Latest / The Adults In The AI Room / Earley AI Podcast - Episode 74: AI in B2B Commerce: From Data Challenges to Differentiation

Transcript

Earley AI Podcast - Episode 74: AI in B2B Commerce: From Data Challenges to Differentiation · The Adults In The AI Room

• 0:01
  SPEAKER_00: This is the Early AI Podcast hosted by Seth Early.
• 0:04
  SPEAKER_00: Join us as we delve deep into the passions, expertise, and experiences of thought leaders and practitioners to talk about what's possible with artificial intelligence.
• 0:14
  SPEAKER_00: The Early AI Podcast is sponsored by Early Information Science, your digital transformation journey with design and deployment of innovative technology solutions, as well as vector.
• 0:26
  SPEAKER_00: Now enjoy the show.
• 0:28
  SPEAKER_01: So welcome to the Early AI Podcast.
• 0:31
  SPEAKER_01: I'm your host, Seth Early, and in this episode, we want to explore how AI is impacting and reshaping the world of B2B commerce.
• 0:39
  SPEAKER_01: From customization to replatforming to product data management and customer experience, AI is fundamentally changing how businesses operate.
• 0:46
  SPEAKER_01: We'll dive into the challenges and the opportunities and the strategies that leaders need to stay ahead.
• 0:51
  SPEAKER_01: Today, our guest is Rudy Abbottball.
• 0:55
  SPEAKER_01: Did I get that right?
• 0:56
  SPEAKER_01: Perfectly right.
• 0:56
  SPEAKER_01: Okay, uh, who's a recognized expert in B2B commerce and AI?
• 1:00
  SPEAKER_01: Rudy's worked with leading global companies to tackle some of the toughest problems in digital transformation, from product information management to large-scale adoption strategies.
• 1:09
  SPEAKER_01: He's very passionate about making AI practical, accessible, and effective for real-world business challenges.
• 1:14
  SPEAKER_01: Rudy, welcome to the show.
• 1:15
  SPEAKER_01: Thank you.
• 1:16
  SPEAKER_01: Thank you for having me.
• 1:17
  SPEAKER_01: So the world is changing quite a bit.
• 1:19
  SPEAKER_01: You've been in this space for a very long time.
• 1:22
  SPEAKER_01: So let's start with kind of the big picture.
• 1:24
  SPEAKER_01: From your perspective, how is AI transforming B2B e-commerce and where are you seeing the most immediate impact on adoption, on customer engagement, on operations?
• 1:34
  SPEAKER_01: So, what's your big picture perspective?
• 1:36
  SPEAKER_01: And I know there's a lot of efforts underway.
• 1:38
  SPEAKER_01: There's a lot of well-publicized failures, but we also know their successes.
• 1:43
  SPEAKER_01: So give us your sense of uh where the industry is and what's going on.
• 1:47
  SPEAKER_03: The impact is uh is really um being seen right now with the probably the biggest pain point was for B2B e commerce to to produce content.
• 1:56
  SPEAKER_03: You have tons of specs to consider, tons of elements to consider, and AI came um a bit saving the B2B word because it was extremely tough when you have 100,000 SKUs and catalog to produce all that content and the expertise that comes with it.
• 2:12
  SPEAKER_03: Now it's getting much, much better, much easier for B2B Commerce Manager to produce such content.
• 2:18
  SPEAKER_03: At the same time, it's about the tools that are joining there and integrating within the their own digital ecosystem.
• 2:26
  SPEAKER_03: So we've seen tools that are meant for B2C come to uh B2B and bringing that agility, bringing that flexibility of AI integration.
• 2:36
  SPEAKER_03: So obviously the biggest example is Shopify, and then you you have other tools that are that are more meant to be able to centralize all that data in one place and start building quickly uh POCs.
• 2:49
  SPEAKER_01: So so you mentioned Shopify, and what is it, what is it specifically about Shopify that uh you believe is kind of an exemplar, or give me your uh dive into that a little more deeply.
• 2:59
  SPEAKER_03: Yeah, it's it's about how you build it and the the the ecosystem.
• 3:04
  SPEAKER_03: Anyone in the B2B will say, well, it's not perfect, it's not incomplete, but the big thing is it's flexible, and for the team internally, it has been a great help.
• 3:15
  SPEAKER_03: B2B in general, it's extremely hard to hire someone.
• 3:18
  SPEAKER_03: Suddenly you get a platform that everyone knows, and I'm able to hire someone coming from the B2C knowing the experience, and that person is really able to do all kinds of modification and change and updates within the platform because it's easy to use.
• 3:32
  SPEAKER_03: And that's a big change with no programming requirements.
• 3:36
  SPEAKER_03: Obviously, it's it's the power of the ecosystem that they have, like the the amount of app that are being built, and some of them being built for B2B, are kind of adding so many features, so many possibilities that will have required hundreds of thousands of dollars of programming on another platform.
• 3:54
  SPEAKER_03: So that's a big change.
• 3:56
  SPEAKER_03: And I think it's applying also to all the different B2B ecosystems that goes along, like we see the change in PIM and search engine and though that are kind of more aligning and integrating with Shopify right now.
• 4:08
  SPEAKER_01: So that's interesting.
• 4:08
  SPEAKER_01: And you know, what you're saying is because the uh the is being woven into the DNA of these products and these tools, that it becomes much simpler to make modifications, to customize, to adapt to your particular environment.
• 4:22
  SPEAKER_01: Plus, there's that ecosystem that's all being accelerated by by AI and by the the uh possibilities.
• 4:28
  SPEAKER_01: So that's kind of a code versus configure perspective, right?
• 4:32
  SPEAKER_01: And and one of the things that we had talked about uh previously was in in our prep was vibe coding and load code development and making these development tools more accessible.
• 4:42
  SPEAKER_01: So, how do you see that trend shaping how companies are doing, uh are innovating and replatforming and so on?
• 4:49
  SPEAKER_01: You just talked about the fact that it's faster and easier to do things.
• 4:53
  SPEAKER_01: And how does vibe coding fit in?
• 4:55
  SPEAKER_01: Because you're in the in the Shopify environment, you're you're leveraging the ecosystem of applications that have been created, you're leveraging the ability to configure that application, but then how does vibe coding fit in with that?
• 5:06
  SPEAKER_03: Frankly, it has been my obsession in the last four or five months.
• 5:10
  SPEAKER_03: I haven't stopped.
• 5:11
  SPEAKER_03: And the biggest thing is whenever we add requirements and though, or in B2B commerce, you have very specific segments all the time.
• 5:21
  SPEAKER_03: What if that customer, what if that kind of customer?
• 5:24
  SPEAKER_03: So at the beginning we're saying with the cost of development and all, is it really is there an ROI for it?
• 5:30
  SPEAKER_03: And all we're not sure.
• 5:32
  SPEAKER_03: The second part is there's elements like wording that are typical of specific industry.
• 5:37
  SPEAKER_03: So I used to be at Sonypar, for example, and we have specific words that were that were and specific ways of working that are typical of distributors, electrical distributors.
• 5:48
  SPEAKER_03: With vibe coding, I'm able to build at least a wiframe within the conversation that I had.
• 5:55
  SPEAKER_03: So I can take all the conversation that I recorded, for example, with the different uh clients, with the different uh stakeholders, bring those conversations and regrouping into okay, well, let's build the specs for that.
• 6:08
  SPEAKER_03: And I'm able to build the wiframes, even though they're not perfect or fully functioning.
• 6:13
  SPEAKER_03: Let's say I bring 60, 70% in one place, and that's suddenly becomes a point of conversation.
• 6:20
  SPEAKER_03: And the fact that I'm customizing that fast for my team makes it even more uh easy for them to adopt because they're gonna say, Well, you finally understood what we want, you're able to frame something perfectly for us.
• 6:34
  SPEAKER_03: Now we're gonna start pushing and use and be involved in it.
• 6:37
  SPEAKER_03: So imagine the double game internal reducing the cost.
• 6:41
  SPEAKER_03: Now, is the coding of it perfect, fully integrated?
• 6:45
  SPEAKER_03: No.
• 6:45
  SPEAKER_03: Within one year, it will be there.
• 6:47
  SPEAKER_03: That's one thing for sure.
• 6:49
  SPEAKER_03: But now that I have an interface like Shopify, where I'm able to build standard custom applications, I'm able to integrate those custom applications that I've vibe coded, let's say, into a known network.
• 7:02
  SPEAKER_03: So I my possibility of the optimization, which was the toughest point, is already reduced in cost and all.
• 7:09
  SPEAKER_03: So I build my own solution designed by my people with my own kind of way and approach.
• 7:15
  SPEAKER_01: So that's a really interesting point.
• 7:17
  SPEAKER_01: It it's it's in the old days of gathering requirements and going through numerous interviews and discussions and analysis, you would then turn something over to uh a team of people that might develop in you know just a UI um based system, something like a Figma, but it still took a lot of work and it took a lot of cycles and took a lot of time.
• 7:39
  SPEAKER_01: And and many times, you know, you want to iterate faster than that.
• 7:43
  SPEAKER_01: You want to speed up the clock speed of the ability to do that kind of innovation and that uh adaptation.
• 7:49
  SPEAKER_01: And many times customers they need to see something in order for them to really understand it, right?
• 7:55
  SPEAKER_01: And many times that came very late in the cycle, you know, when you're doing requirements and building prototypes or POVs or whatever using standard techniques versus some of the newer techniques.
• 8:06
  SPEAKER_01: And so now it it's much easier to get people to focus on something, to provide the feedback, to say, yeah, I get that.
• 8:13
  SPEAKER_01: That's what I want, that's what I don't want.
• 8:15
  SPEAKER_01: So it really does close down that cycle or speed up that cycle of development and innovation.
• 8:21
  SPEAKER_01: Uh, so I see that as being incredibly valuable that you're iterating faster, you're developing faster, but you're you're getting to a shared vision of where you need to be more quickly using these types of tools.
• 8:33
  SPEAKER_01: Even if you're doing, even if you're still hiring a dev team to do the development, they're of course speeding up their cycles using these tools.
• 8:41
  SPEAKER_01: But but putting these in the hands of the business person and the analyst is incredibly powerful.
• 8:47
  SPEAKER_03: Yeah, and it's uh really the conclusion of design thinking right there.
• 8:50
  SPEAKER_03: But even more is that you limit the amount of frustration.
• 8:54
  SPEAKER_03: The AI is trained on the documentation of what's possible to do on the let's say Shopify API.
• 9:00
  SPEAKER_03: So already I can flag from the get-go, that's not feasible.
• 9:04
  SPEAKER_03: It will require extra.
• 9:06
  SPEAKER_03: So just in terms of, you know, in terms of uh agile development, I can do my shirt sizing right away and say, this is a big one, maybe we'll do it a bit later, but this is uh at least where we're going.
• 9:18
  SPEAKER_03: Less surprising, less frustration uh for everyone and for the programmers, at least they have something documented so uh everyone is gaining from it, definitely.
• 9:28
  SPEAKER_01: Yeah, and and so what you're saying is that the AI tools, when you when you specify the platform and the technology and the ecosystem, there's enough uh embedded knowledge that it understands, or you support it with you know with documentation to say what's possible, what's feasible, and to eliminate things that are going to be too costly and time consuming right off the bat, right?
• 9:49
  SPEAKER_01: Because it understands the parameters, understands what can and can't be done.
• 9:54
  SPEAKER_03: Something a human could have not figured out without testing and implementing because it's too much information.
• 9:59
  SPEAKER_03: It's it wouldn't make sense.
• 10:01
  SPEAKER_01: Right, right.
• 10:02
  SPEAKER_01: So yeah, yeah, and I'm I'm astounded every day when I use LLMs, when I'm giving uh it complex interviews and discussions and uh inputs and and so on, and asking very deep and detailed questions and and getting very sophisticated responses.
• 10:18
  SPEAKER_01: So it is kind of uh amazing.
• 10:20
  SPEAKER_01: And and the ability to leverage those with specific company requirements and specific company information is really where it's coming from.
• 10:28
  SPEAKER_01: And you know, we talked about this before.
• 10:30
  SPEAKER_01: You had talked about the idea of of differentiation uh versus standardization, right?
• 10:36
  SPEAKER_01: And you're going to be able to, if you do what everybody else does, you get some efficiency and you can standardize, but you're not going to get competitive advantage from that, right?
• 10:45
  SPEAKER_01: You you need to differentiate based on your knowledge of the customers, your knowledge of the environment, your knowledge of your solutions, and so on.
• 10:52
  SPEAKER_01: Uh, how do you kind of balance that with um uh the efficiencies that are coming from uh from vibe coding and from AI with uh the need to be able to do things differently, right?
• 11:04
  SPEAKER_01: So if you do what everybody else does, you get efficiency, but you don't get differentiation.
• 11:09
  SPEAKER_01: Talk a little bit about that.
• 11:10
  SPEAKER_03: But it's a yeah, you know, it's the logic of the vibe coding is able to replace the work that you've done, maybe use with some programmers or wireframing and all.
• 11:20
  SPEAKER_03: But all the insight within the phrasing, within the the way that it's done and all, it needs to come from you.
• 11:28
  SPEAKER_03: You still need to have someone that is a product owner with a great vision, because that's the sole person that is able to infuse.
• 11:34
  SPEAKER_03: But the problem that we had before is that the person with the knowledge of the business didn't have the knowledge of the technical.
• 11:42
  SPEAKER_03: So by the time we pass the information one to another, well, you get something even more generic or that doesn't work.
• 11:49
  SPEAKER_03: Now, because it's coded, it's let's say coded uh between brackets, right?
• 11:54
  SPEAKER_03: But because it's built by people internally that have that knowledge, well, they're able to quickly add more things that are very specific and the way of working specific of the company.
• 12:05
  SPEAKER_01: Right, right.
• 12:06
  SPEAKER_01: And and that is that essential way of doing business is unique.
• 12:10
  SPEAKER_01: Even if two companies are selling to the same market and they're selling the same products, they still have different looks and feels and voice and tone and value.
• 12:20
  SPEAKER_01: And it's understanding what those differentiators are and understanding the very unique needs of your segment and your target customer and how to serve them.
• 12:29
  SPEAKER_01: So that's kind of an interesting uh thing when you think about it.
• 12:32
  SPEAKER_01: You're getting all these efficiencies, but you still need to figure out what that unique relationship is and what those unique uh differentiators are.
• 12:41
  SPEAKER_01: And it's kind of the sum total of everything that you know about the marketplace, everything you know about your customers, about the competitors, about the ecosystem.
• 12:49
  SPEAKER_01: And that is what needs to be captured in this mechanism that you're that you're referring to, where you're getting that business knowledge into the hands of uh developers or getting it into the products themselves and the and the technology itself in order to be able to stand out from the marketplace.
• 13:08
  SPEAKER_01: So it now seems that the uh uh the time to differentiate is is shorter.
• 13:13
  SPEAKER_01: Uh the ability to do that is is easier, but what's holding companies back?
• 13:18
  SPEAKER_01: Like what is it that so so the the ability to do this is fantastic, and the ones who do it most effectively and most quickly are going to be able to win.
• 13:28
  SPEAKER_01: But what is holding companies back?
• 13:30
  SPEAKER_01: What are what are the things that are getting in the way of being able to realize that that vision?
• 13:34
  SPEAKER_03: It's always the same, right?
• 13:36
  SPEAKER_03: It's uh the internal resistance, but even sometimes in many companies, I see the employees using Chat GPT on a daily basis, but on a very basic level, right?
• 13:46
  SPEAKER_03: The improve, correct my mistake, fix my email, or something like that.
• 13:50
  SPEAKER_03: And nobody many times I'm telling uh teams that I'm meeting, but did you try going further with with Chat GPD?
• 13:58
  SPEAKER_03: You know, it can do that.
• 13:59
  SPEAKER_03: You can upload a screenshot, you can upload your your file in it.
• 14:03
  SPEAKER_03: Yeah, I don't know.
• 14:04
  SPEAKER_03: Nobody told me, and oh.
• 14:06
  SPEAKER_03: And the core problem is you have those tools, you're not training properly your team to use it, and you're saying, Oh, that's it, we have AI, we have uh subscription of 20 bucks per month for ChatGPD.
• 14:18
  SPEAKER_03: That's not it.
• 14:19
  SPEAKER_03: There's a lot of departments that are that needs to use it very specifically, and and that training needs to happen.
• 14:27
  SPEAKER_03: And so what's missing today it's saying, first of all, cut the noise, like make sure that the team stays focused on what exactly are you is your mission as a company.
• 14:38
  SPEAKER_03: Don't say we cannot go there because we're gonna uh we're not uh competitive enough and just stick to what you're doing right now and what you're good at, and from there on train your team to be even more efficient, train your AI to be even more efficient by infusing the DNA, your own data, and all and that's the core conversation we're having right now.
• 14:59
  SPEAKER_03: It's uh you know, you mention in the ragged how do you implement all of that within the the company?
• 15:05
  SPEAKER_03: And many times the point of resistance comes from a simple remark that is see, it made a mistake.
• 15:11
  SPEAKER_03: Well, let's look at the 80% that is absolutely correct and that you've delivered in five minutes.
• 15:19
  SPEAKER_03: Did you count correctly?
• 15:20
  SPEAKER_03: Did you add the right data first before saying it made a mistake?
• 15:24
  SPEAKER_03: Maybe you didn't tell him, just like you didn't tell a junior how to do the job and you blame him for something.
• 15:30
  SPEAKER_03: Well, right exactly that.
• 15:31
  SPEAKER_01: Right.
• 15:32
  SPEAKER_01: I was I was just gonna call out that analogy.
• 15:34
  SPEAKER_01: You know, if if you judge an employee by the fact that they made a mistake, right?
• 15:39
  SPEAKER_01: You wouldn't have any employees, right?
• 15:40
  SPEAKER_01: Because you say, Oh, they made mistakes.
• 15:42
  SPEAKER_01: So you so you have to expect that this is gonna happen.
• 15:45
  SPEAKER_01: But I think the the bigger point is what is the root cause and of that mistake, right?
• 15:50
  SPEAKER_01: What is the what is the source of that?
• 15:51
  SPEAKER_01: And that kind of gets into the data side, right?
• 15:54
  SPEAKER_01: Because if we are leveraging uh retrieval augmented generation, which is using your data, using the LLM to the large language model, the chat GPT, to interpret what the user is looking for.
• 16:06
  SPEAKER_01: And then you go and look for that in your data and then retrieve it and bring it back and then make it more conversational, that's really where uh we're gonna get the tremendous, the most value, the tremendous value from leveraging our own data.
• 16:18
  SPEAKER_01: But then that begs the question of um how is your data, you know, and what's the data uh uh situation?
• 16:25
  SPEAKER_01: So, you know, as these uh organizations are generating more uh more customer data, but product data and all sorts of information, you know, they're struggling with the structure, they're struggling with the quality, they're struggling with ownership.
• 16:38
  SPEAKER_01: And and in order for retrieval to work, and you have to think of it at some level as search.
• 16:44
  SPEAKER_01: If you have crappy data, if you have, if you if you're missing the content, if you're missing the data, you're not gonna get a good results.
• 16:51
  SPEAKER_01: And the same thing is happening here.
• 16:53
  SPEAKER_01: So, how are you seeing that kind of evolve?
• 16:56
  SPEAKER_01: And I'll I'll tell you a quick story.
• 16:58
  SPEAKER_01: I was talking to a company about fixing their data, and um, they had a real a lot of challenges, lots of acquisitions, lots of inconsistency.
• 17:06
  SPEAKER_01: And so we talked about, you know, how do you get your product catalog normalized and how do you get your attributes normalized and all of this?
• 17:12
  SPEAKER_01: Uh, and then they um reached out a couple weeks later and they said, but we're going in a different direction.
• 17:16
  SPEAKER_01: And they said, Oh, what are you doing?
• 17:17
  SPEAKER_01: He said, Well, we're deciding to build a bot, right?
• 17:19
  SPEAKER_01: So they brought in a company to build a bot.
• 17:21
  SPEAKER_01: That bot's not going to work very well if you don't have the right data.
• 17:24
  SPEAKER_01: Well, uh, talk a little bit about that and talk about how companies are understanding this or not understanding this and where they are in that kind of journey and lifecycle.
• 17:34
  SPEAKER_01: Well, it goes both ways, right?
• 17:35
  SPEAKER_03: We don't it goes both ways frequently, is that sometimes the company will say, I don't have the data, so I shouldn't start AI, I should build my debt.
• 17:45
  SPEAKER_03: On the other hand, I didn't know that I needed to train and structure my data properly, so I badly trained the AI.
• 17:52
  SPEAKER_03: So I'll go with the two scenarios right there.
• 17:54
  SPEAKER_03: It's that a lot of time companies are waiting to get the data right before starting the AI, which is a big mistake.
• 18:03
  SPEAKER_03: If you have 10%, 15%, how many CRMs today are complete or incomplete?
• 18:09
  SPEAKER_03: How many times can you repeat to the sales team, please fill up everything?
• 18:12
  SPEAKER_03: That's not them, you cannot ask them for that to have perfect completeness.
• 18:17
  SPEAKER_03: That's not in the culture of it, and they have other priorities.
• 18:21
  SPEAKER_03: Perfectly normal.
• 18:22
  SPEAKER_03: Let's just focus on 15, 20, how can AI enrich that part?
• 18:27
  SPEAKER_03: And on the opposite side, are you exactly giving and producing tools where you can structure the data?
• 18:34
  SPEAKER_03: Do you have a uh master data management in place to structure all that data?
• 18:39
  SPEAKER_03: If you say in 2025 you still have some of the company information in an Excel sheet, you have a big problem, a really big problem right there.
• 18:48
  SPEAKER_03: If it's not structured today, you have tons of tools, nearly free, that you can be used, you need to get on that and all and you'll be amazed, even if incomplete, how AI can come and help you from there.
• 19:02
  SPEAKER_03: You're not the only one, you're not the most unique company, contrary to what a lot of small business think.
• 19:08
  SPEAKER_03: It's not that unique.
• 19:09
  SPEAKER_03: Uh AI can help you on that.
• 19:10
  SPEAKER_01: So it's a really good point.
• 19:12
  SPEAKER_01: So, so in the one case where they're with this company saying, well, we're going to build a bot first, that's going to reveal a lot of the challenges that they have with their data, right?
• 19:19
  SPEAKER_01: And so then that'll help them pinpoint where they need to make changes.
• 19:24
  SPEAKER_01: Or if they build it with a narrow enough focus and they do have a curated data set, great.
• 19:30
  SPEAKER_01: But I think the other point you make is very good is that the AI can help with data quality.
• 19:36
  SPEAKER_01: You can improve your data with AI.
• 19:38
  SPEAKER_01: What do you need to have in place to do that?
• 19:40
  SPEAKER_01: Because you still need a reference architecture, right?
• 19:43
  SPEAKER_01: It's not necessarily going to know what your product data uh structure should look like, but you can you can prompt it to build some of those things.
• 19:50
  SPEAKER_01: So talk a little bit about what's necessary in order to use AI to improve uh data quality.
• 19:56
  SPEAKER_03: So dears, you know, we were talking about Excel sheets, right?
• 20:00
  SPEAKER_02: No.
• 20:00
  SPEAKER_03: And and I uh I was working on building a data model for uh for a PIM, and I kept looking into the data once and for all.
• 20:08
  SPEAKER_03: One of the things that the AI or the AI helped me is to say, this seems to be a great access in order to get SKU uniqueness.
• 20:18
  SPEAKER_03: And how we build it is by doing what we call collision test, which is pretty much comparing, like if you choose that axis, how many unique SKUs do you get at the end?
• 20:28
  SPEAKER_03: Just with the AI, that amount of time that I should have spent myself will have taken weeks because it's building Excel model and all one off top of another of another.
• 20:38
  SPEAKER_03: With the AI, it built this within two, three hours.
• 20:41
  SPEAKER_03: I was able to run five different tests.
• 20:43
  SPEAKER_03: From those tests, I was able to come back and say, by the way, for that collection or for those specific type of product, those are the access that we need.
• 20:53
  SPEAKER_03: Okay, and here's why.
• 20:55
  SPEAKER_03: And here's why, following the test that uh AI proposed, they settle the conversation internally.
• 21:02
  SPEAKER_03: And that is also one of the blockends in making those big architecture is the conversation.
• 21:08
  SPEAKER_03: Everyone sees it from their own point of view, but suddenly you get the data at a fast pace, and you're able, you know, the the quote like uh if you if we have data, let's go with data.
• 21:18
  SPEAKER_03: But if we have opinion, let's go with mine, right?
• 21:20
  SPEAKER_03: That that quote from the uh C.
• 21:22
  SPEAKER_03: I like it.
• 21:22
  SPEAKER_03: Yeah, yeah, that's exactly that.
• 21:24
  SPEAKER_03: You can settle a lot of conversation by bringing that data within AI, get the insights from there, and maybe you'll have differences or or don't agree, but at least you have some point of conversation and eliminate it, eliminate the gossip the years.
• 21:39
  SPEAKER_03: Say, I thought we were doing like this for for a long time and all.
• 21:43
  SPEAKER_03: Right.
• 21:43
  SPEAKER_01: Can you can you dig dig into that testing and that analysis a little bit more?
• 21:48
  SPEAKER_01: What specifically were you doing with the spreadsheets and the product data?
• 21:52
  SPEAKER_01: And then how were you using AI?
• 21:54
  SPEAKER_01: Maybe you could just kind of explore that for a moment and go a little bit deeper.
• 21:58
  SPEAKER_01: So uh maybe talk a little bit about.
• 22:00
  SPEAKER_01: What you started with, and then what tests you were trying to perform, and then how the AI helped with that.
• 22:06
  SPEAKER_01: So maybe re-repeat what you said, but maybe go a little bit more in depth, if you would.
• 22:09
  SPEAKER_03: Sure.
• 22:10
  SPEAKER_03: So in that example, you have some kind of products that are very specific to certain niche, but the products are very much alike, right?
• 22:19
  SPEAKER_03: So how am I supposed to customize and say this is that product can be used for that specific application?
• 22:27
  SPEAKER_03: That product can also be used for that specific application.
• 22:31
  SPEAKER_03: How you determine that?
• 22:32
  SPEAKER_03: It's based on the specs, the size, the fit, and everything.
• 22:36
  SPEAKER_03: The way that you approach it, if I look at the extract of all the data, well, it's hard for me to understand all the single applications that can exist because I'm not an expert in the field.
• 22:47
  SPEAKER_03: It's impossible for me to start, let's say, applying for to hospital and ask also um uh EMS company, how you use it, and all the different types of users.
• 22:58
  SPEAKER_03: But with AI, because you load up the specs that are asked in the contracts from the hospital, you load up the contracts or the specs that are asked from the different companies, you ask AI, you know my product, which one do you think can fit within it?
• 23:14
  SPEAKER_03: And because it has all that data and is able to compare, instead of looking at Collins building pivot table and all four hours, it did this for you right away.
• 23:23
  SPEAKER_03: So if I'm able to suddenly get a grasp of because I structure the data, I'm able to sell better my products and even reach new market that I never thought of.
• 23:34
  SPEAKER_03: So uh a product to clean up hospital room, that's great.
• 23:38
  SPEAKER_03: But can you did you think that because it's eliminating uh viruses and all, you can use that same product to clean up, let's say, some place where they need a better hygiene, like the where they cut meat, for example, or things like that.
• 23:52
  SPEAKER_03: Right.
• 23:52
  SPEAKER_03: Well, do you have thought about it?
• 23:54
  SPEAKER_03: Yes, maybe one day, but imagine you get all those answers because you put the data in one place, you suddenly have even more perspective from the outside world.
• 24:04
  SPEAKER_03: And that's the biggest pain point in the B2B, is that we're very close, we're very niche, we just know each other from the same industry, and that's how you you grow your network and your sales from there.
• 24:15
  SPEAKER_01: So that's interesting.
• 24:16
  SPEAKER_01: So you're you're you're saying that you know we'll load our product catalog, we'll load our specifications, our engineering drawings and diagrams, and product details and product manuals, and whatever the guide, the design guides, anything that I have about the products, which which are which is way more than I'm gonna put on a product detail page, right?
• 24:36
  SPEAKER_01: It's gonna be the supporting product specifications and details.
• 24:39
  SPEAKER_01: And then I upload the requirements or the um details around different segments of customers, different types of customers, say RFPs from different industries.
• 24:51
  SPEAKER_01: Then I can ask the AI, give me some new markets for these products, or give me some idea of where I can expand a category.
• 25:00
  SPEAKER_01: And you can uh if you upload your sales data and and so on, your your AI can identify opportunities to expand a category or to expand a market or to reach a new market and a new customer with specific messaging.
• 25:14
  SPEAKER_01: So that's really where that power you could, if you did that manually, as you say, it would take weeks and weeks and weeks.
• 25:20
  SPEAKER_01: And if we have this and we put it into the right environment and upload it, we can ask the large language model those very specific questions and get those results much, much more quickly.
• 25:32
  SPEAKER_03: Absolutely.
• 25:33
  SPEAKER_03: And even more is that you you can you can structure even the content of how you present the product on your website, right?
• 25:40
  SPEAKER_03: Am I selling that product the same way, whether it's a hospital or it's a meat factory, anything like that?
• 25:46
  SPEAKER_03: I'll present it differently, kind of push further in creating even assets and images specific to that industry, maybe, but that's another thing.
• 25:55
  SPEAKER_03: But at least having content with a specific vocabulary from and the lexicons from a specific and it's amazing.
• 26:02
  SPEAKER_01: Right, right.
• 26:02
  SPEAKER_01: Yeah, because because the hospital is going to talk uh very differently about their cleanliness and their hygiene and their uh attempts to keep that sterile environment very differently than a than a meatpacking plant.
• 26:15
  SPEAKER_01: That's the same problem at the fundamental basis, but you're using different language, different terminology, uh, different types of problems for each of those, even though the solution may be very, very similar.
• 26:25
  SPEAKER_03: And imagine internally, you have people that work with hospital for years, suddenly they have a new market, they have all that learning to do.
• 26:31
  SPEAKER_03: Well, I've taken maybe hiring someone, another sales team, and now you can readapt within your own marketing team, existing marketing team, all of that.
• 26:41
  SPEAKER_01: You can educate and support your salespeople with that new messaging and build an assistant that will help them uh sell in that market.
• 26:50
  SPEAKER_01: Because we're used to selling the hospitals, now they're gonna sell to a meat packing plant.
• 26:54
  SPEAKER_01: How do they how do they engage and how do they answer those questions and how do they become proficient in that?
• 26:59
  SPEAKER_01: And that's where the AI can support that as well.
• 27:01
  SPEAKER_03: Yeah, definitely.
• 27:02
  SPEAKER_01: That's great.
• 27:04
  SPEAKER_01: So let's talk a little bit about you know uh AI readiness.
• 27:08
  SPEAKER_01: And you you did hit on this uh a little bit earlier where you were talking about the fact that organizations are not necessarily training their people with in enough detail or nuance for them to use AI in their particular departments very specifically.
• 27:23
  SPEAKER_01: What what are you seeing?
• 27:24
  SPEAKER_01: How are organizations overcoming that?
• 27:27
  SPEAKER_01: There's a lot of training resources out there, but you know, one of my clients paid a lot of money for something to do a workshop and said it was very generic and didn't help them at all, and they spent a large sum of money on that.
• 27:38
  SPEAKER_01: How what are your thoughts about how to kind of go about that within an organization?
• 27:43
  SPEAKER_03: The most of the time when I talk to my my team, it's it's interesting because they they say, Oh, we're not trained enough with ChatGPD and how to use it.
• 27:52
  SPEAKER_03: I'm like, Did you ask ChatGPT to train you?
• 27:55
  SPEAKER_03: Yeah.
• 27:55
  SPEAKER_03: Think about this one because there's specific rules and all, and you can push it, and it needs to become something as a convenience for you.
• 28:04
  SPEAKER_03: The same way you're using your phone every day to make phone calls.
• 28:08
  SPEAKER_03: Well, any decision and all that you're making should in some way go through AI to refine it, to accomplish the message and all.
• 28:16
  SPEAKER_03: So the saying I'm bringing an expert that will train us, and now we'll all become experts in AI, that's impossible.
• 28:25
  SPEAKER_03: You need to do it on your own.
• 28:27
  SPEAKER_03: For that, you need to give time to your people, and you need to be willing to take the risk to have one specific project that will be AI-driven, and it's fine.
• 28:37
  SPEAKER_03: It's a POC, it might fail, it might not be as successful and all, but you have to let them deep dive on it and all.
• 28:46
  SPEAKER_03: And I'm very much in most of my job position and all, I'm always the guy that says we might not subcontract fully that whole project.
• 28:56
  SPEAKER_03: I want the team to learn, I want them to be helped by an expert, build a great relation with an expert, but I want everyone to raise at the same time, right?
• 29:05
  SPEAKER_03: So it's very important that we keep trying, we keep applying very specific.
• 29:10
  SPEAKER_03: Don't say AI in general, just say here's a problem, I think AI can help us fix.
• 29:16
  SPEAKER_03: If you see four or five Excel sheets somewhere running around and someone opening them, just pause for a moment.
• 29:23
  SPEAKER_03: Think maybe AI can help you on what you're doing every day, and that's how you improve the life of everyone and your employee, and that's how you make your business future-proof.
• 29:32
  SPEAKER_03: We keep talking about job being eliminated and oh, yes, it will happen.
• 29:37
  SPEAKER_03: Like, let's not lie to ourselves, and this will happen.
• 29:40
  SPEAKER_03: But on the other end, how ready and how much gain did you bring to your company and the value in exchange that you brought to your company and to your employees?
• 29:49
  SPEAKER_03: That's what matters today.
• 29:50
  SPEAKER_01: So you made an interesting premise or comment where you said it's really up to the individual to train or to get that some of that training to ask chat.
• 30:00
  SPEAKER_01: That takes a very specific kind of person who is proactive, um, you know, is active in learning, is looking to do these things, has that self-motivation and so on.
• 30:12
  SPEAKER_01: And that increasingly is an important characteristic to hire for, right?
• 30:16
  SPEAKER_01: I mean, it it really is today.
• 30:18
  SPEAKER_01: But I would also propose that you still need some structure around that to provide that environment.
• 30:24
  SPEAKER_01: You did say, you know, the company needs to provide time.
• 30:26
  SPEAKER_01: I would also think they need to provide a framework, right?
• 30:29
  SPEAKER_01: To say, here are the types of things that you should be asking, or here are the ways for other organizations, similar organizations, other departments are using this.
• 30:38
  SPEAKER_01: Now, how does that apply with you and your department?
• 30:41
  SPEAKER_01: And even getting some collective kind of brainstorming and and uh collaboration within the team within the department to do that.
• 30:49
  SPEAKER_01: So I agree that it does take the ability to do this in a self-motivated fashion, but it's also important for the organization to provide some structure around that and some foundational training to in order to catalyze that.
• 31:03
  SPEAKER_03: Absolutely.
• 31:03
  SPEAKER_03: We talked about it with the right, but we talked, we need to talk about it also with the politics that you put internally, how it's being used, and oh, I fully agree with you within the structure.
• 31:13
  SPEAKER_03: What's important in it, it's it's hard to define the structure because it's new for everyone.
• 31:18
  SPEAKER_03: Yeah.
• 31:19
  SPEAKER_03: By saying, here's a small project, let's see what it takes, let's see how much it took, what kind of level and interest, and all.
• 31:27
  SPEAKER_03: But to come back to your first point about being proactive and oh, you're right.
• 31:32
  SPEAKER_03: It's more important than education at this point when hiring someone.
• 31:36
  SPEAKER_03: It's more important that your experience in past company because whatever I did 10 years ago don't even apply today.
• 31:43
  SPEAKER_03: Nobody cares about me having experience, a five-year experience building my gentle website.
• 31:48
  SPEAKER_03: That's it, it's over now.
• 31:50
  SPEAKER_03: So I need maybe have great knowledge of e-commerce experience now, but hosting doesn't matter any anymore.
• 31:56
  SPEAKER_03: A lot of uh structuring the data, the design, the way that it was done before and all, doesn't even apply today.
• 32:02
  SPEAKER_03: But at least you need to get teams that are curious and a team that are willing to understand that there might be something that we no longer need and readapt your job right there.
• 32:13
  SPEAKER_03: But the way of hiring, the way of implementing and asking the person internally, how are you using AI for that?
• 32:20
  SPEAKER_03: And there's that.
• 32:22
  SPEAKER_03: Um, I'm inventing a new word, so it's exclusive today, but that AI shaming, right?
• 32:27
  SPEAKER_03: When someone sends you an email and it's super structured, and you can feel that that person didn't write that email.
• 32:33
  SPEAKER_03: Okay, let's let's be honest.
• 32:36
  SPEAKER_03: But at least he made the effort of cleaning up and having something that is structured enough to communicate better.
• 32:42
  SPEAKER_03: So you cannot reply, oh, nice chat GPD, here's my true answer.
• 32:46
  SPEAKER_03: It's structured, right?
• 32:48
  SPEAKER_03: And it's maybe the right people need to learn to prompt not to build very long email for no reason and keep it short and efficient.
• 32:55
  SPEAKER_03: But on the other end, it's very important that we encourage people internally to use it.
• 33:00
  SPEAKER_01: So if you I think you're making a really, really important uh point here is that people don't want to go through, you know, a long AI generated piece of content that does not have the insights and the the so what from human judgment, right?
• 33:17
  SPEAKER_01: Because it's one thing to use AI and it's one thing to another thing to overuse it, right?
• 33:22
  SPEAKER_01: And we have to be very careful about how we calibrate that because I know when I see something that where someone just sends me the AI summary without their insights, without their observations, without their synthesis, I pretty much disregard it, right?
• 33:36
  SPEAKER_01: I mean, I it it doesn't help me.
• 33:38
  SPEAKER_01: It's like, yeah, I know that.
• 33:39
  SPEAKER_01: I can I can see that from the AI summary.
• 33:42
  SPEAKER_01: You don't have to send me the AI summary, right?
• 33:44
  SPEAKER_01: But but you want to send me your thoughts and interpretations and your insights about this and then what we do with it, right?
• 33:52
  SPEAKER_01: And that's something that I think can be missing when people overrely on on AI.
• 33:57
  SPEAKER_01: And and you can see that they're generated.
• 33:59
  SPEAKER_01: But I agree with your point in that you can take a long rambling email and then say to you know one of the one of the large language models, please structure this or simplify or shorten, right?
• 34:11
  SPEAKER_01: Make this more actionable, change the tone, right?
• 34:13
  SPEAKER_01: All of those things are important.
• 34:15
  SPEAKER_01: It's still your content, it's still your ideas, but you're asking it to make some adjustments and put some polish on it.
• 34:21
  SPEAKER_01: But that is different than you know, sending someone, you know, a big yeah, of course, a very generic candle.
• 34:28
  SPEAKER_03: You you can sense it.
• 34:29
  SPEAKER_03: Like I think there was a study of certain words being used by email that did resonance, it's like exactly the thing is AI can also help you refine with bringing data, right?
• 34:42
  SPEAKER_03: So if you build your different project in Chat GPD, there's that whole history of data.
• 34:46
  SPEAKER_03: So you can reply by saying, hey, last year we were at X conversion rate, and you tell ChatGPD recomplete because I I forgot the conversion rate.
• 34:55
  SPEAKER_03: And at least they will enrich with qualified data with that are very precise and though your email without you having to look back into your analytics for hours.
• 35:06
  SPEAKER_01: As long as it's going back and getting the right data, right?
• 35:08
  SPEAKER_01: Yeah, yeah, that's interesting.
• 35:10
  SPEAKER_01: Uh, but I I think that's uh you know, and it also I think there there's also the danger of developers overrelying on on AI and not using that human judgment.
• 35:20
  SPEAKER_01: But but I think across the board, it's like use AI, but be focused and judicious in how you use it.
• 35:29
  SPEAKER_01: Don't over-rely on it.
• 35:30
  SPEAKER_01: And there have been studies that talk about how over-relying on it will reduce productivity, right?
• 35:35
  SPEAKER_01: And and actually reduce your the cognitive capacity of an individual.
• 35:40
  SPEAKER_01: And you start looking at you know how this is really shaping uh reshaping society, uh, there are a lot of interesting implications.
• 35:46
  SPEAKER_01: But I I do agree with you that you know it has a lot of power to provide those efficiencies.
• 35:52
  SPEAKER_01: Uh, we just need to be able to teach people how to use it most effectively, teach them not to overrely on it, teach them how other parts of the organization may be using this and and how it needs to be uh a tool in their toolbox.
• 36:05
  SPEAKER_03: Yeah, but you you know, a quick tip that I that I started doing is when I start prompting, I first kind of, if I want to reply to a complicated email or something like that, I start just typing the email, not building full phrases and all.
• 36:20
  SPEAKER_03: I don't I don't care about that.
• 36:22
  SPEAKER_03: It's just what exactly is my main ID is and all.
• 36:25
  SPEAKER_03: Because by doing that, you're already giving direction to your uh to your prompt and all, and it's very clear for the AI what kind of answer you're looking for.
• 36:34
  SPEAKER_03: So instead of exchanging on something that is very generic, like do this for me, you start already building everything, a context and all.
• 36:42
  SPEAKER_03: Copy pasting, for example, the whole email thread is always more efficient than just saying reply to the list email because you're giving context.
• 36:51
  SPEAKER_03: But you need to infuse your ID of your type of insight within in order not to be generic from there on.
• 36:58
  SPEAKER_03: That's little things like that that you can add.
• 37:01
  SPEAKER_01: Totally agree, totally agree.
• 37:02
  SPEAKER_01: I uh uh uh inserted an email exchange with someone I was working with, and I got some management consulting level encouragement feedback, right?
• 37:13
  SPEAKER_01: Basically said, hey, this is not going in the right direction.
• 37:16
  SPEAKER_01: I think you need to end this, and here are the reasons.
• 37:18
  SPEAKER_01: And it was very, very thorough.
• 37:19
  SPEAKER_01: But it was from giving that long email interchange with this uh individual that was able to provide that context.
• 37:26
  SPEAKER_01: And then it was able to give me a very, very solid, specific answer to my particular situation.
• 37:34
  SPEAKER_01: Yeah.
• 37:34
  SPEAKER_01: Well, listen, uh uh, you know, any final thoughts for organizations and executives who are embarking on this path?
• 37:41
  SPEAKER_01: Uh, what should they keep in mind?
• 37:43
  SPEAKER_01: You know, I like to say you should understand the inputs, you should understand the outputs, be careful about people saying secret sauce or proprietary, right?
• 37:50
  SPEAKER_01: They should be able to explain what what they're doing.
• 37:52
  SPEAKER_01: What other advice do you have for uh organizations going down this path of either bringing in consultancies or evaluating products?
• 38:01
  SPEAKER_01: What are your what are your headlines uh around that?
• 38:04
  SPEAKER_03: I think the the first thing is we look quickly at your digital ecosystem.
• 38:09
  SPEAKER_03: How is my digital ecosystem today can integrate AI well and perfectly and help me scale at a level of refinement and all super easily by integrating more?
• 38:20
  SPEAKER_03: So I'm comparing with I mentioned different tools, Shopify before, Akinio and all.
• 38:25
  SPEAKER_03: And they start, you see the way they're integrating one to another, they see the way they're integrating AI within their tools, and now you can look at their roadmap and say, okay, I think they're gonna help me be more efficient by having an easier integration.
• 38:40
  SPEAKER_03: Because we talked about vibe coding and all, but vibe coding is producing a code on its own that is very independent for now.
• 38:47
  SPEAKER_03: So you need to have a tool that is enough, flexible enough so that you can integrate via APIs, via different things, and make sure that you automate.
• 38:55
  SPEAKER_03: So that will be the first thing is look at the different solutions that you have.
• 38:58
  SPEAKER_03: The second thing is in terms of change management and all, make sure that every single person in the company knows that they will be affected by AI.
• 39:09
  SPEAKER_03: Okay, so that needs to be clear.
• 39:11
  SPEAKER_03: It's not because your job is this in the warehouse and all, there will be some kind of AI behind it, okay?
• 39:18
  SPEAKER_03: So, in the way you receive information and you treat information and all.
• 39:22
  SPEAKER_03: So it's very important that it's not by bringing an outside consultant, it's by having really the person being helped in deep diving on certain things on his job and all.
• 39:32
  SPEAKER_03: And it's sometimes basically asking the question for him, how do you do that?
• 39:37
  SPEAKER_03: Why does it take so much time?
• 39:38
  SPEAKER_03: What is your biggest pain point?
• 39:40
  SPEAKER_03: And if you're there inside the company to help them fix that pain point, then suddenly it's gonna get better.
• 39:46
  SPEAKER_03: You will get resistance and it's normal, but you need to move forward by building those POCs, as I mentioned, and all.
• 39:53
  SPEAKER_03: And the last part is look at the data that is being used.
• 39:56
  SPEAKER_03: Where is it hosted?
• 39:58
  SPEAKER_03: What kind of tools are you using to host it?
• 40:00
  SPEAKER_03: Is that tool flexible enough?
• 40:01
  SPEAKER_03: How many things, how many times do I see, oh, it's just an SQL database?
• 40:06
  SPEAKER_03: How come we don't build it in Snowflake one and for all so that we can use it or big crates, things like that that needs to be put in place and structuring your data?
• 40:15
  SPEAKER_03: The rest it's day-to-day use, and that's it.
• 40:18
  SPEAKER_01: Yeah, and and I I agree.
• 40:19
  SPEAKER_01: You you mentioned before not just subcontracting something, but building the core competence, right?
• 40:24
  SPEAKER_01: Absolutely.
• 40:25
  SPEAKER_01: You need to understand what you can't outsource your core competence, right?
• 40:28
  SPEAKER_01: You need to say what part of this has to be part of our core competence, and how do we build that capability and how do we build that maturity and evolve the organization?
• 40:37
  SPEAKER_03: And a great example that I read is that there's a big consulting company.
• 40:42
  SPEAKER_03: I'm not gonna mention their name, but they took all their employees for a three uh for a weekend, a three days weekend, just about AI, right?
• 40:49
  SPEAKER_03: And the week after that's it, the company was now offering AI consulting solution because everyone is trained on AI.
• 40:56
  SPEAKER_03: One knows at this point.
• 40:57
  SPEAKER_03: Okay, it's brand new for everyone.
• 41:00
  SPEAKER_03: So don't expect anyone to come from the outside and help you build.
• 41:05
  SPEAKER_03: And remember, the sooner you start building, the the safer you are in your growth because the competition will come from everywhere.
• 41:13
  SPEAKER_03: Before we used to say, oh, it's it seems that it's a Chinese uh product, or it seems that it's not anymore.
• 41:20
  SPEAKER_03: Everyone can learn from you and build assets directly from you because translation is not a problem, visuals generating is not a problem.
• 41:28
  SPEAKER_03: So you can bring your brand at a very mature level very quickly.
• 41:31
  SPEAKER_03: So your competitor will start popping from everywhere.
• 41:34
  SPEAKER_01: And leveraging the right expertise, right?
• 41:36
  SPEAKER_01: If you don't have the expertise, you need to find that, build a trusted relationship, and then use that to accelerate your own internal learning.
• 41:44
  SPEAKER_01: So that's great.
• 41:45
  SPEAKER_01: Well, listen, Rudy, thank you so much uh for sharing your knowledge and your expertise and in the space uh around e-commerce uh and uh B2B and AI adoption.
• 41:56
  SPEAKER_01: Uh, thank you to our listeners for tuning into the early AI podcast.
• 41:59
  SPEAKER_01: You can reach out to to Rudy uh through LinkedIn.
• 42:03
  SPEAKER_01: That's in the show notes, and it's uh Rudy Apidball, R-U-D-Y-A-B-I-T-B-O-L.
• 42:11
  SPEAKER_01: So you can find it on LinkedIn if you'd like to connect.
• 42:13
  SPEAKER_01: Thank you again, Rudy.
• 42:14
  SPEAKER_01: Thank you uh for our listeners, and uh, we'll see you next time on the uh Early AI Podcast.
• 42:21
  SPEAKER_01: Thank you.
• 42:23
  SPEAKER_00: Thank you for joining us on another deep dive into AI innovation.
• 42:27
  SPEAKER_00: Tune in next time when we introduce another industry expert and discuss how to maximize AI in your world.
• 42:33
  SPEAKER_00: The Early AI Podcast is sponsored by Early Information Science and Vector.
• 42:38
  SPEAKER_00: To learn more, visit early.com.
• 42:41
  SPEAKER_00: That's E A R L E Y dot com.
• 42:44
  SPEAKER_00: Thank you for tuning in.