Inspired with Alexa von Tobel
00:22:18 · Daniel First, Alexa von Tobel
[00:22:18] S0: I think that people really get wrong the way that AI is gonna impact manufacturing. I think that a lot of the discussion today is all about how AI is gonna be... Bring robotics and automation, and that's true. But that all really is about reducing costs. And I think that that's not actually how, you know, the CEOs that I talk with across motorcycles, medical devices, or consumer electronics, none of them are looking to compete that way. The way that they're they're less looking to to copy lower cost regions at a lower cost, that... You know, that's not really how they're looking to compete. I think that the future of American manufacturing is one where people are... They think about empathy with their customers at scale and understanding customer needs that nobody else understands. And I think that AI is enabling us to have much more rapid feedback loops between customers experiencing issues and iteration, where I think that it used to be people would try to be first to market. I think what actually is going to define American manufacturing going forward is the speed at which you can learn about your customers' needs and iterate your product in response to that. [00:23:33] S2: That sounds amazing. What else can you see? What do you feel like, you can see maybe speeding up, happening more frequently? And also, what problems do you think go away entirely?
00:23:33 · Alexa von Tobel, Daniel First
[00:23:33] S2: That sounds amazing. What else can you see? What do you feel like, you can see maybe speeding up, happening more frequently? And also, what problems do you think go away entirely? [00:23:44] S0: I think that, you know, right now, I think that, you know, many companies, they have, you know, portfolios with hundreds to tens of thousands of SKUs. I think that the public doesn't understand the degree to which most of these, you know, hard, manufacturers or consumer brands, because solving... Because learning about a problem and solving it is so arduous and time consuming, usually, the humans who are working on this are only solving problems on the top 10% of the SKUs that are most important to that manufacturer. So imagine instead a world where every SKU that a manufacturer creates is constantly iterating and evolving in response to customer needs. I I think that that world is gonna look really different, that I think it's gonna be actually a much more empowering world for for people who are using products to solve their problems because you can give feedback and then see the next version that that things are solved.
00:23:44 · Daniel First, Alexa von Tobel
[00:23:44] S0: I think that, you know, right now, I think that, you know, many companies, they have, you know, portfolios with hundreds to tens of thousands of SKUs. I think that the public doesn't understand the degree to which most of these, you know, hard, manufacturers or consumer brands, because solving... Because learning about a problem and solving it is so arduous and time consuming, usually, the humans who are working on this are only solving problems on the top 10% of the SKUs that are most important to that manufacturer. So imagine instead a world where every SKU that a manufacturer creates is constantly iterating and evolving in response to customer needs. I I think that that world is gonna look really different, that I think it's gonna be actually a much more empowering world for for people who are using products to solve their problems because you can give feedback and then see the next version that that things are solved. [00:24:40] S2: That's so interesting. What you're basically saying is right now, if you're manufacturing company, your... All of your brain effort focus goes to the top 10%, wherever the biggest headache, pain point, etcetera. And because of the proactive nature of what Axion is doing for your customers, they're actually gonna have a lot more surface area. So... Sorry. Is the punchline of that just like far better products for all of us in every SKU, every sort of on every dimension?
00:24:40 · Alexa von Tobel, Daniel First
[00:24:40] S2: That's so interesting. What you're basically saying is right now, if you're manufacturing company, your... All of your brain effort focus goes to the top 10%, wherever the biggest headache, pain point, etcetera. And because of the proactive nature of what Axion is doing for your customers, they're actually gonna have a lot more surface area. So... Sorry. Is the punchline of that just like far better products for all of us in every SKU, every sort of on every dimension? [00:25:10] S0: Yeah. Absolutely. I think it's gonna be much better for customers. I think innovation is gonna happen much faster. I think also that the humans who work at, who work at these manufacturers, instead of spending all day reading line by line through data, you know, AI is gonna take care of all that. AI agents are gonna be learning about customer needs, proposing iterations to products. And I think it's gonna be much more exciting to work at a manufacturer, which, you know, today, it is for some people, but it may not be for others because they have to spend all day slaving away at the data. I think in the future, humans will be presented with complex decisions about customer needs. And I think that... And and and how... They're gonna decide how to react to that. Think it's gonna be a much more exciting future to work at a manufacturer. [00:25:50] S2: I'm pumped about when my my stroller no longer breaks and I have to deal with trying to figure out how to get a wheel replaced. That sounds like a beautiful feature for people here.
00:25:50 · Alexa von Tobel, Cameron
[00:25:50] S2: I'm pumped about when my my stroller no longer breaks and I have to deal with trying to figure out how to get a wheel replaced. That sounds like a beautiful feature for people here. [00:26:00] S3: So, Daniel, there's there's studies out there that show that the vast majority of enterprise AI pilots fail, yet Axion has been quite successful in converting real contracts in industries that have historically been very slow to adopt technology. What do you think people are missing or misunderstanding most about AI? [00:26:19] S2: Yeah. How did you... Tell us more. Like, what did you get right there?
00:26:23 · Daniel First
[00:26:23] S0: I joined McKinsey in 2016. And, you know, it's really funny reading this this study that came out, you know, this year about the ninety five percent of AI pilots failing. I mean, I didn't need the paper. I, you know, I I watched myself and my friends at, you know, companies like c three or BCG Gamma that were building all these enterprise AI pilots. We watched them fail in real time. And it was very easy to make a flashy demo, but then when it came to actually bringing things to production, it often wouldn't make it. And it was really, it was really soul crushing, you know, bringing all these really brilliant colleagues to enterprises, having this big flashy project, and then seeing time after time, it didn't make it to production. The the projects that failed most consistently were ones that, you know, I think of as cross functional intelligence workflows, which is bit of a mouthful. But, basically, people would would hire, McKinsey or QuantumBlack pretty much always with the same problem formulation. It's a very unique one. It's, hey. We have a lot of different groups with different data across the enterprise. They... You know, everybody... If they could just link up all their datasets into this omniscient pool, then suddenly everybody would be able to do their work much better. So I worked with a professional sports team that had twenty twenty six different datasets. They're trying to make decisions about who to play and game strategy and injury. Could we link up the 26 datasets? I worked with a clinical trials study. Can we link up all the data about clinical trial audits? Ultimately, the reason that these projects were failing, it wasn't because the AI didn't work. It was because there were these very big gaps between the prototypes that were created and, and what you actually needed. So there was a a knowledge gap, which is that when you showed the results of the machine learning to experts, they would say, you know, my expert knowledge is not incorporated. There's a process gap, which is that people had very legacy processes. And so, you know, they they they... There was no way for them to really fit the AI into their... Into a new process. And number three, you know, you might have AI models, but humans would need to dig through and find insights and somebody would need to take action.