Latest / The Adults In The AI Room / Earley AI Podcast Episode 64: Understanding AI: From Misconceptions to Effective Product Mindset

Transcript

Earley AI Podcast Episode 64: Understanding AI: From Misconceptions to Effective Product Mindset · The Adults In The AI Room

• 0:00
  SPEAKER_02: Well, good morning, good afternoon, good evening.
• 0:02
  SPEAKER_02: Welcome to the Early AI podcast.
• 0:04
  SPEAKER_02: My name is Seth Rulley.
• 0:05
  SPEAKER_01: And I'm Chris Featherstone.
• 0:07
  SPEAKER_02: And today we're really excited to introduce our guest to discuss a lot of the issues in the marketplace around AI, misconceptions, how people need to think of a product.
• 0:18
  SPEAKER_02: Technical teams need to have a product mindset.
• 0:21
  SPEAKER_02: And we'll talk about how that's different than just looking at applying the technology as a product versus thinking in product mindset.
• 0:29
  SPEAKER_02: Those are different things.
• 0:31
  SPEAKER_02: And we'll talk about data storytelling.
• 0:33
  SPEAKER_02: So our guest today brings over 27 years of experience from corporate roles in tech and pharma spaces.
• 0:40
  SPEAKER_02: He's now a respected advisor and speaker.
• 0:42
  SPEAKER_02: He focuses on improving data storytelling and fostering a product mindset among technical employees.
• 0:48
  SPEAKER_02: He's lived in Poland, US, and is currently based in Munich, Germany.
• 0:52
  SPEAKER_02: Jack Blenka, welcome to the show.
• 0:55
  SPEAKER_02: Thank you very much for having me here.
• 0:57
  SPEAKER_02: I'm glad to be part of the show.
• 1:00
  SPEAKER_02: That's terrific.
• 1:01
  SPEAKER_02: So I wanted to start off.
• 1:02
  SPEAKER_02: You know, we like to understand what you're seeing in the marketplace in terms of misconceptions, in terms of lack of understanding of fallacies.
• 1:11
  SPEAKER_02: So what are you seeing now?
• 1:12
  SPEAKER_02: That's been changing quite a bit, but um what is your experience showing you right now that people are still not getting or that are not getting now?
• 1:21
  SPEAKER_02: Maybe they maybe it's changed some in the last uh six or twelve months.
• 1:24
  SPEAKER_04: Well, I'm gonna take a 15-year step back because 15 years ago I had this big epiphany that developing AI solutions and basically assuming they will be used, it won't work.
• 1:38
  SPEAKER_04: And we developed with my team a promotion response modeling for a tech company.
• 1:45
  SPEAKER_04: And we went to the marketing director with this great ideas, we could save 20% of expenses.
• 1:52
  SPEAKER_04: But the marketing director said, so what?
• 1:56
  SPEAKER_04: What we didn't do was explain the benefits.
• 1:59
  SPEAKER_04: We came with this technical specs, how it works, and then what the code is and why this model derived these recommendations, but we didn't really think about what's in it for me from the user perspective, from the marketing director perspective.
• 2:16
  SPEAKER_02: Wasn't a powerful enough uh example, or was it just not believable?
• 2:23
  SPEAKER_04: Probably not believable because the response was, well, we've been doing allocation of marketing budgets for years, my team is great.
• 2:32
  SPEAKER_04: Why would this model, this black box, tell us that we could save 20% all of a sudden?
• 2:38
  SPEAKER_02: Okay.
• 2:39
  SPEAKER_02: So it was a lack of of really believing that something like that could work the way it is.
• 2:46
  SPEAKER_02: And when you were explaining the technical architecture and the details and how cool the coding was and why the model did this or that, that wasn't that wasn't doing it.
• 2:56
  SPEAKER_04: That wasn't really that wasn't enough because we focus on really what we focused on on this solution, but not on the customer benefits when discussing this, when presenting this to this marketing director and his team.
• 3:10
  SPEAKER_04: So that was the challenge that led me to from then on to really think about how you sell AI solutions internally within the company, how you as a data team, and this could be even simple dashboards.
• 3:27
  SPEAKER_04: Why should I use why should SiceRap use a new dashboard?
• 3:32
  SPEAKER_04: We've been doing making decisions without these new tools, without the new dashboard, or without new new AI solutions, and that has not changed what actually from my point of view it got even worse because now with Gen AI, there is so much hype around the technology.
• 3:55
  SPEAKER_04: Every day there is a new LLM coming out, there's new features added, and people focusing on this technology, people focusing on the speed or on the accuracy, instead of thinking about what it's the business problem that this AI solution should address.
• 4:16
  SPEAKER_04: Because if you start from the business problem, then you start thinking how also you communicate to the business users about the benefits of that.
• 4:26
  SPEAKER_02: And so, and you're you as you observe, we're increasingly seeing this with Gen AI because people have a misconception about not only how it works, but what it's good for, right?
• 4:37
  SPEAKER_02: And right, and uh you know, in and just unrealistic expectations about what it can do and and and not clearly understanding the preconditions that need to be there.
• 4:49
  SPEAKER_02: So, Chris, did you want to add to that?
• 4:51
  SPEAKER_01: Yeah, well no, I I think that you were spot on.
• 4:53
  SPEAKER_01: I was gonna say, in terms of you know, focusing on the business problem is is the most important thing.
• 5:00
  SPEAKER_01: And at the same time, they don't folks just don't know what it can do, like Seth, like you're teasing out.
• 5:05
  SPEAKER_01: They don't know the limits of it, they don't know the um where to stop, where to start, what it can do, how to actually apply it, what it's gonna actually replace, and or can it augment?
• 5:17
  SPEAKER_01: You know, there's so much, like you said, just in terms of the first question that we usually ask folks is misconception misconceptions that you're seeing.
• 5:25
  SPEAKER_01: So, how do you then go help them figure out what that use case should be and what the limits or the governing, you know, the governances that need to be in place should be.
• 5:34
  SPEAKER_02: And and just to add to that, I mean, a focusing on the business problem and the use case is not necessarily new, right?
• 5:41
  SPEAKER_02: But it requires a new emphasis with Gen AI because I think people do have a misconception about what it can do, right?
• 5:48
  SPEAKER_02: They they have this preconceived notion of what Gen AI is, yet they're not necessarily saying, well, how exactly will it apply to my business?
• 5:57
  SPEAKER_02: So you want to talk about how it this is different than prior technologies?
• 6:02
  SPEAKER_02: Because again, we're always saying we should focus on the business problem and the outcome and the use case.
• 6:07
  SPEAKER_02: But why is it different now with with AI and Gen AI?
• 6:11
  SPEAKER_02: Or how is it?
• 6:12
  SPEAKER_04: So, from my point of view, it's uh it's different because there is more hype and more kind of formal fear of missing out.
• 6:22
  SPEAKER_04: The CEOs fear that they miss out if they don't use AI, whatever, whatever AI is, like in the 90s, where the CEO saying, I heard about this website, build me one.
• 6:35
  SPEAKER_04: I don't know what what I need it for, but I need a website.
• 6:39
  SPEAKER_04: Today it's the same with AI, and and there's misconception what AI is, and people now equate a lot AI with generative AI.
• 6:52
  SPEAKER_04: People forget the analytical AI that has been around for 20 plus years, but now with this AI hype with the FOMO, CEO saying, Well, we need AI, and the most of the time they they think of chat GPT, and the challenge comes in when IT is then tasks to implement that because IT comes usually from the technology point of view, not from the business needs, but from technology, and the challenge becomes how do you take that technology and make it useful?
• 7:30
  SPEAKER_04: IT is tasked with implementing gen AI, but business is not is like kind of almost left alone of what should we use it for.
• 7:40
  SPEAKER_04: So if there is not enough resources on that business side who, even the CEO driving the use of Gen AI or or individual business units, the technology tools will be there, but nobody will be using them.
• 7:55
  SPEAKER_02: Right, right.
• 7:56
  SPEAKER_02: I was just talking to an organization that is doing some investigation in this area.
• 8:02
  SPEAKER_02: And when I showed them an approach, which is really an assessment of looking at the business problems in the organization and looking across multiple systems and applications and really doing this maturity evaluation, understandable sources, understanding where their friction points were, information leverage points.
• 8:20
  SPEAKER_02: They said, Wow, that's much more than we expected.
• 8:22
  SPEAKER_02: And I can see that because people aren't they they have a preconception of what this stuff can do, and they're not necessarily looking across.
• 8:31
  SPEAKER_02: They said, Well, this is a very holistic approach.
• 8:33
  SPEAKER_02: We like it, but we're afraid it's too big, you know.
• 8:35
  SPEAKER_02: And it's like, well, it's not too big because you have to look across the business and say, where can we apply this?
• 8:41
  SPEAKER_02: Where do we have the problems?
• 8:42
  SPEAKER_02: So, you know, what do you say to an organization that uh that may think that looking broadly is is too big, you know, what's the alternative?
• 8:50
  SPEAKER_02: I mean, obviously you have to prioritize, but you have to start off by looking and casting a wide net, looking across the organization, unless they're very, very clear about a business problem.
• 9:01
  SPEAKER_04: So that from my point of view, it it's always starting with business needs will get you to the right first step.
• 9:12
  SPEAKER_04: And different organizations are different levels of maturity when it comes to data and AI.
• 9:18
  SPEAKER_04: So almost every organization has this low-hanging fruits, fruits in a sense that activities or processes that could be improved to reduce expenses or increase revenues.
• 9:33
  SPEAKER_04: So um it has to be looked at this kind of horizontal view in terms of identifying somebody who has the overview of the organization, understands the business needs or the biggest challenges, identifying and at the same time, of course, knowing what AI can do to address those challenges.
• 9:54
  SPEAKER_04: Somebody or a team like that identifying what are the top three low-hanging fruits that could be addressed with AI, whether it's gen AI, whether it's analytical AI, once you identify that, then you of course need to identify what is all the elements needed to be successful with that implementation.
• 10:16
  SPEAKER_04: What is already in the organization, what needs to be acquired, whether it's additional data, with whether it's technology, although most of the time companies have enough sufficient technology in place.
• 10:29
  SPEAKER_03: Yeah.
• 10:29
  SPEAKER_04: So starting with this identifying the low-hanging fruits usually leads to fruitful results when it comes to the first EI AI implementations.
• 10:42
  SPEAKER_04: And after that, it kind of you go to the next step.
• 10:45
  SPEAKER_04: What are the next with higher hanging fruits and so on?
• 10:48
  SPEAKER_02: And when you go into organizations, what are they what are the uh uh what are the misconceptions around the capabilities of the tools themselves?
• 10:59
  SPEAKER_02: Like when you look at Gen AI, what are people not getting uh when it comes to how to use Gen AI?
• 11:06
  SPEAKER_04: So they they think it's they when a topic of AI comes up, they think for when the topic of AI is is comes up, they always think of technology.
• 11:20
  SPEAKER_04: So they think about and they know when it comes to technology, most people, most executives too know obviously the the chat GPTs of the world.
• 11:31
  SPEAKER_04: So they think, okay, we need that tool to be implemented here, and then they don't understand how these tools were trained on what data they were trained on.
• 11:46
  SPEAKER_04: Many larger companies have so much data, so much content that it's difficult for employees to even comprehend and use five percent of that.
• 11:58
  SPEAKER_04: I mean Gen AI is a perfect tool to implement a large organization that is then trained on the internal data, so employees have uh easy access to that information.
• 12:12
  SPEAKER_04: But the misconception is we're gonna implement Chat GPT in-house, and we'll all of a sudden have the same capability like applying to our internal data, although that's probably the most critical part now, the next step to address how you train or retrain or adjust the model to take into account your internal data.
• 12:38
  SPEAKER_02: So, how do you look at that process?
• 12:41
  SPEAKER_02: Because people have different ideas and concepts of what training is and what tuning is.
• 12:46
  SPEAKER_02: And so, you know, what in our experience, we've come back, you know, so we mostly are proponents of retrieval augmented generation, and we also call it IA-directed information architecture retrieval augmented generation.
• 13:00
  SPEAKER_02: So, in other words, you're using the information architecture, the data structures, the data, you know, the data architecture, and structuring that content and that information in a way that can be retrieved, and also breaking up large monolithic documents.
• 13:14
  SPEAKER_02: You don't want a 300-page document, you want an answer within that document.
• 13:17
  SPEAKER_02: And so that means, you know, building a component model that means ingesting into a vector space, it means enriching those embeddings and so on.
• 13:25
  SPEAKER_02: Do you think people have an understanding of those steps to the process and and what goes into because chat because any large language model can be trained on lots of stuff, but it's not trained on your data, right?
• 13:37
  SPEAKER_02: It doesn't know your organization, your processes, your IP.
• 13:41
  SPEAKER_02: So where are people with that level of understanding, would you say?
• 13:47
  SPEAKER_04: I think it it depends on the organization.
• 13:50
  SPEAKER_04: So I've seen organizations that actually the last company I was with, where there was a large technical team responsible for exactly what you described.
• 14:02
  SPEAKER_04: Uh although it was going slowly, the understanding was there.
• 14:06
  SPEAKER_04: I heard or see from other companies that don't have that understanding.
• 14:11
  SPEAKER_04: So they think uh just applying a large language model at least gives the thing of this gives the first step or the first step for employees to use large language models in-house.
• 14:25
  SPEAKER_04: So maybe they they put it within the company infrastructure so the data doesn't go out, but still it's that that LLM trained on the internet data, not on our company data.
• 14:39
  SPEAKER_04: So far, what I've seen is a mixed uh mixed uh response.
• 14:44
  unknown: Right.
• 14:44
  SPEAKER_01: What do you what do you think then, Jack, when we're getting into this notion too now of large action models and large genetic models, too?
• 14:52
  SPEAKER_01: Right?
• 14:52
  SPEAKER_01: Because we've got LMs, that's just one piece of it now.
• 14:55
  SPEAKER_01: And to Seth's point, now we've got all of this really neat, you know, inter intro interesting infrastructure.
• 15:00
  SPEAKER_01: And to pivot on unstructured and structured data, the only way to do it was with a rag architecture, because then that gives you the accuracy you need.
• 15:08
  SPEAKER_01: But now we have agents that are going off and autonomous doing work, and we've got actions that are doing work.
• 15:15
  SPEAKER_01: They're all calling different models and things.
• 15:17
  SPEAKER_01: I mean, and to your point earlier, like the rapid notion of how fast these technologies are coming on board is such is in such a feverish way that it's hard for these folks to stay on top of things.
• 15:31
  SPEAKER_01: And you know, to your point also, LLMs like pattern recognition has been here for we'll say eight years, but LLMs were the first thing that was consumer-based.
• 15:39
  SPEAKER_01: So, yeah, of course, everybody has the perception that that's AI and it's not, it's one of many, right?
• 15:45
  SPEAKER_01: Thousands, but now we have agents.
• 15:47
  SPEAKER_01: What's your perspective on that?
• 15:48
  SPEAKER_01: On those aspects, too, where we have now agents that are going out and doing things autonomously and pulling back relevant information and context and neat and interesting things.
• 15:58
  SPEAKER_04: So, because as you said, this with agents, it gets the requirement for understanding how how the system works, they become higher.
• 16:08
  SPEAKER_04: So if companies don't understand LLMs, they just won't be able to use agentic AI.
• 16:18
  SPEAKER_02: Yeah, and I think people are focusing on the wrong thing when they say we need agentic.
• 16:24
  SPEAKER_02: I talked to a company recently, they said, Well, we really want to build agents.
• 16:27
  SPEAKER_02: It's like you really want to build agents, why?
• 16:29
  SPEAKER_02: Right?
• 16:30
  SPEAKER_02: Because agents are built into many of our processes.
• 16:33
  SPEAKER_02: So we have agentic technology that's just part of our our stack, you know, when we are doing retrieval augmented generation, it's these agents are going off and doing things by by integrating with other systems, by pulling other APIs.
• 16:48
  SPEAKER_02: And so that's already part of it, but it's not the focus of it.
• 16:52
  SPEAKER_02: I think that's too much focused on the technology, not on the problem.
• 16:56
  SPEAKER_02: So when you go into organizations and you talk about uh a product mindset for technical teams, so you want to be careful about people implementing a product to solve a problem without understanding the nature of the problem, right?
• 17:11
  SPEAKER_02: That's where you have the cart before the horse.
• 17:13
  SPEAKER_02: But tell me what your what your concept is for getting that mindset in the part of technical teams, the product mindset.
• 17:21
  SPEAKER_02: What does that mean?
• 17:22
  SPEAKER_04: Um before I go to that, so one more word on the agentic AI.
• 17:27
  SPEAKER_04: From my point of view, that that's kind of kind of this gen AI hype is now being replaced with the agencai.
• 17:36
  SPEAKER_02: Yep.
• 17:36
  SPEAKER_04: Where where people basically creating agents and call them agents, but it's uh basically like the old APIs, software models, systems, talking with another software, talking with other software.
• 17:49
  SPEAKER_02: It's a oriented architecture, right?
• 17:54
  SPEAKER_04: Coming back now to the product like mindset.
• 17:56
  SPEAKER_04: The the product mindset for me is uh technical teams, data teams, data scientists, understanding not only what they develop, I mean understanding business needs, understanding the solution, and not only developing that solution, but also being responsible for making sure that the solution is being used.
• 18:19
  SPEAKER_04: So instead of developing a solution solutions and throwing over the fence and hoping that it's used by business users, product mindset involves having a clear marketing plan.
• 18:34
  SPEAKER_04: How do you market this AI solution to the non-technical users?
• 18:40
  SPEAKER_04: So in my pride company, we implemented a Next Best Action system for Sales Reps.
• 18:46
  SPEAKER_04: So, in addition to developing it, we also develop a marketing plan with a big lounge of this, with implementing many other components to ensure that the business users understand the value, understand how it improves their daily business, and understanding how they can reduce the time they spent on some activities by using this and next best action.
• 19:17
  SPEAKER_04: So product marketing is basically product mindset includes product marketing.
• 19:21
  SPEAKER_04: And then here you can apply typical the five piece of product marketing concepts, so product price, placement, promotion, and people to apply the concept to market AI products as well.
• 19:35
  SPEAKER_02: And is that on the business?
• 19:37
  SPEAKER_02: Is that more on the business who understands their needs and their people, or is it incumbent on the technical team?
• 19:45
  SPEAKER_02: My guess is the answer is both, but I wanted you to expand on that.
• 19:49
  SPEAKER_02: What's the role of the business versus the technical team?
• 19:52
  SPEAKER_04: So ideally, you would have some roles that are kind of the intersection of both.
• 20:01
  SPEAKER_04: And these roles can can be part of the business, these roles can be part of the technical teams.
• 20:07
  SPEAKER_04: Where they report to is is actually less relevant, but it's important that they are the kind of the intersection between the technical and the business side.
• 20:16
  SPEAKER_04: And this is and they will bring the requirements in terms of the business needs, but they also the people who will help to sell that to the business users.
• 20:29
  SPEAKER_04: So these roles are required.
• 20:31
  SPEAKER_04: I call them data translators, but could be you can call them whatever they want as long as they they address that kind of the missing link between these two different functions.
• 20:44
  SPEAKER_02: So that's kind of a rare combination to find in a technical team, right?
• 20:48
  SPEAKER_02: Usually you'll need a business analyst or a liaison, but again, I can understand how you're trying to instill this mindset on uh technical team to understand that look, you are building a product for a user that's solving a problem.
• 21:01
  SPEAKER_02: So, how are you going to get that message out to those users and how are you gonna help them understand the value to them and why they should they should change their work process and why should they should you know trust your system?
• 21:14
  SPEAKER_02: So I think trust is probably a big piece of that.
• 21:16
  SPEAKER_02: What else is important about the product mindset?
• 21:19
  SPEAKER_04: What what we had good luck with during the development of this Next Bus action, we brought in two SAS reps into the development team.
• 21:30
  SPEAKER_04: So they were part of a weekly calls with the data engineers, data scientists, ML engineers, and they basically helped them, those technical people help them understand how sales reps operate and help them understand how each of these functions that has been requested, how that function helps sales reps to achieve their tasks.
• 21:57
  SPEAKER_02: How are they operating?
• 21:58
  SPEAKER_02: They're coin operator.
• 22:01
  SPEAKER_01: Yeah, exactly.
• 22:02
  SPEAKER_01: You would think you would think though, too, Jack, that because I've done a number of these systems in the past, that one of the greatest pieces of data and information is going back through the historical records of the transcripts of the actual reps themselves.
• 22:16
  SPEAKER_01: Because the problem with what you're talking about too, though, it's a double-edged sword, is the reps and the way they behave are incentivized, and they behave the way they're incentivized, I should say it that way.
• 22:27
  SPEAKER_01: So when we get into these types of situations and scenarios, what we often see too is what they tell the developers in terms of the user stories and what they need to be able to develop are different from actually how they really act.
• 22:40
  SPEAKER_01: And the problem with that is are we solving symptoms or are we solving problems?
• 22:43
  SPEAKER_01: And we want to be able to solve problems.
• 22:45
  SPEAKER_01: And so the we we saw so many times the discrepancy between when we would actually do, you know, dive into the pattern recognition of the processing of their actual transcripts and their outcomes.
• 22:58
  SPEAKER_01: And then those outcomes would result in multiple callbacks.
• 23:01
  SPEAKER_01: And those multiple callbacks were exponentially more expensive, as you can imagine, because they were never solving the problem, they're solving the symptom.
• 23:08
  SPEAKER_01: And then what we also saw was then they would actually get into this notion of next best action was automatically give them a discount.
• 23:16
  SPEAKER_01: Right?
• 23:16
  SPEAKER_01: And that was then just in fact just haemorrhaging cash because, again, solving the symptoms and solving the problem.
• 23:23
  SPEAKER_01: So part of that I'd love to get your take too, in terms of solving what's when we start thinking about AI to replace humans, we also need to take into consideration the behaviors and how we've incentivized humans to actually do jobs and outcomes and output.
• 23:37
  SPEAKER_01: And how does that change when we think about this in terms of removing redundant work with AI?
• 23:42
  SPEAKER_01: Does that make sense?
• 23:44
  SPEAKER_04: Yeah, but so I I would not see I don't see next best action as replacing size reps, but helping them make better decisions.
• 24:00
  SPEAKER_04: Okay, so I'm a little biased being in Germany with all the regulations and all the workers' council prevent many things from being implemented.
• 24:12
  SPEAKER_04: So so in that situation, because you mentioned the symptoms versus the causes, the I would say the feedback we got from the sales reps were really addressing the the lowest-hanging fruits in terms of how they operate, like to optimize the the the order of the activities or to better understand what preference does this customer has have for specific content or specific channel preference.
• 24:42
  SPEAKER_04: So so so it was was not as as as uh tricky as as you described.
• 24:49
  SPEAKER_01: Oh, I did this with Deutsch Telecom too.
• 24:51
  SPEAKER_01: I did it with you know British Telecom, did it with uh Vodophone, did it with a number of organizations across the world.
• 24:58
  SPEAKER_01: And part of that, what we were trying to do is come up with when they were coming in, were they fraudulent or not, and doing classification.
• 25:06
  SPEAKER_01: I think that's where a lot of these organizations miss.
• 25:08
  SPEAKER_01: And this is one of the things when maybe I want to go back to is that the folks that are trying to understand these systems now, would you agree or disagree that they need to be more technical or understand?
• 25:21
  SPEAKER_01: It's like moving, like you said, from business to when we were doing on-prem and and uh brick and mortar to the internet, it up-leveled the entire conversation in terms of what that meant.
• 25:32
  SPEAKER_01: Now people were talking about IP-related things and talking about um, you know, secure sockets and things like that from a business perspective, because it was just nascently part of the business now.
• 25:43
  SPEAKER_01: And I feel like we also have to uplevel that whole mindset around the conversation of being more technical around the ins and outs of AI.
• 25:52
  SPEAKER_01: I don't know if you agree or disagree with that, but that's where we're seeing things going.
• 25:56
  SPEAKER_04: So it depends on how you define technical parallel to different AI solutions.
• 26:02
  SPEAKER_04: One of the first things I started doing when I joined that company was to develop the implement a data literacy program.
• 26:11
  SPEAKER_04: And data literacy, and later expanded with AI literacy, was meant to show how data, what data is available, what that data represents, how you can access data, but also how you can do a quick analysis.
• 26:30
  SPEAKER_04: And then the third component of that was how do you, you sales rep, you marketing the person, how you generate data.
• 26:40
  SPEAKER_04: So you you Chris, you kind of alluded to this CRM systems, how sales reps document their activities.
• 26:47
  SPEAKER_04: We spend lots of time explaining how clear documentation helps to really understand the customer.
• 26:56
  SPEAKER_02: And favorite activity.
• 27:07
  SPEAKER_03: Yeah.
• 27:08
  SPEAKER_04: I know.
• 27:09
  SPEAKER_04: So so that's why we we actually that was a big component of this data literacy program.
• 27:15
  SPEAKER_04: And if you think about elevating the technical capabilities, that's what I would say the data literacy program is for to improve the data mindset of non-technical people for them to understand uh how how data and how AI plays a role in business decisions.
• 27:39
  SPEAKER_04: I don't think they need to go deep in a technical um know-how that you would reserve for the engineers or IT, but they need to understand the the data mindset at a level that's sufficient to use for them to use data and AI.
• 27:58
  SPEAKER_02: Yeah, it's like how does it apply to my job and how does it apply to my success?
• 28:02
  SPEAKER_02: And the secure sockets layer may not be as relevant unless they're selling that, right?
• 28:08
  SPEAKER_02: Um, I want to switch uh gears just a little bit and talk about, you know, and you started to allude to this, the storytelling piece.
• 28:16
  SPEAKER_02: So, you know, telling the story can be challenging, you know, for teams and for business people.
• 28:22
  SPEAKER_02: What has proven to be effective in helping um both technical people tell that data storytelling, or even uh business people to understand and tell that story?
• 28:33
  SPEAKER_02: So talk a little bit about that, if you would.
• 28:35
  SPEAKER_04: So uh that's part of this what I call product marketing, how you market your solutions to the non-technical users, and data storytelling helps to turn these specs into a a story about how a user like you I'm gonna use here in an express action system again for size reps.
• 29:02
  SPEAKER_04: Size reps like you, how she had trouble being always or often rejected by customers when when she went to visit, and how a data story that explains what she did can help then the other sales trips understand the benefits of next plus action.
• 29:27
  SPEAKER_04: So a data story can be built around different um characters, different stories, almost like a Hollywood movie that can be different, have different plots, and leveraging that concept, you can create a story where using an example, so Setraps being where customers cancel on her when she starts realizing how this next best action can help because she can identify five other customers nearby she can visit, she can identify which content the customer wants she is going to visit tomorrow.
• 30:08
  SPEAKER_04: She can identify through which channel to engage with that, and that helps to improve the engagement with the customer and reduce the frustration that the sales rep has.
• 30:19
  SPEAKER_04: So that's in a much nicer way.
• 30:21
  SPEAKER_04: That's that is an example of how you can package a data story that you can tell then to other sales reps so they can identify themselves with that character in that story and better understand the value what AI solution provides.
• 30:40
  SPEAKER_02: I'm sorry, keep going.
• 30:42
  SPEAKER_04: And basically, what we did, we hired um a data storytelling expert who came in for a three-day session, and with a bigger group, basically, everybody who had time for that from the technical side, from the data people, came in and learned about the process, learned about the different plots you can use.
• 31:05
  SPEAKER_04: Basically, we created several stories for several different data and AI solutions we had at that time.
• 31:12
  SPEAKER_02: Interesting.
• 31:13
  SPEAKER_02: And so that was done through working sessions.
• 31:15
  SPEAKER_02: So it's it's really the day in the life, right?
• 31:18
  SPEAKER_02: It's a really talking about how people will interact and what they're going to do.
• 31:21
  SPEAKER_02: So those are user stories, those are use cases, those are user journeys, and things like that, right?
• 31:27
  SPEAKER_02: So you can also put them in those contexts.
• 31:30
  SPEAKER_04: Right.
• 31:30
  SPEAKER_04: And it's it's usually when you put the the main character being that person using it, that helps for the other colleagues to identify with that person.
• 31:41
  SPEAKER_01: Yep.
• 31:42
  SPEAKER_01: Jack, did you guys classify the customers coming in as well?
• 31:46
  SPEAKER_01: In terms of setting up and and putting some pre-context around if they're a high-value customer, if they're a fraud customer, that kind of stuff, right?
• 31:55
  SPEAKER_01: Because it's super important as they're coming in to be able to do that.
• 32:02
  SPEAKER_02: Yeah, exactly.
• 32:04
  SPEAKER_04: That was actually the very first thing I did when I joined the company because there was no customer segmentation in place.
• 32:09
  SPEAKER_04: So that was uh very first thing my team has done, uh identifying customer profiles and then then identifying segments, breaking the those into different, as you said, high value, low value, or or more receptive, less receptive to new products and so on.
• 32:27
  SPEAKER_02: Jack, what else do you cover in your your keynotes and your speaking?
• 32:31
  SPEAKER_02: What are the topics um are of interest to folks uh these days in the marketplace?
• 32:36
  SPEAKER_02: What are you seeing or asking for?
• 32:38
  SPEAKER_04: So I was surprised how many are interested in really understanding what AI is.
• 32:43
  SPEAKER_04: So I have a keynote between Chat GPT and Terminator.
• 32:51
  SPEAKER_04: So that's actually the very interesting to the audience because many people have different perceptions, people at different levels, C level, engineers, marketing, they have different perceptions of what AI is based on their experience.
• 33:08
  SPEAKER_04: Many things, many times they think it's Chat GPT.
• 33:12
  SPEAKER_04: So so in that keynote, I basically explain machine learning a little bit, explain deep learning, explain the small subset of that being generative AI, and and show also that it's not only two years old, that has been used in business for 20 plus years, so people get the sense, oh okay, that's uh there is more to that than just chat GPT.
• 33:41
  SPEAKER_02: Tell me what else uh do you do uh for fun?
• 33:43
  SPEAKER_02: What do you do outside of work?
• 33:46
  SPEAKER_02: I travel, I like to travel.
• 33:49
  SPEAKER_02: My spare time, I wonder why I have no spare time.
• 33:51
  SPEAKER_02: What else do you do?
• 33:54
  SPEAKER_04: Yeah, I like to travel.
• 33:56
  SPEAKER_04: I like the um now it's been the winter, not that great for biking.
• 34:01
  SPEAKER_04: I like to bike to keep myself my fitness level up.
• 34:05
  SPEAKER_01: Nice.
• 34:06
  SPEAKER_01: What city in Germany did you say you were in?
• 34:08
  SPEAKER_04: Munich.
• 34:09
  SPEAKER_01: Oh, you're in Munich, okay.
• 34:11
  SPEAKER_01: Awesome.
• 34:12
  SPEAKER_01: It's a great city.
• 34:14
  SPEAKER_04: So yeah, it's a it's a great city, and uh I mean the many things to do nearby.
• 34:18
  SPEAKER_04: There are lakes, there are there are mountains.
• 34:21
  SPEAKER_04: I like to hike as well.
• 34:23
  SPEAKER_04: Um also in Switzerland.
• 34:29
  SPEAKER_02: This has been great.
• 34:30
  SPEAKER_02: Thank you for sharing your your insights and and your knowledge on today's early AI podcast.
• 34:37
  SPEAKER_02: I want to thank uh thanks for everyone tuning in.
• 34:40
  SPEAKER_02: But uh Jack, thank you so much for being here today.
• 34:42
  SPEAKER_02: I really appreciate it.
• 34:44
  SPEAKER_04: Thank you very much for having me here.
• 34:46
  SPEAKER_04: It was a great uh discussion.
• 34:48
  SPEAKER_02: Yeah, and and I want to thank uh the folks for tuning in.
• 34:51
  SPEAKER_02: We'll catch you on the next episode of the Early AI Podcast.
• 34:55
  SPEAKER_02: We'll really talk about AI across industries.
• 34:57
  SPEAKER_02: And by the way, where can people find you?
• 34:59
  SPEAKER_02: You are at uh you have Jack Lampka, and that's J-A-C-K-L-A-M-P-K-A.com.
• 35:06
  SPEAKER_04: And then Jack Lampka.com is where you most people will find me.
• 35:11
  SPEAKER_04: Okay, and then I'll link it.
• 35:13
  SPEAKER_04: Thank you so much for that.
• 35:14
  SPEAKER_04: Maybe one more thing.
• 35:15
  SPEAKER_04: So uh aside from keynotes, uh so after leaving the corporate world, I'm now focusing on keynotes, but also provide advisory services and and executive sparing for executives who just want to see talk about AI, better understand AI, and and how this can help them to profit.
• 35:36
  SPEAKER_02: Absolutely.
• 35:36
  SPEAKER_02: Well, that's great.
• 35:37
  SPEAKER_02: So um folks know where to uh find you.
• 35:40
  SPEAKER_02: Again, uh thanks for being here.
• 35:43
  SPEAKER_02: Really appreciate it.
• 35:44
  SPEAKER_02: And uh thank you to uh Carolyn and thank you, Chris, for uh the other pieces.
• 35:49
  SPEAKER_02: So great to see you, and we'll we'll talk again soon.
• 35:53
  SPEAKER_00: Visit early.com to find links to the full podcast on all audio platforms to listen on the go.
• 35:59
  SPEAKER_00: Thank you for watching.
• 36:01
  SPEAKER_00: Contact us today at infoearly.com.