- 22:01products themselves. They don't have inventory data.
- 22:03They, you know, so they're selling things that are out of
- 22:05stock and, and sort of making all of these assumptions because
- 22:09they sort of have a tiny bit of knowledge.
- 22:11But in this sort of answer LED discovery world, in effect, you
- 22:15need an 80 page PDF for each of your products worth of data.
- 22:21So, and to ensure that the answer engine knows when to
- 22:24recommend your products, when to, to return your products
- 22:27based on a prompt. And, and I think that's the big
- 22:29piece, like in a, in a content product data syndication model
- 22:35through things like, you know, Pims and, and what have you out
- 22:38to the, to the market. There's way too much information
- 22:41for a marketplace or search engine engine to consume, right?
- 22:45You're not going to push that you through your Google merchant
- 22:47center, you're constrained and how much information they can,
- 22:50they can, they can take in. And so that leads even when you
- 22:55have it set up effectively, there's only so much you can
- 22:58share, but that in that, you know, what you end up doing is,
- 23:02you know, you operate within the constraints and you leave all
- 23:04this other information that's relevant out.
- 23:07And I think the opportunity in particularly in B to B is by
- 23:10definition, these products are more complex.
- 23:12They have more information required, you know,
- 23:15interdependencies, you know, how products go together, you know,
- 23:21compatibility, think, you know, all sorts of elements like this
- 23:25regional differences, you know, in those are things that that
- 23:29are aren't able to be well handled in sort of traditional
- 23:32sort of SEO and search LED processes that I think creates a
- 23:36new opportunity in an LLM world to to to manage.
- 23:41Plus, I do think, you know, we've always see the our
- 23:45personal behaviors and sort of lead into our professional
- 23:48behaviors. And as people get, you know,
- 23:51more and more comfortable writing prompts and querying
- 23:55answer engines for how to how to find things, they're going to
- 23:58want to apply that. That's that in B2B use cases.
- 24:02And so I think the opportunity here is how do you best
- 24:05represent the important details of your products in as much
- 24:10depth as you can to the LLMS. So that way you know it can
- 24:14behave the way it's designed to behave, which is to make
- 24:17predictions about what the next best word is in A, in a
- 24:20structure, right? And sort of complete the
- 24:22sentence, if you will. And that I think that works has
- 24:26the potential to work exceptionally well in B to B, in
- 24:29the same way it works exceptionally well in software
- 24:31programming. And I'm keen to, to explore a
- 24:35bit about where you've gone technically in the, in the
- 24:39product on this, because you've got AI optimised product
- 24:42experience manager feeds and you've got, you've basically got
- 24:48the structure there to now merchants to create LLM ready
- 24:52product catalog pricing, etcetera, to, to push into
- 24:56engines like ChatGPT perplexity. Do you want to talk us through
- 24:59both those those solutions? What do they do?
- 25:02What do they enable for merchants?
- 25:05And also, I guess for people to listen, like what, what work
- 25:08does the merchant need to do beyond putting the source data
- 25:11in? Or is all of the connectivity
- 25:13automatically enabled out-of-the-box?
- 25:16So, so you know a couple of things that we've done.
- 25:20So first, our our product experience manager, you know, we
- 25:23made this decision years ago before this answer LED discovery
- 25:27was even a process for us. But effectively what we did is
- 25:30we decoupled the traditional product catalog and an ecommerce
- 25:34system into separate discrete services.
- 25:36What that opens up is sort of all these combinatorial
- 25:40opportunities and all this flexibility to add information,
- 25:43whether you know, so we support 10,000 variations on a
- 25:47particular product, which you know, doesn't MIT matter in
- 25:50shoes per SE, but it might MIT chat matter in machine parts.
- 25:54And the ability to to configure and assemble those together and
- 25:58and and pre describe combinations becomes really
- 26:01important. And you can combine that with
- 26:03pricing and inventory. So, so our catal, what it has
- 26:06done is made our catalog information model, data model
- 26:09really, really flexible. Then a couple years ago we added
- 26:13capability called commerce extensions, which we can really
- 26:16think of frankly as databases as a service.
- 26:20So you can store any type of information.
- 26:22So now that is again it explodes the amount of information that
- 26:25you can store around your products to inform typically
- 26:30inform workflows or processes, recommendation engines and now
- 26:34can go and work exceptionally well and well suited for LLMS.
- 26:39And so this combination of an A very elastic pun intended
- 26:46catalog structure of product data model that allows you to
- 26:49capture tons of information is at the core of this.
- 26:53And we built all this before so the LLMS were thing.
- 26:57So you know, good on us if you will.
- 27:00And now we've we've made it so that way you can all your
- 27:03product data is already structured for consumption by
- 27:06open AIS product feeds or for perplexity engines.
- 27:10And so really all the merchant needs to do is to point those
- 27:13product information feeds to ChatGPT and perplexity.
- 27:18And there's a, there's a well documented sort of transact
- 27:21handshake process for that. And now the LLN knows where to
- 27:24get the structured information. And the, the nice thing is we
- 27:27tied it into our catalog delta features.
- 27:31So every time you make an update, you don't republish all
- 27:35the information in the catalog, you just publish the delta
- 27:37changes. And so that same process works
- 27:41with the LLN. So you're giving much smaller
- 27:44bits of information, delivering it to the LLNS, which is
- 27:47obviously much more consumable on on their part.
- 27:51So yeah, you have the one time process of all the information
- 27:54goes up and then you're just sending deltas from there on
- 27:56out, which makes it much easier to for to ingest over time.
- 28:00Yeah, that is very nice. Yeah, the pain of having to re
- 28:05publish or re export in bulk every time is is yeah.
- 28:09Should be yeah, yeah, exactly, exactly right.
- 28:13You know, and that's resource and time intensive for both us
- 28:16and for the merchants so that you know, so, so we've taken
- 28:19that problem off the table. The other thing that we we we've
- 28:22built into this is we have a whole sort of rule structure on
- 28:26our catalogs, right. And this is really important in
- 28:29B to B because it determines who gets shown what products, what
- 28:33pricings based on what attributes and you can assign
- 28:35all that based on accounts. And so that makes that gives you
- 28:40really tight grain granularity over who sees what information,
- 28:45in what context they see that. And all of that sort is
- 28:48performed automatically in the background.
- 28:50You know, we're still working through how do you, how do you,
- 28:54how do you take advantage of that in, you know, in some form
- 28:56or another or do you in a, in a sort of LLM scenario.
- 29:01But you know, I think this is a rapidly evolving market.
- 29:03So some of the nuances around how businesses really operate,
- 29:08we're still exploring in conjunction with our customers.
- 29:12Yeah. Do you know that's I had AI had
- 29:13a question that popped into my head when I was prepping for
- 29:16this and that was exactly is there there is some content that
- 29:19is a hidden based on specific account credentials.
- 29:24I mean, pricing is a classic one, but there's also additional
- 29:27information that's only relevant to certain customers that wasn't
- 29:30made mass market and it's got commercial sensitivity.
- 29:34So I guess, I guess in my understanding from what you said
- 29:37about the structure of your catalogues where the you have
- 29:39all the different components so that only not all have to be
- 29:42published is AI guess you can store account specific and
- 29:47commercially sensitive data and not expose it publicly, right?
- 29:50So LLM. Exactly right.
- 29:52Doesn't have to consume what it shouldn't be consuming.
- 29:55Yeah, exactly right. Through authentication models
- 29:58and account ID, you know, account controls as well as
- 30:01through our catalog rules, you can designate a very
- 30:04specifically which information in this case you'd share with an
- 30:06LLM for discovery purposes versus which is only available
- 30:10to, you know, customer cohort one versus cohort 2.
- 30:14Yeah, that makes a lot of sense. The the question that that most
- 30:18people when they listen in to podcasts have is this makes
- 30:21sense. I understand what this this
- 30:23means to business. What do I do about this?
- 30:25So and I know that you you've lasted past prepared like it's
- 30:30it's view on what a practical payback for B to B brands
- 30:33preparing for 2026 is and like a three-step readiness plan.
- 30:37Do you want to talk us through what you think these B to B
- 30:40businesses and the E com teams should be doing?
- 30:46Great question. You know, I think that, you
- 30:50know, I think it is all very much in line with our
- 30:52conversation here is 1 is I think the first is, is get an
- 30:57understanding, you know, if we just take the answer LED
- 31:00discovery process is what percentage of information or
- 31:05traffic and you know, and get a developed hypothesis on how much
- 31:08of that is coming from your, from answer engines.
- 31:11You know, this is a conversation we're asking customers all the
- 31:14time across all use cases is, you know, how, how are you
- 31:18seeing your traffic dynamics change?
- 31:20And it's a little bit different in B to B and it's going to be
- 31:22slower. But there's increasingly, I
- 31:26think you're going to see that in general in the discovery of
- 31:29your information of your, of your company and your product
- 31:32suite and, and the problems that you solved.
- 31:35Then I think once you feel like you'd figure out where the sort
- 31:38of tipping point is, is like, hey, we need to really be here.
- 31:41Then I think the second question is, is how well suited is our
- 31:45data for sharing out to LLS? And it's really sort of the data
- 31:50analysis, product data analysis exercise to, you know, And, you
- 31:57know, I think one of the questions I, I would encourage
- 32:00people to ask is what are the constraints that our existing
- 32:04systems, be it our ERP system, our commerce system, our PIM put
- 32:09on the amount of information and the types of information that we
- 32:13have and what additional information sources or, or get
- 32:18content data pieces would be valuable to improve the
- 32:22discovery, the understanding and the, the relevance of our, of
- 32:26our products. You know, because I, I think
- 32:29clearly, you know, we see this a lot because in B to B, you know,
- 32:33we predominantly see home grown systems.
- 32:36But the nature of the, the, the beast, if you will, is there's
- 32:39always constraints on there, right?
- 32:41And so there's things that are left out of the system that are,
- 32:44you know, in marketing collateral and sales people's
- 32:47heads and training information, training guides, what have you
- 32:50in support, you know, manuals or FAQs somewhere outside the
- 32:54context of the buying experience that are supporting the product.
- 32:58Well, now suddenly you have, how can we add that in to inform the
- 33:03discoverability of our products in this sort of new answer LED
- 33:06model? And then from there I, you know,
- 33:10then the, the really the question is what's the best way
- 33:13and the most cost efficient way to push this information into
- 33:16the LLS, right. And I think, you know, our value
- 33:19proposition is we can, you know, very quickly do a catalog
- 33:24modeling exercise and then our, all of our data is already
- 33:29enabled for consumption by LM. So you don't have to do any
- 33:32extra work to structure your data for consumption by these
- 33:36discovery LED engines. We have a whole guide on how we
- 33:38do this in our documentation site around how to to provide
- 33:43your product feeds into the Open AI product feed.
- 33:48What one of the one of the notes I've got here, which I think is
- 33:51really, really smart idea and I'd love to hear a bit more
- 33:54about how how people do this or what you've used it for is well,
- 33:59internal small tests to help you gauge readiness for submission
- 34:04using like small copilot agents, for example, to run some simple
- 34:07questions to see what sort of information is coming back.
- 34:11Is it pulling in relevant information?
- 34:13Does the data come back in a struction of way?
- 34:15Do you want to talk us through that?
- 34:17Because I think that sounds like a really good practical way for
- 34:20people to start experimenting and testing the quality of their
- 34:23data. Yeah, I mean, you know, one of
- 34:26the things that, that I think the easiest way to do it is, is
- 34:31we all sit in front of our machines all day is start
- 34:33querying your products in, in, in chat CPT and in perplexity,
- 34:37right. You know, and it's, I do this
- 34:39for our customers. It's money.
- 34:40I did, I said to myself, I'm going to do all my Christmas and
- 34:43holiday shopping on chat CPT and, and you very quickly start
- 34:48to see where the sharp edges are around, whether it's, you know,
- 34:52you're actively in a purchase process or you're evaluating the
- 34:55efficacy of your current data that's in there.
- 34:59You know, how do you, and how can you better structure that
- 35:02information? So I think, you know, and this
- 35:04has been a big, certainly a big, you know, theme of mine over the
- 35:12last year is the only way we're going to get better and
- 35:15understanding how to use AI is by using AI, right?
- 35:18And all the unlock moments were there.
- 35:20I think on the product side, you know, our organization that
- 35:24we're participating in the mock alliance AI initiatives.
- 35:28And, and so we've been doing a lot, you know, working with
- 35:31hackathons that they've done to understand how to best structure
- 35:36information for consumption by LLMS and then running tests
- 35:40like, hey, is it actually working?
- 35:41Do we, are we just seeing the results that we expect?
- 35:45You know, I think one of the, the, the questions and
- 35:48challenges is, is the LL Ms. you know, sort of have a, you know,
- 35:52specific training schedules. And so we know that they consume
- 35:55information and get better with inferencing.
- 35:58But I think that's the, the, the key thing is, is it's unclear
- 36:02what the lag looks like, how much you push information and
- 36:04how long before that shows up in the LL, in the LLM results.
- 36:09But we've been doing a lot of testing to to, you know, as
- 36:13we're building the product today, does it deliver?
- 36:15It says it is the information showing up.
- 36:16And I think, you know, we're starting to see more a lot of
- 36:20investment in sort of, you know, AEO or Geo tooling in the same
- 36:26way we saw with SEO to provide you some sort of qualitative
- 36:30assessment of the, you know, your data information.
- 36:34And one of the, one of the things that we've been working
- 36:37on toward working towards building is product content like
- 36:41evaluation, AI evaluation tool right inside PSM.
- 36:45So that way we're providing you assessments on, you know,
- 36:49there's easy stuff where you're missing data and where the gaps
- 36:52are, but also, you know, starting to get better and that
- 36:55sort of qualitative and measures around data structure and what
- 36:58have you and data, you know, content quality.
- 37:02So that way you're, you understand immediately how well
- 37:08suited your information is for consumption in the L in the Ln.
- 37:13So I think there's a lot, there's a lot to unpack there,
- 37:15but I think, you know, we're, we're excited and, and certainly
- 37:18it aligns really well with our sort of engineering philosophy
- 37:22is, you know, start small and, and experiment and learn along
- 37:25the way. Yeah.
- 37:27And I, I think you're 100% right about I'd, I would urge any
- 37:32businesses out there who haven't started playing around or I'm
- 37:36not doing much to use it more and more just to understand how
- 37:39it works in principle and see where information gaps go.
- 37:41I mean, I've, I've done it myself, my own consultancy.
- 37:44I want to know, do I get visible for key things about who are,
- 37:47who are the recommended independent E com consultants in
- 37:50the UK? Who are the top Reed platform?
- 37:51And do you know what I have learned very quickly is if you
- 37:54keep challenging challenge saying I had it on a couple of
- 37:57queries that I should be found for because I'm top in Google
- 38:00on. And I was like, why didn't you
- 38:01consider digital juggler? Why not?
- 38:04Did you know that X and I now when I check and I've done it
- 38:07through through like hidden IPS as well.
- 38:10So it's not just on my machine. So they're not detecting me
- 38:14spamming all the time, but I'm now included in answer results
- 38:17that I wasn't before because I'm suggesting.
- 38:20And then it's gone off and looked at my website.
- 38:22It's cruelty. I found the influence on.
- 38:23Oh, OK, you're now in my consideration set.
- 38:26So actually I think the the time to response of updating is and I
- 38:31know that's that in some ways is is slightly concerning because
- 38:35it could also use crap information, but.
- 38:37Sure. Yeah, yeah, yeah.
- 38:39No, you know, that's a, that's a, that's a great thing to hear.
- 38:43You know, and one of the things, and this is very much sort of an
- 38:45inside baseball thing is, you know, you look at sort of
- 38:49traditional enterprise software companies like us, you know, the
- 38:53gatekeeper on that type of recommendation has always been
- 38:56the analyst firms, right? And there's a whole game you
- 38:59play to be to, to, to, to score effectively in the analyst
- 39:04firms. And so now, you know, we've seen
- 39:07this is, is the, the, the democratization of that sort of
- 39:14gatekeeper piece is happening in real time in the LLMS all the
- 39:18time. And so we're, you know, we, we
- 39:20do very similar things to to in some ways to trigger the machine
- 39:24to go to go learn more about specific areas.
- 39:27So that's, that's very useful. Yeah, you know, and it's, it's
- 39:31in some ways it, in some ways it actually changes like some of
- 39:35our marketing tactics because of the things that we know the LLMS
- 39:39want to see. They want to see things like
- 39:40press releases and G2 reviews and what have you.
- 39:43And so you go and, and work on those.
- 39:45And so I think that applies to every business, B to B or B to
- 39:48C. Yeah, exactly.
- 39:51The the the credible external sources in the same way that
- 39:55Google's link graph has been incredibly important and an
- 39:57incredible extent, not just volume of links like it used to
- 40:00be in the Wild West days of link files, but now credible
- 40:04authoritative links from reputable domains still is one
- 40:08of the strongest signals and increasingly like if it's it's I
- 40:11do your do your foundations right essentially.
- 40:16So. The, the next question I've got
- 40:19is around the, the classic crystal ball thinks, yeah, you,
- 40:24you specialized in B to B and you work with some complex B to
- 40:29B businesses as well. What is your broader 2026
- 40:33commerce predictions angle for from AB to B and links that?
- 40:36What do you think will be the biggest surprise in terms of
- 40:39like B to B trends for this year?
- 40:43That's a good question. You know, I think the a couple
- 40:48of things, certainly one is I think answer engines have the
- 40:53opportunity to become sort of the new product detail page,
- 40:56right? And so, you know, as we've been
- 40:59talking about all throughout this, this discussion, you know
- 41:02that product information that populates the details around
- 41:05your products is going to be critical in the answer, right,
- 41:08because that's people where people are are going to to find
- 41:12and learn information, right. So that becomes a, you know, you
- 41:15know, we've all commerce has always been sort of a multi
- 41:18channel thing. Yet that's another product
- 41:19detail page that we have to, to think about from in a, in a BB
- 41:22commerce piece. And do you think that that AI
- 41:27sort of, as we, you know, compresses the funnel from
- 41:30discovery comparison configuration can all happen
- 41:33much, much faster and maybe in one or two steps or as today,
- 41:36those are, you know, much more elongated, right, 'cause the
- 41:40answer engines aren't, you know, putting in a query, then hitting
- 41:43the filter and narrowing the results.
- 41:45And, you know, and all the sort of sort of manual things that we
- 41:48do today, they're processing the information and doing that on
- 41:51the fly. I think, you know, in general, I
- 41:56think marketplaces, this really is A, a potential to disrupt
- 42:00sort of traditional marketplaces, you know, whether
- 42:04they be B to C marketplaces or B to B because I think AI is the
- 42:09potential to shift loyalty away from channels to outcomes,
- 42:13right? In, in, in, in in some ways
- 42:15disrupt the transaction. I mean, one of my big
- 42:18predictions, and maybe this is more BC is, is the, the, you
- 42:24know, the problem that the, the open AIS of the world have is
- 42:27they need $50 billion revenue opportunities.
- 42:30And there's just not that many in the world, but transactions
- 42:33taking the transaction and taking a, a cut of that is a
- 42:36huge opportunity. And I think we're going to see
- 42:38more of that in the short term, you know, rather than, you know,
- 42:41you know, in addition to likely sort of traditional ad models,
- 42:44which just turned AI into television.
- 42:48But I think that the, the, the interesting opportunity here and
- 42:52maybe the biggest surprise is there's a, there's an
- 42:54opportunity for B to be, to really lead the adoption curve
- 42:57of AI in digital commerce because the, the complexity and
- 43:03nature of their products are well suited to the types of
- 43:09outcomes and, and responses that LLMS are really well suited to
- 43:13deliver and well suited to understand.
- 43:17And there, as we talked about with quotations and other
- 43:20pieces, there's a ton of workflow elements in the B to B
- 43:24buying process that are right for automation with AI.
- 43:28And so I think those two elements we, you know, really,
- 43:31you know, say to say to me that that, you know, B to B could
- 43:33play a strong role in leading the adoption, practical adoption
- 43:37of AI in the digital commerce world for sure.
- 43:40Interesting. We Paul and myself recently did
- 43:43our our general not be to be specific trends of 2026.
- 43:47And I'm looking forward to going back in six months and seeing,
- 43:50seeing what what we were actually talking sense on and
- 43:53what we're talking nonsense on. It's always fun, isn't it?
- 43:56Look, Brian, I really appreciate you coming on.
- 43:58I know you're very busy, always a pleasure to hear what you're
- 44:01up to and also your perspective on the market.
- 44:04For anyone listen who wants to learn more about Elastic Path,
- 44:08the product capabilities and the B2B solutions, Brian's contact
- 44:12details are on the landing page, as is a link to the Elastic Path
- 44:17website where you can submit enquiries.
- 44:19So, yeah, thanks for joining us today.
- 44:22James, really thanks to that for having me enjoy the
- 44:24conversation. And thanks as always everyone
- 44:28for listening. Keep your ears appealed.
- 44:29For next week's episode, you're listening to the Inside Commerce
- 44:33podcast. Hosted by experienced
- 44:35consultants James Gerd and Paul Rodgers.
- 44:38Inside Commerce is your fast track to digital knowledge
- 44:40feature in interviews with leading e-commerce retailers,
- 44:43industry thought leads and the C-Suite key tech companies to
- 44:46explore the latest digital strategy, customer experience
- 44:48and tech. Discover what the latest trends
- 44:51really mean and get practical advice based on experience to
- 44:54help fine tune your E Comm strategy and tech.
- 44:57Follow us on LinkedIn and reach out to James of Paul if you'd
- 45:00like to discuss an ecommerce project.
- 45:02We hope you enjoy the latest episode.