AI is everywhere in manufacturing right now, but most companies are still stuck in pilots, hype cycles, and half-finished experiments. In episode 83 of the Missing Half podcast, Bill sits down with Sebastian Chedal, Founder & CEO of Fountain City and Co-Founder of TestFox AI, to break down what it actually takes to move beyond proof-of-concept and into repeatable, scalable AI outcomes inside mid-market and enterprise manufacturing companies .

Sebastian has spent decades leading digital transformation initiatives and now works directly with manufacturing leaders implementing AI in real operational environments. Together, he and Bill unpack why so many AI initiatives stall after early wins, where executive expectations disconnect from technical reality, and how poor data hygiene and undefined processes quietly sabotage promising projects. They explore the importance of humility in AI adoption, why smaller, focused implementations outperform grand transformation mandates, and how leadership maturity ultimately determines whether AI becomes a competitive advantage or an expensive experiment.

The conversation also dives into one of the most pressing issues in manufacturing today: capturing tribal knowledge before it walks out the door. With a generational workforce shift underway, AI presents a powerful opportunity to preserve institutional expertise, democratize it across teams, and accelerate onboarding for the next generation of engineers and operators. Bill and Sebastian also discuss how private equity and investors are evaluating “AI-enabled” companies, how to distinguish real operational leverage from performative AI branding, and why the asymmetrical upside of AI makes experimentation not just worthwhile, but necessary.

Bill: Thank you for joining the Missing Half podcast where we're discovering what's missing. Today it's AI. Today, manufacturing leaders hear a lot about AI, but very little about what actually works. Today's focus is about moving beyond pilots and hype to repeatable, scalable outcomes. And I'm joined by a very special guest today, Sebastian Chedal, who is the founder and CEO of Fountain City, co-founder of TestFox AI, and he works primarily in mid-market and enterprise manufacturing companies on AI strategy and implementation. Sebastian, welcome to Missing Half.

Sebastian: Thank you so much for having me, it's a pleasure.

Bill: Sebastian, before we get tactical, let's just do a little quick overview of your background and how did you end up where you are today?

Sebastian: Thank you. Yeah. So I've had a pinball career, so to speak. I started pretty early in tech and IT back in the mid nineties. And then in 98 started Fountain City. And the first few years was doing a lot in different areas. I'm, I love learning new things all the time and both creative and technical. I kind of have both sides to me and, I mean, I've had two nonprofits, two tech companies, which are Fountain City and Test Fox, and also one game company in my past. So I'm entrepreneurial and yeah, things have just, I would say just evolved over time, different opportunities. Sometimes pivots are needed because of how things are evolving with tech. I think very early on I realized in tech you're paid to learn and if you stop learning or stop evolving, you're just out. So that's kind of where we are today and things have just kind of led this path. I mean, I could go deeper into origin stories and things like that, but that's the quick one.

Bill: Interesting that you talk about creativity and technology and IT because isn't it, aren't we at a moment right now with AI that it is, there's certainly a component that is very technical and very detail oriented, but there's also room for creativity because we are still defining how AI is going to operate, how it is going to hopefully help us achieve these outcomes. Do you see that as being true?

Sebastian: Yeah, absolutely. I mean, I view everything ultimately from the creative lens or perspective. I would say I'm at a core kind of like a creative teacher hybrid kind of thing. That's kind of like my calling. Some people it's like, there's, I remember reading once there are three main callings you can have and a mix of them. It's either creative, teaching, or healing. And so kind of more of primarily creative secondary teaching. And so creativity comes through in running a business, creating solutions for people to solve problems. We're doing primarily digital transformation for the last while and primarily manufacturing. And that is a very creative endeavor because you're, you're trying to help people change the way that they're working, bring in new tools and technologies, methodologies. And then the whole process of roadmap planning and then with AI figuring out what is our North star, what are we looking to achieve? And then how do we break that down into actionable steps? It's a very creative process to be doing all of that brainstorming and then identifying and then, and then how to use AI effectively right now is also pretty creative. You have to really understand what it's good at and what it's not. So that's, yeah. So it just comes through in everything that I'm doing. So yeah.

Bill: That's great. Sebastian, whenever. So you're a serial entrepreneur, right? A serial founder. And what I've found the most successful serial founders are those that focus on solving a specific problem for a specific customer group. And certainly you've you've done that recently with Fountain City by focusing on helping manufacturers with AI. What were the problems you were seeing that manufacturers were experiencing with AI that really led you to focus on this specific niche?

Sebastian: So the evolution towards this niche for us has been coming from two directions. So the first is the technological direction. We were already doing digital, like I said, digital transformations. And then we've developed both the very strong internal interest into AI. And then it's also where the market is at. There's a lot of not only educational needs, but opportunities within the AI transformation space. And then in terms of the vertical focus, we've just historically had over two decades, a lot of manufacturing companies that we've worked with, not necessarily always all the time, but it's been a recurrent theme. And so last year, it was really last year where we decided that we were going to just narrow down even more and position ourselves really in this space. So it is a fairly new, I would say, verbalized focus. It's been, it was more of a just, that's the, you know, this is where we're at focus before, but we weren't positioning ourselves on our website and so forth saying, you know, this is our niche, so to speak. So yeah, a convergence from two directions, I would say.

Bill: Great. And from your seat, you know, you've dealt with manufacturers for a long time now with digital transformation. And now you're seeing them grappling with the AI transformation and the really early stages of that. Whenever you're talking with the C-suite or an owner or a founder, what do you think the biggest misconception that those folks have around AI today? Where are you seeing like dissonance or disconnect with their vision and what we can achieve with AI practically today?

Sebastian: I would say the biggest theme is a disconnect between the expectation and reality. And that can actually go both ways. It could be that there's an assumption that AI, you just throw it in and it's going to immediately create tons of productivity gains and you just install the thing and start using it and you should see 30% 50% or whatever percent increase in productivity or the opposite, where there's very strong skepticism and resistance and thinking, you know, it's just, it's just hype and it's not going to really change things or it's not that important. So I would say both directions and really at its core kind of linked to the educational needs that are existent right now in terms of, and then there's many layers to that, you know, understanding how it works, but also how it can be applied and then how it can be applied at different layers within people's individual productivity, but also within creating automations or agentic systems is a different way of applying it. So lots of different ways to apply these new tools within an organization. And so that all needs learning and experience.

Bill: So you and I both have a number of years of experience in digital transformation with manufacturing companies. When you think about the adoption rate and the adoption curve, when you think about the adoption rate and the adoption curve of just digital, right? Good websites, good content, SEO, video, the things we've been working on for the past 15 years. Do you feel that the AI adoption rate is going to be higher and faster by companies or in 20 years, you know, in 2046, are you and I going to go back on a podcast or probably our AI avatars will have a podcast at that point. Will we be sitting there looking back on 20 years of AI adoption and still seeing what, a tremendous 50% of the market still lagging dramatically similar to what occurs in digital? I mean, we still run into clients who in 2026 have not paid attention to SEO for years. Do not have a good video strategy. Do not have a GTM motion. They're just wait, their digital maturity is very, very low. Do you think that will be true of AI adoption?

Sebastian: Well, first, if we have digital avatars speaking on our behalf, I think we would have very high adoption if we're already there. But joking aside, I would say it's, it's hard to say for sure in that vein. I have predictions in other areas that I would be able to give more confidently, but I think manufacturing as an industry has had major transformational moments, like with robotics and so forth. There's a lot of companies that I visit or walk through and robotics has made a very big impact. And so the whole production process. I think us though. I think in that regards, AI is going to make a dramatic impact to workflows and, and other processes as well. The, the reason that I'm not sure about certain businesses is just that there are certain level. So manufacturing as a whole is kind of, it's just has this change resistant predisposition that's a little more prevalent than in say some other sectors. But, you know, if I think that through, I think with the new generation that's coming through in the next 20 years, they're going to be bringing in a lot of, for them, the new generation AI is just going to be so normal. And a lot of the things that today seem controversial or questionable are not even going to be, they're not even going to make sense to the next generation. You know, like, um, I was listening to a podcast recently about AI generated music and the person on the podcast who was being interviewed is a critic of music. And his first response was AI music, I can't understand it because there's no soul in this. So it's, it's useless. You know, I don't, it doesn't connect with me. And so he decided that for the next month, he was going to only listen to AI generated music to just see what happens to himself. What if his opinion changes? And then by the end of the month, he was completely convinced that, that he, not only was he, did he change his opinion in the sense that now he wants to make AI generated music all the time. Like he prefers it over normal music. But he said, you know, the shift kind of changed instead of thinking about what soul is in this music that I'm hearing from someone else, it became the AI generated music is the soul of it is me who's wants to make a song about these things. So it's kind of changing the reference point of the music. And also he said, you know, the kids that are growing up today are just going to kind of raise their eyebrows with confusion of why we even worried about there being soul in the music or not. But anyway, so sorry, I'm derailing a little bit from manufacturing, but, but yeah, I would say, you know, so that generational shift is coming as well. And I think that's going to have a big impact. And then of course, you know, there's just market pressure. If other competing companies in manufacturing are all applying AI processes and add automations and systems, you have to keep up and adapt in order to stay relevant in the market. And, and with the pace of how things are going in AI, I think it's, it's going to take over in so many areas. It's, it's just as big as when the internet first came out and you know, nowadays it seems unreasonable that a business would not have a website. You know, it's, it's. Yeah. And it's being embedded in so many things too. That's the thing with, AI is not only going to become or is becoming more powerful, but it's also becoming smaller over time, meaning that you can get smaller and smaller agents embedded within little processes. So an agentic system doesn't necessarily always mean a super smart, powerful one. It can be a very small element that makes important decisions within a larger process. And so those kinds of embedded AI elements are just going to be everywhere. It's just as big as having, you know, right now we used to be writing code with like one plus one equals two and it's very deterministic. And this just lets, essentially it just lets us make probabilistic determinations within automations with an increasing level of confidence that we weren't previously able to make and therefore we're able to achieve new automation types that previously people had to do. Right. So we're relieving people of more micro decisions that they can focus more on straight strategic choices and human relationships and other things.

Bill: No, that's great. And let's take that like more holistic perspective and now like kind of dive into some some of the struggles that you're facing. You're seeing others face and that we're all experiencing as we experiment with AI as we try and adopt it without waiting for someone to, you know, kind of top down, give us this solution. But we're, it's more of this uprising where we all have access to these tools. We're all able to ask it to do things and program and do those sorts of things. So why do you think, and with your experience, what have you seen? Like, why do so many of these AI initiatives stall at proof of concept stage? Like, what are you seeing as you're developing a lot of projects for manufacturers?

Sebastian: So I would say that there are different areas for failure. One of them is a lack of AI expertise, so not understanding its limits and capabilities, or not bringing in somebody who understands that really well. So you might get a very good result within a sandbox, but then once you apply it more broadly or within an automation, it doesn't work, and then it can be frustrating or confusing, and then it just fails there. Another area that needs to be considered is data and process. So if you don't have good processes defined, the AI is so, different things are going to happen. First, the AI is going to be given too much leeway to kind of come up with a solution to what it needs to do. And that's going to lead to non-repeatable outcomes. So you're going to have quality issues in the automation. So you really need, a lot of what we end up doing is we look at the processes that exist or don't exist. And sometimes they're in people's heads, their tribal knowledge, and they need to be mapped out. And that process is either, or it can be both, it's either a workflow process. So it's you need to do A, B, and then you have a decision point. You go C or D, map that all out. Or it could be a logical tree as well. Working with one company right now this week to map out what's in people's heads, or in this person's head to kind of map it all out into a logic tree. It's a really interesting exercise when you get someone to try and really break down how they think through their solutions and problem solving. It's a really fun exercise, but it can be challenging for people to do that because it's not something you're asked to do usually. But yeah, so logical tree frameworks or workflows, combination of those. And then the other area of failure is giving the AI too much agency, too much leeway. You want to really narrow it down. Also lack of testing of these AI systems. People sometimes, and when I say sometimes, it's quite often, will install an AI system. They'll throw some training data at it and then they'll put that, let's say it's a chat bot, they'll put it on their website and then haven't tested it. And then they, you know, and then we who two weeks later, they're getting all these upset customers and then they think AI is not ready because they just expect it to work. And then they take it down and they're very upset. You know, I've seen that a number of times, not necessarily from us, but just in general. And, so too, and so too much agency and data. So if there's not enough data or the data quality isn't there, there's not enough context and training. You're going to get a lot of issues as well. And then, not focusing on ROI is another one. So having too loose of, so if you're going from experiments to practical application, how are we measuring the success of this project? And it could also be too big, is another one as well that I've seen. So it is actually important to have a North Star, some kind of transformational vision and outcome that you're looking for. But it's really important to then break that down into smaller achievable modules or stepping stones towards that outcome. And then I would say also connect an ROI to each of those so that you can see that you're getting a positive return or positive outcome on these steps as you're progressing towards your long-term goal. And then the last one, which you kind of already hinted at, is if it's only experimentationally driven. So I've seen also companies where someone's really excited about AI, maybe they're working in one department, they pilot something that they want the whole company to use. And then it just kind of gets spread across the company, but there's been no, there's no change management. There's no governance policies in place. There's no thought about how we're going to maintain this, how we're going to get, let's say, iterative feedback to improve the system. Maybe we forget to connect it to some, I don't know. There's just a number of things that are just not thought through once it's just experimentation. So there needs to be some formalization on. Experimentation is not bad. I think it's good as part of the change management process and also for the innovation and adoption and excitement. But there needs to then be a process that takes those experiments, if that's the approach your business is taking, to then look at how you're going to formalize that as an actual project and looking at ROI and all those other things.

Bill: Sebastian, one of the biggest mistakes I've seen in AI project launch and like bringing into fruition is treating data as a companion to the project as opposed to launching a separate initiative where you focus on data to make sure that your data is accurate, it is crisp and updated. What have you seen? Did you see people say, we're going to do this AI project, let's clean up the data as well while we're doing this. And then it's just too big of a an ask to resource rich, the requirements are too resource heavy because they're trying to fix two problems, data and then how AI interacts with it. Have you seen that?

Sebastian: Yeah, so I would say yes. I mean, in my sample set, it's been maybe slightly different problems that I've seen around data, but I totally can see that being the case that it balloons. What I would say is there's two things to think about with the data as I brainstorm that with you. I would say the first side of it is what data do we need for this particular project that we're doing? And so if you're trying to clean all the data in general, then yes, you might end up with too large of a cookie bite that you're trying to take in one go. Maybe there's a way to kind of, maybe that project is too large. You know, how can we get so, you know, more practical kind of example, let's say our, our North star is, we want to create a virtual engineer. This virtual engineer would be able to answer internal questions, client questions, but also come up with custom B2B quotes for, you know, what I'm thinking of in my head is one that, you know, client-related, makes quotes for custom heating solutions or welding equipment or, you know, large scale industrial systems, but you have to customize each one for each client. So, you know, can we get a system that helps us to get a quote that's 80% or more there, right? You know, from the get-go, you just give the parameters and the virtual engineer gives you a starting point, but also train engineers, right? So the new ones that you're hiring and then retain knowledge of your engineers that we're leaving. So if we unpack that, there's a lot in here that it's trying to do. Can we take elements of this and start working on them as first projects? So maybe the first step is looking at what are the common kinds of questions that clients and internally are being asked to engineers all the time that is taking up a lot of their time to keep answering? And let's start doing a data capture of that data and you know, meta let's say meta, data engineering being done then on that data set. So meta tagging, organizing data around that category. And then you could create some kind of an internal and or external cat system that can help answer common engineering questions that then alleviates the engineers from having to do that. You're not tackling yet the customized quotes for solutions, but you're already alleviating time from the engineers. You're building, like I said, these steps towards having the virtual engineer and you're building elements of that knowledge. So that, that's the one direction. The other direction of data is that I do think it is important for organizations to be looking at data capture, data retention and organization as a whole. And I think the sooner you start doing that, the better. But I think that can be more of a practice of improving how you approach data and store and categorize it. And then knowing that the better, the more data you have and the more of it you're capturing and organizing, the more services, in this case, AI services, you can then plug into that data over time once you have that data. So data itself can be very valuable to create lower hanging fruit or more easily actionable projects later once you have it. So that can be, you know, having a policy that every single meeting is now recorded and stored as a transcript or written transcript. Emails have better retention policy so that you can also use those as a later capture source for, in this case, you know, if people are asking questions all the time over email or chat systems could be having a data storage policy around them as well. And then you could use AI to help meta tag and organize that data as well if you have an idea. So one, for example, easy data categorization would be categorizing the data based on the role of the person who's asking or answering that question. So, you know, if one person asks another, Hey, how do I onboard to this new customer? The answer is going to be nuanced based on the role of that person. So let's meta tag the answers based on that, for example, smaller example. So I would say. I would split, you know, the monstrous mountain of all data organization into its own initiative separate from the data that's required for the specific project that we're doing. And I think by splitting the two up a little bit, you can maybe make that that project or that, that problem of data enormity be a little more manageable. So yeah, those are my thoughts on that.

Bill: No that's great. And one of the things I've seen with AI projects and implementation, the key seems to be something that's very simple and has nothing to do with technology, but has to do with attitude. And the word humility comes to mind. It is very difficult. I'm an entrepreneur. I want to think about big things and solving big problems. But AI is not to the point where we can just download it and our avatars are showing up for this podcast and doing our jobs and we're off in some dystopia and all of that. So what, where I've seen the most success is where the folks who are involved making the decisions and building the systems are humble enough to do the small things and solve the small problems one step at a time, not trying to create the next most amazing do it all, end all product. Do you see that as being true in the projects you've been involved with?

Sebastian: Yes, I think the smaller it is, the more focused it is, the more successful it tends to be. And the large–you have to, I don't think it's really possible right now to take a very large, you know, three year type endeavor with AI and know that you're going to confidently implement it. Also things might change in the middle of that. That's going to throw you completely off course. I mean, to that point, you're making me also think about. So when we approach digital transformation or, you know, problems we're looking at, if there's, so for example, one engineer, they were, they're very busy. It was kind of connected to what I was saying, but it's actually a different client I'm thinking of. So engineer, very busy getting lots of questions. First principle, is there anything we can do to alleviate their time that may not even have anything to do with AI? You know, and in their case, we, after, you know, through the interviewing process realized they're just answering questions all day long. And it's like, well, let's stop that. Let's, let's suggest they batch their answers and they deal with them, you know, once or twice a day in a batch type format. And that immediately has, has had positive impact on their work week. They're saving already an hour or two of work minimum from just a little bit of a process improvement. So, you know, and that's not something fancy we can put on the website, like improved this, you know, this non AI solution for, but I mean, it's important to kind of always take, you know, the holistic look at the problem you're trying to solve and not just try and use a hammer on everything that doesn't need one. And that's not to say that, you know, from there, you don't start looking at AI solutions to then create further enhancements, but you know, those can be very quick wins. And sometimes we'll go in and we'll put up an automation and it has nothing to do with AI, but that's what's needed, right? There's just, you know, or the AI is like a really small part of the project, but that, again, it's like, you just don't want to get sucked into seeing everything as a nail because you've got the hammer. You just need to hammer everything in.

Bill: I love that. And all of the wins we generate for our clients, whether it's in digital transformation or in AI, will not make the case study. Right, they will not be on a KPI that we promote that we've achieved for them.

Sebastian: Yeah, exactly.

Bill: Sebastian, when you think about repeatability and scalability, what are you seeing that's breaking down first when companies are trying to scale AI?

Sebastian: Breaking down first, yeah, I would say… The biggest hurdle I've seen, because I saw this a number of times last year, is that upper management executives lacking enough personal experience and understanding with AI. And that could lead to all kinds of different problems. I've had situations where they thought they, they'll come into a process, thinking of one company, and then overturn the recommendations to try and get it to be done by a way smaller team, maybe. I mean, in one case, it was kind of very surprising. They just decided to bring in one person who was, I think, to do the whole AI transformation that they had had us scope out, which was a really substantial project. So that was very, like, it didn't even make sense. But they were just like, oh this person seems like they really understand AI. We're going to try using them first. We'll see how far we get. I've also seen that, one company came to us to help them double their business because they had or have lots of demand and lots of capacity, but their bottleneck is being able to handle quotes quickly enough. And so they wanted AI to be able to help in that aspect of basically sales enablement for their sales engineers. But the, and then they wanted, I think it was four different types of important security considerations on the system. So very enterprise-grade build. And they kept adding more and more requirements onto the system. And when we came back, we're like, look, this is not going to, you know, this is going to have a price ticket. I think our quote was in the six, six figures lowish. I mean, not, not very low six figures, but low six figures to get it all done. And, know, we ROI calculated it. Like, look, if this is going to help you double your business, you'll make ROI in one month. You know, it's very, very quick ROI, but it's still not, you know, this is still going to take some time to build. It's not a quick solution. And when they saw that number, they were like, dear God, I thought AI was going to be way cheaper than this. And we're like, you're asking us to double your business. I mean, so I don't know. So it's that kind of, and then, sometimes, I mean, we, so I also did a presentation talk in November to a group of executives down in Silicon Valley, who they themselves work at big tech companies who are also making AI models to not mention any names. And there's a lot of middle upper management that has barely used AI themselves, even in these companies. So I think there's, things are moving really fast and there's almost like this, it makes me think of the coyote who runs so fast that his skin is left behind and has to kind of catch up with the body, or like the soul is the little soul has to catch up. But I think right now we're still in this phase where there's just this, different kinds of disconnects that are going on within organizations. And then we're also in that hype trial where things have been overhyped a lot over the last year or two. Last year was a lot of the disillusionment kind of phase that I was seeing where… So I think what we're as we're starting to see more and more successes that are permeating around not just manufacturing, but in general, in AI adoption, it's going to create more and more positive case studies that are going to help more companies to jump on. And because companies also have resistance sometimes to being on that bleeding edge, you know, they want to see, they don't want to jump in until they're seeing like 51% of the market’s ahead of them. And they're like, no, no, I have to start moving. But that moment's coming and it's really more question of months or a year or two rather than decades, I think.

Bill: No absolutely. The next 24 months are going to be a roller coaster ride. There's no two ways about it. One of the things I think that is a real missing piece to most executives’ and owners' philosophy and approach to AI is there's this preconceived notion that, oh I have I can get a Chat license. I get a Claude license. I can get a whatever for a couple hundred bucks a month. So it should be able to do all this stuff. But then like you're talking about, if they were approached, with if companies were approached with a three, four hundred thousand dollar investment in any other category that would pay back in a year, let alone 30 days, they would, you wouldn't be able to get out of their office without them throwing money at you and having a contract that tied you down to ensure and guarantee that you delivered it. But then when it comes to AI, wow, that's going to cost money? It's not free? Are you seeing that disconnect on ROI comparisons between AI investment and other non-AI investment?

Sebastian: I am and like I said, I think a lot of it has to do with a lack of understanding of what it can and cannot do. And I think it's also though connected with, so there's a lot of fail rates that are communicated on the internet right now. And you know, and on these shows, I mean like this one. So I think people are very afraid of those fail rates. It's interesting though, because so I had one, I was

listening to someone else's presentation and they talked about something called the asymmetrical upside of AI. And it was a really good point. So I'm going to repeat it here. I'm going to take it on for this podcast. So the idea of an asymmetrical upside is even if you do have a project, let's say you have a project and you're trying to increase the output of one person's role by, I don’t know, just for argument's sake, we're to say it's going to double their output with this AI project. And you might fail on it. Maybe you're not that good at doing it and you have an 80% fail rate. And so you try it five times and then you finally get it right, like an average of 20% success. Worst case. Even if you failed on it four times, succeed on the fifth time, you're still doubling that person's productivity forever. And then any new person in that role has doubled productivity. So the reason that companies are trying a lot and want to get this to work is also because the upside is permanent, scalable, and relatively infinite in the sense that you can just keep creating more and more upside in more and more areas as you keep going. So the risks, even if they're high, are totally worth it. And so it's, but it takes, so two things are going on. One is that that percentage of fail rate is going down as more people get experience and learn about how do you approach things and you know, lot of what I profess is here's the way to do it that works, that kind of looks at all the reasons that they could fail so that you don't, we don't do that. But, but, you know, so that's reducing the downside, but the upside is still there and is very enticing. So, and this is part of the reason why it's going to transform everything is because of that upside. It kind of locks, unlocks the limitations that businesses have had until now. You know, usually in a business you can attribute the maximum revenue of the business as some kind of formula related to the number of employees within the company. But with AI, you kind of uncap that. You can have, so for example, in agency space, there's kind of this unspoken limit that maybe about 200k per person is kind of the ceiling of what you could achieve. That's not true anymore. You could be having 400, 500k per employee if you have an augmentation of productivity that is being unlocked from these systems. So yeah, there's just a ton of upside, I guess is what I'm trying to say.

Bill: No, absolutely. I think often we forget how much we have failed in the past when we start on something new. So AI is the new thing. Well, why can't we just get it right the first time? Well, digital we didn't get right the first time. And before that, every iteration, whether it was robotics and manufacturing, whether it was advanced mechanization, whether it was pneumatic or whatever, those revolutions, no one got them right on the first time either. So we shouldn't expect that, that somehow there's this outsized expectation. One of the things that I think AI is going to do that is going to be amazing is it is going to capture and democratize tribal knowledge in organizations. And certainly we're seeing with the baby boomer generation transitioning out of work at a tremendous rate, there's a lot of like intellectual capital that is just, there's a potential of losing all of it. What are you seeing and how are you seeing AI being used to democratize that tribal knowledge and not only capture it, but then distribute it in organizations?

Sebastian: Yeah, so that is something that we've been doing a lot of or had a lot of interest in is that topic specifically. So it's, like you said, it's, it's a process of just taking people's tribal knowledge and then figuring out ways to store that data so that it becomes, legacy data or part of the business's knowledge and how you approach that can take different angles. It could be a mixture of interview process around topics. You can also synthesize data around certain subjects and then have the topic expert review and criticize it. You can use iterative processes where you have data and then the knowledge expert reviews that data and then flags it as they're going as been being inaccurate or improves the answer. But other techniques we've used is also looking at someone's email history over the last 20, 25 years. And then you can use a process of going through all of that data automatically. Hint, it uses AI to go through that. And then you go through all the data and then you can categorize, you know, oh this was a Q&A on this, this was related to this topic. And so you can start to pull pieces out of that as well. That can be difficult. So it depends, you know, but that is at least something you can look into. And of course, you know, specifications, other things, all their webinars they've done. If they're, you know, a person who's done lots of that can be used. So there's lots of different ways to get that data so that it can be retained. But, yeah, it's something that is very powerful because it not only helps the organization to retain that knowledge, but it also helps for being able to train new people who come on board by being able to have access to that knowledge, whether it's for the training itself or afterwards to be able to ask the ghost of the boomers past, you know, what was their expertise, you know, now that they're retiring. So yeah, it's a really exciting innovation. I think it's fascinating that culturally it's appearing exactly at the time that the baby boomers who are the largest segment of our population right now are all retiring. So it's kind of, it's like the miraculous entry at the end of the of the story that comes in to save the day. But yeah, it's a very highly demanded or high interest area right now, especially in manufacturing. And there can be some resistance to it as well, because there is also AI resistance in general. I could talk about that because it relates to change management. But some companies are really into it. Sometimes the engineers are really into it. Sometimes they're the ones who are really not into it. So you have to kind of get through those parts of the process as well. If you're interested in knowledge retention. How you frame it is really important. You know, like it's, not like we're trying to steal your intellectual property, but it's more like we're trying to retain your knowledge so that we can continue to help train excellent engineers. It'd be the same as asking someone to write, you know, training materials, but you're doing it in a easier, quicker way than we would ever be able to do before. I think that's the big advantage is that this used to be so costly and so difficult to do that you would just never be able to do it. And now there's a way to do it that makes it be cost attractive or not as costly to achieve.

Bill: No I think that's great. And I think there's a tremendous opportunity. It's low cost. It's low friction. Adoption is going to vary based on the organization and the people involved. But there's definitely an opportunity of people willing to resource it and put some emphasis behind it. So whenever we look at AI sophistication, we're starting to see it really impact M&A, private equity and valuations. And some of it, as we've seen recently, is super hype. Like every deal thesis tagged on the two letters AI. So yeah, we're AI. We're, we have an AI component or we’re AI enabled. And I think the market has already sniffed out really quickly the, been able to differentiate between what's smoke and mirrors and what's real, because there's, well, they can go into AI and ask it and see is this true? But how do you see that impacting investment strategy as we move forward?

Sebastian: So I think what I would say to investors is that looking beyond AI, I think that's important to look at is what is the management team or executive team or the innovative team in that company, what are they really trying to achieve and how is that going to really translate into growth of the business, whether that's improving their top of the line or the bottom of the line. And what's the culture of innovation and change within the company? Because a comp, whether it's AI today or VR, AR tomorrow, or, whatever that thing is that's going to be the next hot thing. You have companies that are resistant to change and those that are early adopters and innovative and, and nimble. And then in terms of leadership itself, you know, is the leadership able to take something that maybe starts off as a mandate, you know, we want to see 20% growth from all these or 30% growth from all these AI things or, whatever the target is, is leadership able to take that and transform it into a vision that is something that's more aspirational, that is inspiring, that people are able to rally around because they themselves see benefit in it. And if so, you know, that's, that's one indicator that there's going to be a better chance of success in that organization to actually achieve those targets, those changes, you know, so for example, a vision that's like, you know, we're going to use AI in order to help you to unlock more of your time to be doing the things that you love to be doing. And, you know, so we're going to go through different departments and figure out how we can enable your team to free up your time. So you're not doing as much drudgery. Something like that is way more inspirational than just saying, you know, all right, so we need to figure out how to increase your productivity by 20%. And, you know, this AI is going to come in here. Don't worry. It's not really going to take your job, wink, but it's going to start doing your job. You know, that's.

Bill: Yeah, if they wink and they nod, that's when you have to worry, right? If it's just a wink, you're okay. Well and I think one of the things we're seeing is the speed of AI is also impacting the ability of companies to window dress with AI. If they have a open-minded and forward-thinking approach to how AI is going to impact the business, it can have a direct impact on the valuation. If it's just a bold on window dressing, I think the market has already been able to sniff most of that out and we're not seeing as much hype or the hype that is out there is getting blown apart really quickly and found out because we're able to see through with a lot of examples of success and failure that are being experienced in the marketplace with various implementations.

Sebastian: I would say that evaluations inflated from AI only exist as long as the buyer isn't educated enough to be able to evaluate the company properly. So I think any uneducated buyers, beware, but educated buyers who are using consultants, advisors, experts who can evaluate the company's abilities, capabilities, and also successes with AI correctly shouldn't be in a trap of over-evaluating an AI-enabled company incorrectly. But that does require some investigation. I think certain companies, I mean, there's way too many companies that say they have AI this, that, and… And the AI in it is really not even that useful. And I definitely don't want to pay for that extra feature. I've seen that too, and some tools that add AI in it. And I'm like, I don't, no. This is not how I want to be using it. But you know, that will subside. There's going to be some bubbles bursting. We're at a, it's a very similar moment to when the car was first invented. I think there was something like, I forget if it was, I think it's around like 2,500 car companies existed for the first few years, and we're down to what? Like five, and there's only been one new car company in the last two, three decades, Tesla. So it's a very different market now, and there's so many AI companies. I think the statistic is from last year sometime, but it was something like 95% of these AI companies that are new have gone bankrupt in the first two years of their existence or one year of their existence. So.

Bill: Pretty high failure rate.

Sebastian: it's a very high failure rate. It's, it also reminds me of when computer graphics 3D was a big thing in, in 2k and they just 3D everything, you know, and it was, it wasn't even that good yet. So you had like lots of movies with kind of subpar out of nowhere, 3D modeling everywhere. And it was just like, ugh. You know, and now it's gotten way better. And I think we're all so not as obsessed with using that paintbrush on everything. But yeah, there's just this obsession with needing to put AI everywhere. And then, you know, and I think the problem also that you get, sorry, I'll switch subjects soon, but I think the, when a financial investor or when the board of investors or whatever are like, Hey, this AI thing is really important. We need to be doing it. It can become one of those. So you're creating an incentive environment and then the incentive is, we need to put AI in order to, you know, have our evaluation still be correct or be stronger or meet a target or whatever. And so it becomes just performative metrics. And then the company that's putting those mandates in place, maybe their incentive was just, oh there's AI. Great. There's AI. We're all happy now. Maybe our investors aren't going to investigate them. Our shareholders aren't going to investigate deeper either. So we're done. But there's a big difference between that and actually creating real market change, you know, market penetration or being able to break into new markets or innovate new products or, you know, the, if the incentive is coming from the business itself in order to actually actualize you know, 20%, 50% gains or doubling the business in certain areas or something like that. Like that's when you create real change is when it's driven by a real motivational force and not just hitting a target.

Bill: Well, Sebastian, this has been a fantastic conversation and I want to give you the opportunity. We are all about shameless plugs, so I want to give you the opportunity to give a shameless plug about your business, what you're doing, who you're helping, and then also how you can be reached. All of this information will be shared in everywhere we put this and throughout social media, but I want you to give you the opportunity to verbalize it.

Sebastian: Yeah, thank you. So Fountain City is our digital transformation company, primarily focused on manufacturing companies, but we also help other companies too, midsize and larger. And on Fountain City, there's a newsletter that you can sign up for, there's a blog, and I also have a YouTube channel that I post videos on roughly weekly, but I'm more focused on quality than quantity. So it's whenever the next one comes out. And then if you're interested in looking into a tool that helps you test AI systems to make sure that they're compliant, not hallucinating, giving you good knowledge answers and so forth, we have a SaaS platform. I'm co-founder of a business called testfox.ai. It's still early stage business. So it's actually free if you sign up. So you can try it at no risk. It's not like it's free and then we make you pay after a month or two. It's just completely free right now. So you just use it. And then you can of course also connect with me and follow me on LinkedIn. I'm not the most prolific LinkedIn poster, but I, you know, I'll post there when new blog posts and come up or related to newsletter things and things like that. So those are the, then if someone was, I also do help companies independently in quotes with advisory or consulting services. So if someone is interested in my help for, it could be consulting, advising, but also workshops I have right now I'm preparing, well, two workshops actually I have coming up. So I also do that. And if people are interested in that, they can reach out to me directly for that too. So yeah, I'm very happy to be here. Thank you so much for inviting me.

Bill: No, Sebastian, this has been a wonderful conversation and I want to extend an invitation. We need to run this back again when we have more time. I say that, we've already been an hour into this and I think we could go on for two or three more. But it's important to have these conversations and really tackle AI with as many as much collaborative effort as we can so we can all extract as much value and provide as much value as we possibly can to our clients. So, yes, thank you again for joining us. It has been a pleasure.

Sebastian: Yeah, likewise. And yeah, anytime you want to invite me, I'd be happy to join again. So thank you so much. Been a great conversation. Thank you.

Bill: Excellent. Well, thank you for joining the missing half where today we've discovered what's missing in approaches to AI and manufacturing. Like, share and subscribe. Have a great day.