Announcer: You are listening to augmented ops where manufacturing meets innovation. We highlight the transformative ideas and technologies shaping the front lines of operations. Helping you stay ahead of the curve in the rapidly evolving world of industrial tech. Here's your host, Natan Linder, CEO, and co-founder of Tulip, the frontline operations platform. Where are we? Madilynn: So we are in hall 15. On a little stack of cubes looking at AWS and the built for industrial AI sign. So it, it wasn't enough to do it in the lobby and then the restaurant, we had to actually bring it to the fair. Mm-hmm. Today Natan: it's like maybe fourth year in a row. Madilynn: I think it's the fourth year. Yeah. Natan: Time flies. Madilynn: Yeah. Which you were mentioning. You have a little bit of a rant because time flies, but maybe also some things, yeah. Aren't changing. Natan: Our industry pretends to move fast, but in reality it is kind of slow. Definitely. When you compare it to the past, say couple years with all the rage out there on AI and friends, I'd say Madilynn: yeah, Natan: it's just not the same pace, but it's. If I go back a couple years ago when we sort of first time put Frontline copilot out there with Microsoft, it was the only one in the show. This show is like Copilot Festival of Sorts. Madilynn: Yeah. Maybe that's a new name for Hanse Copilot Festival. Natan: Copilot Festival, yeah. But we'll get to the rent, I promise you. Okay. Okay. But before the rent, it's kind of a strange ville because I was walking around yesterday and I was thinking about it. It used to be that you would walk into, um, some hall and let's just say you're in, uh, the water industry and you can get all the things for water there, pumps and pipes, and any widget you need and control and software. This is like the second year. I think that Hall 15 and 17 are practically digital. Yeah. And I was thinking this is the only conference in B2B Enterprise that you can actually see that I. Cross vendor action. You know, you want Salesforce, you go to Dreamforce, you want AWS, you go to reinvent. So all the IT world stuff is very segregated, kind of by nature. Mm-hmm. And here they're almost like forced to collaborate. It's kind of weird. I gotta say. Yeah. Madilynn: Yeah. I see what you're saying. It's like, because it's manufacturing, everything can be in the same area, but there's actually a lot of difference. And so that was one of the things when I was walking around yesterday, I actually had a, a lunch conversation about it where I buy software, you know? Mm-hmm. I don't buy manufacturing software, but I buy and evaluate a lot of software. Natan: You have a software stack? Madilynn: Yeah, I have my, I have my own stack. Like we have our own ops team. Like I understand what that looks like, and I've been to some of those conferences. You have citizen developers. Yeah, exactly. You, you're Natan: into Vibe Marketing. Madilynn: We use ai, you know, of course. But when I walked around here, you'll have two companies that like service completely different industries, do completely different things, use all the same words. It's like, oh my God, what a stressful and like, unhelpful evaluation and shopping experience to just be like the digital hall. Like there are so many different things in this hall. Natan: So my conversation, so of course we meet a lot of customers here and uh, partners. It's great. But, uh, I was talking to some in the conversation, more or less goes like this. I'm really sorry for your experience in Hanover. Customers are just confused, like you come here to learn all that kind of stuff. Yeah, and I think you come out completely confused and it's reflective of this state of the industry. And maybe that's the, could I segue to my rent? Yeah, Madilynn: no, this is a good, this is a, Natan: so why are customers confused? Well, first I think there's this. Catch up moment, like industrial, everything wants to be with the cool kids, you know, open AI and Nvidia, and they are to some degree, but it's so early. Yeah, so my rant starts, I've been walking around and I've been collecting all these. Things that companies say we we're not gonna get specific because we love everybody. Yeah. And a bit of disclaimer, we're biased. Madilynn: Of course. I mean, and, Natan: uh, we do it too. Madilynn: This podcast is sponsored by and paid for by Tulip. Natan: Yeah. And, and, uh, we forget Tulip. We probably do the same to be completely open to a degree. Oh, of course. We'll be open about how Madilynn: Yeah. Don't send me a mean message. I get it. We, we also do marketing. Natan: We try to do better. Yes, exactly. We do try. So with that, the general syntax is like X AI X. So it's like there's a prefix and there's a suffix and anything goes. So you can say Cat ai freeway. There we go. That works too. So you can try it. You can say, I'm just looking around light way. AI intelligence. Madilynn: Yeah. A lot of intelligence. Natan: Twin AI factory. This, this is kind of like the general thing. So now you see stuff built for industrial ai. Okay. And then there is bring your AI to the factory reinvent core value changes as engineering from Azure production with AI data and AI expand data center capacity for ai. It's crazy. It's like the humans going to work for ai. We'll get to that in a moment. Yeah. 'cause this is kind of really annoying me. Power AI with edge computing. Unlock ai. Unlock it from what? Where is it locked? Actually do, do we know Madilynn: It is, you know, locked away. Not given access to all the data to just do whatever it wants. Probably because it's too early. Right? Natan: Maybe it's like a teenager. You don't wanna let it loose. Yeah. Or toddler. Madilynn: Maybe Natan: more likely. Toddler. It's the terrible tools. Hey, it's the terrible tools of ai. There we go. There we go. I think we got this. Madilynn: I think the funny thing about what you're saying is two years ago where like everyone's late Natan: Yeah. Madilynn: To AI and now two years later it's like AI everything. Yeah. And it's too early for the scale of the ai, Natan: but Madilynn: marketing, Natan: however everyone is promising scale and impact. Madilynn: Yes. Natan: So again, so we're building up to the, so the crescendo of this rent. So then a few more, but we had enough. Madilynn: I think we've had enough. I think people get it. Natan: People get it. Okay. So there's a lot of noise. They're tired. We're all tired. There's a lot of noise. So there's a lot of noise. And a lot of confusion. Two things that I see that are completely missed. They're kind of there in the very high level positioning at best, I'd say. But where is the AI to work for the people who actually do the work? So we have our point of view. We'll talk about it in a second. Yeah. But the crazy thing is that people say edge and data and break the silos and uh, all these kind of things. For what? To create a smart factory, intelligent factory, this, that and the other. What makes it a smart factory above all the people in the factory? Yeah. You know, the automation they put in the tools, they choose whatever, and it's, at the end of the day, it's people and. The crazy thing, you go around and we might get heat for the, what I'm gonna say right now, so just, okay, let's, let's be ready for that. All right. Like you walk through all these booths and you see basically legacys of sorts. I've seen less of them. Yeah. The booths look the same. I call 'em the talk therapy booth. So they have the positioning. People sit there. And I don't even know what they're showing. And they talk and they're pretty empty. And then you go to some of the large vendors, you see the little booth inside. You know what often you find there? Uh, the system integrators. So you wanna do this smart factory thing. Some folks still don't get it. That ai for the people means one thing, it's been true for decades. It's about developers who's driving, who's pushing the pace for what happened in the internet, you know, since the beginning of that revolution. It's people who are adopting this technology on behalf of their organization. Yeah. So what we think about is frontline AI and building tools that eventually will become real, you know, at the risk of our own marketing step function, a step function in productivity, and what people can do. It really lies with can they use it to act like developers in their own operation? Because that's the only thing that would make it sustainable. Madilynn: And I think the thing that you're pointing out, like I think there's some nuance when you talk about system integrators. Like we love system integrators, we work with them, we do. They provide like great service and insight and they're like a good source of knowledge and leverage to the industry. There are not enough system integrators and you cannot have them on site all the time to. Run operations and deal with the amount of change that manufacturers are dealing with, and so it becomes really necessary to. Think past the systems and like your technology strategy and really think to your point is like how do I give ownership and power to the people that are actually gonna make decisions that impact my business? And those decisions are a bunch of little tiny decisions that add up to this big thing. And one other note from a conversation, I've been meeting with a lot of analysts here. It's like a lot of really insightful conversations. We're about halfway through, so like eight different firms I think so far, and two different conversations. Someone asked me if the skills and labor shortage still mattered in manufacturing. Natan: Oh my God. Which Madilynn: shocks me, right? Because every single one of our customers. Cares about that. And for that to even be a question shows how removed, I think some of the analysis on technology and market becomes from day-to-day operational realities. And it is so important that there isn't that gap that that's bridged and that it becomes a real principle in the way that we design and think about product development. And that ties into ai, like how do we think about AI as a. Technology or a force that we can incorporate in our technology that will enable those people to do their jobs better. Natan: Yeah, totally. Tons of conversation. Sounds like this. When you go and see stuff, it's like people are kind of referring to that as if it's like this magical, mystical character. The AI does something, but, and I was asking Che on the ride, just full disclosure, Madilynn: hearing consulting with the enemy. I was consulting insider information. Oh, Natan: the partner? Madilynn: Yeah. My buddy chat, Natan: whatever you wanna call it. But uh, I was asking like. How do you say nicely lipstick on a pig? I'll explain in a second. The answer was, uh, a fresh pane of coat on an old structure. Okay. And the reason I'm saying that is because it is the same freaking demos from the past three years for the same software that is likely 15 or 20 years old with, um, some chat bot experience. And I've seen several demos that generally look like. Here is five simple steps to get ai, and our AI will do X, Y, Z. But what it does is just shovel stuff across siloed product experiences in one's sort of semi walled garden, you know, garden with holes 'cause everybody says API and ecosystem whatnot, fine. But it's really nuts. Architectures don't lend themselves for this type of stuff. You see literally like a legacy type of workflow builders of all sorts that a process engineer would say like, Hey, design me a process and we'll go to text. And then they go, I'll tab in Windows. They get a studio, build something then, and then they go create this workflow, and then Al Tab again goes to get another tool that then talks to the digital control system. And it is just such a fresh pen of code and old structure. Yeah. Madilynn: AI washing. Natan: AI washing. Yeah. It's just not gonna do it. And independent ai, and, and we've been saying it all along, like, why focus on Frontline? And because a lot of the conversation here are about impact and ROI and you know, legit conversation. Why the heck invest in a platform and what does it do? And. And so we talk about ROI productivity and call all the usual things. And we all have the numbers. We all know the stuff works. There are things that are harder to measure, like what's, and you know now more than ever, everyone thinks about budget and the world is not getting less complex and that's critical. And TC o's a form of ROI. But how do you measure what's left after the platform has done its impact and you have augmented your workforce? What can that workforce do next? Madilynn: Yeah. Like how do you measure opportunity when you're thinking about return? Yeah. Because return is really focused on the past. Right. And when you're making decisions that set your company up to be able to adapt in the future, we're not really set up to think about value. Yeah. In that way. Natan: And all of this was like. The first problem, just to recap it, it's like there's no focus on AI for the humans. Mm-hmm. And maybe this is the, the point to say, what are we actually doing? How are you positioning what we're doing? And maybe share a bit of that. Madilynn: Yeah. And you and I talk about this, right? I think there's like a tulip cultural focus that seeps into our marketing on being precise and authentic. Yeah. And so. You know, when we talk about principles, when it comes to ai, it's how do we enable people to make content they need and get data they need in a better way, right? That there's a lot of low value kind of administrative work. And you know, in other systems that administrative work becomes like outsourced customization in a quote unquote, like out of the box implementation. I mean, Natan: people today work for the crappy software. They bought 10 15, that's a better way. Put it years ago. Madilynn: People do end up working for those systems we work for. The systems don't work for them. The system don't work, don't work for them. And so when we think about, I mean we have a no-code platform. Yep. So no-code is like, how do we reduce the admin? Right. The cost of you working for the system. A lot of the AI work that we're releasing now, like our AI composer is how do we make that even easier, right? Let's take away like 80% of the work even then, but still have a person there, like reviewing what they're doing and making sure that they agree with the decision and they're doing like transformative updates. But if you need to translate to 28 languages, what if that took you like 60 seconds to do 80% of it, and then the rest is like final reviews. And I, I think that's true when it comes to agents too. Like that's where I've been, the punchiest. I'd say Natan: they're not here, by the way, the agents, that's not, Madilynn: well, I'm seeing agent stuff all over. They're here in maybe, uh, they're here in spirit in, they're here in marketing, Natan: but they're not here. In essence, Madilynn: what's the difference? Natan: Like, this is an agent, blah, blah, blah. But you know, for me, first of all, I think this is temporal. So eventually we'll stop using this word agent. Yeah, I think it's a trendy bird. It could take a long time, but to me it's just the ability to offload some. Real piece of work to some degree of autonomous compute, we call them back in the day, quant jobs. So like scheduled some procedure that goes and talk to your systems, make some decision, maybe gets an input from user. You know, even back in the day of system administration, technical operations done all day long. We just call them agents now and the dream of like. We will be able to directly replace humans that are doing stuff like this. You know, it'll come, I think, totally Madilynn: unsupervised here, but it's not here. Decision making is like kind of what people are selling and also on, we're not ready for it. Right? Maturity wise. And two, I think in most industries there's a lot that you don't want to just. Automatically start happening without your people there. But a, a vision of one person who previously like, had all this like administrative burden and like, you know, mental load of keeping track of things that you could just schedule out. Like giving them basically a digital team to help them do that and let them focus like. Their critical thinking and like unique human intelligence skills. On the other part, like one, I think it's critical when we talk about actual like workforce size and skill levels, but two, that is where you get value from your team as a business. That's where you want the bulk of their time to be spent. But Natan: you know why it's not happening as quickly here. There's two reasons basically, like you can't even compare. It's like the amount of people who sit behind a screen all day long. They have their platforms, whether they do marketing or hr, whatever, and they can have an ai, teach them how to build it, what we call agent workflow in safe and kind of useful platforms that don't jeopardize sort of, maybe they do, but maybe they're less regulated or the consequences of making an error lower. They sit there and they can build all sorts of agents and, oh, I built an agent. It looks at whatever, does market research for me and automates my email flow. Yeah, that's nice. Until we'll find the deficiency in having that, you know, company with tens of thousands of people not in manufacturing has like I. 50 million agents and like now, you know, deal with that. So good luck to them. We'll, we'll see. But at least now it's like the pace is picking up and obviously software people are leading the charge with all the code and all that kind of stuff, but what's slowing it down in operations or kind of the distance between the positioning and the marketing and the impact is one simple thing. It's just trust. The people in operation don't have safe and useful enough environments to iterate quickly to get results that they can actually put to work on a consistent basis. And so this is holding back what we call agents. I. Independent of, you know, there could be a nice PLM tool with like nice bomb information that can annotate a CAD and that's nice, that's taking some work and or you can use it to like create PLC code. You know, that's no difference than making JavaScript code from my perspective. And those are real things, but by and large, what we've seen, what people call AI is just the evolution of enterprise search. Injected into the existing experiences, and that's just not good enough and it's not gonna get people in their organization to a more perfect state of autonomy, which I think is the new word in the new utopia that people are talking about. Where operations, you know, factories and labs and assembly floors, all those places where physical work happens, they actually have a state that's captured in between all the tools that are implemented in between the data models, in between the actions people take. And if you want those things to withstand the test of time, you have to deal with how people can continue and iterate with this stuff because this is not one and done. You don't just like, now I have an AI that does this. Yeah, that AI is like as good as the last contextualized knowledge model that you had yesterday. But guess what? 15 people on the other side of the factory changed something. And what happens now? So I think platforms, and this is again, we're gonna let our products speak for itself and people, you know, feel free to send us hard questions and we'll take them. But, uh, platforms definitely help. And the other thing that can help, and this brings you to item number two, so it's not gonna be as long as item number one, I promise you, because it's a very short, simple thing that we're missing here, which very few people talk here about the requirement to having interoperable ai. Madilynn: You know what? I maybe disagree with that statement. Really. I've only been having conversations about interoperable ai. Okay. I wanna hear the rest of your rant. I, I would just challenge that statement and it is definitely your experience, but I think that maybe we're just talking different folks I, that it is a topic I examples, Natan: like, again, good stuff, you know, you want to annotate. Give context to, you know, CAD models that jump between systems and one LM can look at that, another L can look at that. And that's nice. And I guess what I'm trying to say, I, I hear people talk about it, but I just in complete disbelief in how it's not a more central topic that people actually working on. And I think it reflects the state of our industry where. Developers are not ruling here. That's my main rent. 'cause like there's all sorts of ways standards emerge and obviously bigger companies and small companies can, you know, bend together, do stuff and create a new standard. But the thing that makes standard real is developers voting with their feet out hands, fingers. But whatever they use the program now. Madilynn: Yeah. Their voices Natan: nor lynx, whatever they vote with. But right now it's like voting within your clan. Madilynn: Yeah. It's, it's really siloed and there's that combination. I, I, I think it's right for people to be distrustful. I think that the tech isn't quite there, that it's really focused in, I think the word platform's actually showing up everywhere because people wanna do a land grab for kind of that orchestration of different agents that you're starting to get at. But I do think that that's kind of like a vision for the future. And so there's a lot of both. Like, I mean, this is probably the shiniest version of the marketing at Hover essay I've ever seen. Like there's a lot of. Cool, but also sameness around the show and I think it's because there's some wait and see in terms of what people respond to and how tech continues to develop, which is a little disappointing, I think. It feels like we've made so many strides with the royal we, while also being kind of in the same place. Yeah, Natan: but what are you hearing about interoperability that now you got me? Yeah. Very curious. Madilynn: So one of the conversations I had was about, um, MULTICONTEXT protocols and what anthropic is doing there, and it's like, well, how are you guys thinking about that? Like when you think about agents, when you think about your just like AI strategy, how are you thinking about how it's going to interact with other agents? How does it interact across your ecosystem? Those are good questions. And I appreciated the conversation I had, but it came up a few times. Natan: Yeah. But when I talk about interoperability, it's not such a difficult sort of future to predict where Yeah. Multiple models of all sorts, by the way, not just generative models, not just LLM, but also S LMS that run at the edge. But also stuff like classic deep learning of all sorts that, you know, how we orchestrate that, which is, there's a name for that. It's good old fashioned APIs. Mm-hmm. And so we don't have to overcomplicate this, but what's happening today in your ability to iterate between models, pass the context along and make it easy for people who want to run one NLM within certain platform. Let's just say tier one ERP, companies of sorts that have their everything AI built in. That could take care of, you know, the classic ERP functions, supply chain planning, all these kind of things, and approach the manufacturing execution, whatever that means to whomever's listening here. You know, we think the definition of MES is rapidly changing, and that's great, and that's critical for that millions and millions of dollars of in multi-year investment there. But then there's like 15 other things that talk to that tier one ERP that have LLMs and this and that and the other. That is the kind of interoperability I'm talking about. Yeah. People are now flying the flag high. We have an AI for X, our AI will do as if it's like another character, Madilynn: right? I mean, some of them have names. They are characters. Natan: Yeah, they are characters. But I don't think there's enough work on that level interoperability with thinking about the people who live in between those systems or platforms or what have you. Madilynn: I think that's a good note to maybe leave our listeners on and maybe a little, little more of a positive note. Uh, yeah, just something to think about. Thank you for joining me here. Announcer: Yeah, it's been a great hot over. Madilynn: All right, until next time. Announcer: Until next time. Thank you for listening to the augmented Ops podcast from Tulip Interfaces. We hope you found this week's episode informative and inspiring. You can find the show on LinkedIn and YouTube or at tulip.co/podcast. If you enjoyed this episode, please leave us a rating or review on iTunes or wherever you listen to your podcasts. Until next time.