Rick: What we would call MES today. 80 85 percent of the spend, I mean I saw this front line because I was in the integration business, was on custom solutions. Now, I think we're starting to see this slow but meaningful shift to empowering the, eventually the end users to do this themselves. 

Announcer: You're 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. 

Natan: Welcome back to Augmented Ops.

This is especially exciting moment for the show. We are ready to start a new season. This season, we have a bunch of great guests lined up and no shortage of fun topics we're going to cover. We're going to talk about composable architectures, evolution of the shop floor technology stack. What is industry 4.

0 transitioning to 5. 0 and do we even actually need to talk about this? And what is a quarterback platform? It's unclear, but we're going to talk about all those things. And as usual, we're going to bring the community, the ecosystem together, and we're going to talk to people who are running around the shop floor, busy doing things like system integration or understanding their data ecology.

Whatever needs to get done to change a company's work and optimize their operation day in, day out. I'm also equally excited that Erik Mirandette, our Chief Business Officer, is going to join us, co hosting the show together with me. There's just more stories to tell, more factories to visit, and more exciting things to cover.

So, here we go. Hey, Rick. 

Rick: Natan, how are you, buddy? Oh, I'm good. 

Natan: It's been a while, man. You were one of the OGs that helped, like, hey, we gotta start talking about this stuff, frontline operations, a couple years ago. 

Rick: It's been a busy two years, too. A lot has changed. 

Natan: A lot has changed, and yet so much has remained the same.

So, before we jump into this, for those of you who, for Whatever reason, in our space and beyond, don't know Rick Bellotta, who, you know, earned and have taken for himself many, many interesting titles, currently running by Resisting the AI Oligarchs, but formerly known as the Bullshit Detective or the Bullshit Police, and in full disclosure, one of our early backers at Tulip.

Welcome to the show, Rick, again. It's great to have you. 

Rick: Great to be back. 

Natan: And more seriously, for those who really don't know you, like, how did you get into this wonderful world? I want to say, I don't want to embarrass you, but is it coming up on four decades in industrial software? 

Rick: Maybe a little more than that.

Natan: Maybe a little more than that. 

Rick: Now, about 40 years in the industry, so, I mean, I was blessed and lucky enough that one of the first companies I worked at was very progressive. This is mid 80s now, we're talking about implementing, in a way, frontline worker capabilities. And we had mini computers and, you know, screens for the folks to guide them through their work.

It was pretty amazing. But anyway, I started out on the operations side of things. And then this company was Really aggressive and putting technology on the plant floor. I was managing the heat treating processes They were implementing all kinds of automation and basically what we would call MES today You know very rare and about that same time if you recall, that's when the PC explosion was starting Intrigued me the one thing led to another where I basically said hey this technology side of stuff is super interesting It's going to be a key enabler for manufacturing becoming better.

And so I got hooked and I started, uh, you know, systems integration. As you know, I worked at Wonderware, which is real pioneer, learned so much there. I think you've appreciated this over the 10 year journey at Tulip. It's one thing to create great products, but it's also just as important, the go to market machine, the messaging, marketing, all that, and I learned a lot about that there.

Natan: It's been a pretty interesting case study, you know, I think the first time we met, I was like showing you, hey, I have the one last copy of the Wonder Way, like the physical book, like, and, uh, I remember thinking about how much history you need to know when you're trying to build companies. 

Rick: Well, if I recall, I was your second choice.

You were trying to get Phil Huber. I know. 

Natan: But to Phil's credit, he introduced us really quickly. Of course. And can't thank him enough for this introduction. So Wonderware, Catalyzing Years, maybe you circle back on like what made that company so special a second, but how did you evolve after that? 

Rick: It's interesting.

I worked a bunch of different roles there. First of all was sales. So it was a great. Like, these are all things that kind of added to my personal toolbox later when I made the decision to start companies, understanding all the aspects of what makes a company successful or not. So, being out there, you know, having a quota, having to sell stuff.

You need to deliver. Right? Like, I know as a technical founder, we often think that like the winning moment is when you've done a presentation, the customers are all nodding their heads, and they're mentally committed to buying your software. You quickly realize if you do a real stint in sales, that's when the hard work begins, right?

Now you have to get a purchase order and negotiate, you know, all the fun stuff. Unfun stuff, but that's what sales professionals and so on do. 

Natan: But I'd add to that specific about operations, because like, these are people who live in the real world and they're like, we're going to implement this thing. And what does that mean?

And it gets more fun when you're talking about cloud and SaaS and all that good stuff. 

Rick: And we didn't have, obviously we didn't have to deal with that dynamic, you know, but I think the other big difference then was maybe the other takeaway for me, and again, it's just luck sometimes, I happen to be involved with a company that had two things going on, almost a cult like following.

And I definitely see aspects of that with Tulip. And I say that in a very positive way. Passionate customers, sort of a, you know, this is something new, interesting, exciting. You would have people at trade shows stacked up, you know, 10 people deep to watch the presentation. Demos, 

Natan: yeah. 

Rick: Yeah. It taught me that, you know, That ease of use and capability, functionality need not be inversely proportional, right?

And that's, I think, what Underwear did extremely well. Again, I see a lot of that manifest itself in the Tulip platform and the importance of marketing. I mean, these advertising back then was like sidebars and magazines. And they had underwear, you know, negligees and boxer shorts saying, this is not wonderwear.

Well, it created brand recognition and then they were the first, uh, first public company in our space. 

Natan: 1987, if I recall. It's always so mesmerizing to me with the ginormous space that spans Operation Globally, you know, its portion of global GDP, et cetera, that in fact, there are very few, you know, Public companies in the space that are standing on their own while they're super large in the mirror universes, like, you know, the RPA companies, the ServiceNows and all those.

And they're large, successful companies supporting customers next to, you know, the galaxy of SAP or the galaxy of Siemens or the galaxy of whomever. Why is that happening in our marketplace? Like, you know, we're not seeing like, Secular public companies, you know, put on my VC speak filter, you know, pure plays, as they love to say, in Silicon Valley.

Rick: You could certainly say that between like the late 80s, early 90s to now, the investment thesis has changed dramatically. I mean, the whole internet craziness, dot com bubble. I think the investor expectations switched from being perfectly happy helping build a successful company, and successful at that time was measured by revenues and profitability.

I mean, you would be punished if you weren't profitable. That's changed, right? For a while it was growth at any cost. 

Natan: Until 2021. And 

Rick: you know, in the swing for the fences mentality, I want 100x return. What's wrong with a five x or 10 x return? And then, you know, the last piece is it's, let's base it. It's easier.

The technology enablers we have makes it easier for new entrants. So markets that are attractive get fragmented very quickly. They're just look at a, look, look at AI right now. 

Natan: Well, we'll get there. Let's finish your journey and come back to the, to the Tulip story because I think, you know, we're starting to gear up for the decade for Tulip celebrations.

And, um, you're kind of the first one to start the party, I guess. 

Rick: Yeah, I mean, that's so exciting, man. And credit to the team. A lot of companies don't make it that long. And not only have you made it, you're doing all the right things. You're on the growth path. 

Natan: Yep. Lots of work ahead, though. I have to say, but yeah, absolutely team play here.

Rick: Yeah. So other stuff, then I kind of went back and forth, did some more integration stuff, came back as a product manager, got involved in some M& A activities. So we were acquisitive, learned a lot about that process, decided a couple of years later that there's this big gap between all, we had this, all these fragmented plan for systems that people were deploying, you know, homegrown MES.

Skata, HMI, Historians, Quality, Maintenance. I'm like, why isn't someone tying them together? That's kind of what led to Lighthammer. This was like 97, 98. That one really grew well. You know, we're up to about 50 people. Had no plans to be acquired, but we're starting to become a little bit of a pain in SAP's butt stealing some revenues from them.

Get some good visibility there and they came and basically said, Hey, we'd like to buy the company. So spent some time there, which is also great. Now I'm seeing the business backend big business processes, ERP, the way they go to market, how people buy, those kinds of solutions at the executive level. That was just a fantastic experience.

Got to work in some interesting research projects there. You could call it Precursor to the Internet of Things and all that stuff led to, to start. Yeah, 

Natan: potentially all of the Precursor to UNS, like just a million years ago, but, you know, event driven architectures, PubSub, all the good stuff. I remember you shared, was it publicly or you just showed me like some deck from a million years ago on Lighthammer?

Rick: And it was probably still relevant, right? Yeah. 

Natan: I remember that. That was an awesome deck. 

Rick: Might have been the one I did for Tim O'Reilly way back at like a Web 2 conference. Like tried to explain him and then I actually, funny, I still have the original big sheet of paper when we started ThingWorx, which was like, here's what the requirements need to be.

Natan: Yeah. 

Rick: That pretty much describes what the product looked like. 10 years later. So it was, it was interesting, but that was more of around that, that shift from industrial IoT to IoT. That was kind of the dynamic happening then. And we tried to serve both masters with the same product. And that 

Natan: became 

Rick: ThingWorx?

ThingWorx, which, uh, again, growing well. And then a big co with three letters came in and said, Hey, we'd love to make you part of our solution. Super visionary CEO, Jim Appleman, the guy just one of the most interesting people I've worked with. Didn't always agree with each other. And He's made it actually a lot of 

Natan: part of the deal.

Rick: He saw that kind of like IOT plus, you know, the back office processes plus service. And he just had an amazing vision for that. So that was fun. Took a few years off, restored this old house, came back off the bench to do a year and a half on the IOT team at Microsoft. But most of my time now is just mentoring, advising, you know, technology research and riding my bike.

Natan: That's good. So maybe before we dive back, there's like two kind of mini topics I want to try and cover that I think are pretty interesting. So, have you seen the reindustrialized conference? 

Rick: No. 

Natan: You should really check it out. The gist of it was like a bunch of people, about 700, came together in Detroit to kind of talk about, hey, we need the U.

S. renaissance, we need it now. You know, the urgency, not just from the economic perspective, but, you know, the geopolitics globally are big. Less fun. Kind of the gist is like, we're not ready for the next big crisis that comes, whether it's a, God forbid, a global war or another pandemic with its implication on supply chain.

So, you know, you're in the heartland, literally, right? Yeah. You never left, actually. What's your take on that today? What does the U. S. need to do to get back on path? 

Rick: Obviously it's a passion for me too, because I grew up in the steel industry. So I got to see. A lot of great jobs disappear. And some of it was the fault of the domestic steel industry, for sure.

Natan: Yeah. 

Rick: It left a impression on me that still affects me greatly to this day. So some of the things I'm trying to help with, I'm involved with the smart manufacturing executive council through SESME. I volunteered for some other kind of agency type stuff, and they're making progress, they have good initiatives underway, but you know me, I'm the poster child for ADHD before it was cool, so.

Natan: I would also say you're a natural diplomat with your communication skills. No, you never upset anybody, for example, with hard opinions on anything. So that helps. 

Rick: Anyway, but for me, you actually hit on a few things that I think we forget about. It's a national security imperative. I mean, God forbid we need to make boats and planes and things like that again.

They're good jobs. They're fundamentally good jobs. And I think those jobs have morphed dramatically from More physical labor to a blend of physical and intellectual power. Nothing pisses me off more than hearing people refer to people in the office as knowledge workers. 

Natan: Yeah, that means other people are not knowledge workers.

Rick: Exactly right. It just torques me to no end and there's so much we can do and we are doing with companies like Tulip and you're offering. Let's empower these folks. If anybody doesn't do jumbo walks or doesn't like actually spend time down with the people on the floor is missing out on so much learning and knowledge capture.

Anyway, we're obviously all painfully aware of the challenges with hiring people into manufacturing. And I think that's half the work itself, or maybe less than half. It's also just the human things. Respect, fair wages, flexible work schedules. One of my questions Technology that allows people to be more multi skilled or more flexible in what they do.

Can that help with, you know, more flexible work schedules for folks? So we can attract a different breed. 

Natan: I don't think we'll get to this in this episode, but like we'll point people to the episode we recorded with, uh, Liz Reynolds, that we cover some of that in the links. But you know what's interesting when you were talking about this, and that will take us to the second micro intervention here.

I'm kind of thinking about your generation, and we're a little bit apart. You know, on one hand, if you think about it as a chart, like, the computing industry is exploding, 1980s and on, on one side of the equation. But all the jobs are moving, and we're turning into, oh, let's believe the service economy, and, you know, now the bill is served.

You know, and the whole generation actually like contributed so much to modernizing manufacturing, you know, adding the compute, building the right architectures, all that kind of stuff, you know, you kind of turn back and it's like, Hey, where are all the factories? Where'd they go? Like, what are we doing about that?

Rick: I know there's all the buzz about bringing back semiconductor fabs and yeah, makes great sense, right? But all the other. You know, it's not bad necessarily that some commodity industries just kind of found their natural homes. Right? Sure. There's something that just, it just makes sense. But another related thing, you see all these charts that show U.

S. manufacturing productivity or even global manufacturing productivity, and they show some chart. If you notice what the divisor is, is labor hours. What the heck does that even mean anymore? Is it a 7 an hour person? Is it a 200 an hour PhD? I don't know. Tell 

Natan: me what value stream they're running and we can calculate that.

Rick: Exactly. So my point is that to make that generic doesn't make a ton of sense. And are we evolving to the point we're going to have higher paid, but more empowered people augment them with all, I mean, you know, the answer to this question and with automation and information technology. I think that's how we have to go to offset the labor.

Natan: So that brings us to the main dish of today. We're going to try and cover in the time we have left three big topics. Check it into Frontline Operation. Are we in the era of AI and what does it mean? And what does that mean for architectures? So let's check in on this term, frontline operations. Like, I remember all sorts of conversations we had.

I can't count how many lovely mutual rants we had, like, what the hell MES means, and we're still trying to figure it out. Like, would people get the no MES nomenclature that we've been pushing or augment MES, but it was like this fundamental belief that there's something missing. So, what's your bullshit police take on frontline operation?

What is frontline operations for you? 

Rick: Back up a little bit too, because you can't look at that without the M E. You know, even the Tulip stories come back to, let's not fight the term, let's say how we're doing it a hundred times better, right? With composable MES. Totally. Even back in the day, the largest individual providers were maybe 20, 25 million in revenue, 30 million in revenue, and in the dawn, the analysts said it's going to be a 2 billion industry.

And on aggregate, it probably was, and it's probably even greater than that today. But big observation was that 80 to 85 percent of the spend, I mean, I saw this front line because I was in the integration business, was on custom solutions. Have to ask them, what's the why, right? Right. Well, there's at least 14 different sub segments, like what we do in aerospace and defense is totally different than what we do in Medicare.

Medicare. And yeah, exactly. So you start breaking it up, it becomes a bunch of little vertical solutions. But the play was a platform play. So I, you know, I spent a lot of time doing that. We didn't get to the point of composable NDS like Tulip did. We focused more on a higher level set of tools that people could use to build these kinds of applications.

SPC and quality stuff, you know, connectivity to plant floor, flexible user interfaces, integrating stuff from ERP and other systems. So, it was really about allowing the engineers that were building these custom solutions. Now, I think we're starting to see this slow, but meaningful shift to empowering the, eventually the end users to do this themselves, right?

The other thing, when I started, big Fortune 500 to 1000 manufacturers had engineers in the plants. Yeah. They had a big corporate OT, you know, engineering, MES, Those don't exist anymore. You know, there's a few. So empowering other people to do this and innovate. Second thing is these systems aren't static.

They have to be kind of like, hey, we're bringing a new product online, or we have this quality issue or whatever. These systems need to be super adaptable. The old systems of old are more like basically shrunk down little ERPs. Rigid, pain in the ass, big bang implementation. So we're addressing a lot of those issues.

And then putting the people first, right? Thinking about what kind of tools do they need to do their job most effectively. And that's different if it's a frontline worker, a supervisor, or a planner, or whatever. 

Natan: Yeah. I always think about, you know, you mentioned before something that we say a lot, paraphrasing, you know, the knowledge worker and the non knowledge workers.

And if you do away with that, basically like a constituency without a platform today, and the specific one I'm talking about is the people that spend their time in the operational environment, be it like assembly floor, machine shop, lab, remanufacturing hub, warehouses, whatever it is, they're kind of orphans, technological orphans.

And what I've seen the past decade is that Companies start to really care and they don't necessarily care about this vendor, another vendor, you know, in the Tulip context, they actually care about, well, if we really want to change how we work, then we better harness those full knowledge workers that we have there and give them tools.

And like, that's the hardest thing. And, you know, we've gotten to a scale where sometimes we meet customers and like part of the way we have a discussion with them, it's a little bit like corporate therapy, you know, are you sure you're ready for this change? Maybe you should take time to contemplate it.

You know, talk to your partners, see how you're feeling about it. There's a lot of emotions and it's been very focusing for us because, you know, startups, we need to commit suicide on every deal, but you get to a point where you really want those who are ready. Cause that's like 50 percent or more of the success.

I mean, the platform, we got the engineering down, still a lot of work to do, but you know what I mean? It's like the change management is going to be with us, I think forever. 

Rick: And, you know, the way you implement, right, it's like if it's just a system enforcing some corporate policy, 

Natan: how are they 

Rick: going to take ownership of that, right?

If it's something to help me do my job better or ideally be more productive and share in the economic benefits of that increased productivity. Yeah, we did stuff like that even in the steel industry. We had incentive, performance based incentives on a shift basis and to the point the best that we could allow the operators to actually see how they were doing that shift, right?

So you could drive behavior in near real time. 

Natan: And you did that without AI? That's what you're telling me? 

Rick: No AI required. Just humans talking 

Natan: about what's 

Rick: important. Math and humans. The other element, you know, you look at how a lot of us are applying technology solutions, like when we build applications, we tend to codify the no, right?

Meaning we're going to enforce this certain set of processes and procedures. You're spending up time on the plant floor. You realize that 70 percent of your time is firefighting, materials didn't show up, machines down, people didn't show up, customer changed their spec, something's going on. So, the shift from codified applications combined with tools that let people deal with those issues.

Ad hoc, situational issues, give them power room with information, analytics. 

Natan: Yeah, tools to solve problems. And that's also a good shift to the next phase of this. You know, I hear this term, this is working out of the box with relation to MES. And I'm like, shut up. Yeah, right. At best, you're at, this is ready to be customized.

And you know, in the MES space, you know, we've taken our gloves off a long time ago. No need to explain it to you. You choose an MES. Basically, what you're choosing is whose development environment you're going to work with, with what constraints and decisions they've made on things like data model, how you integrate to me, what UI I'm allowing you or not allowing you to do, what's the constraints of my, uh, Typically on prem, even though lifted and shifted the cloud architecture is.

And after you took on all those constraints, that's when you'd start to programming within the boundaries of said out of the box features, because it doesn't do what you need it to do. So the term I was starting to, Push on the industry and also like kind of being pretty religious when we're talking about composable MES.

Stop saying that. At best, you say, you know, ready to be customized, ready to be applied to the customer's problem. And you know, there with all my bias, I think composability holds a huge promise. But that's the reality of MES. That's the conundrum of MES. 

Rick: And the inverse was true, unfortunately, that it's not that they were ready to be configured.

Some of them were If you wanted to modify it, it was like modifying old ERP systems. It's going to cost you so much and you're going to have to maintain it. The other epiphany we had a number of years ago, and I credit, you know, Russ Fidel, my co founder at Lighthammer for this idea, was that apps have different life cycles.

There's nothing wrong with a disposable app. Totally. If we have a situation and we want to apply technology for ramp up production quickly or whatever, go for it. Traditional approaches didn't let us do that. Tools like 

Natan: Tulip 

Rick: do. 

Natan: So now we come to the next era, right? You know, we're wrapped half of uh, 2024 and I feel like the Gen AI hype is On one end dying down, which is nice, it's like less noisy, but you know, I think people start using it very regularly and I'm going to just talk about myself, you know, the different search engines and whatnot.

And it's like a weird use case. It's like sometimes extremely useful and sometimes it's just a way to pass time. So it's like a combination of Google and YouTube sometime. You know, that's how I feel. You're like, either like surfing it, like doing all sorts of weird shit. Like we had a whole, we're going to talk to each other in Japanese haiku.

Remember that? We would not be able to do that. I made 

Rick: up all myself. There was no AI involved in the production of my haikus. 

Natan: I had only AI involved, but it's about the prompting that matters, because like, I didn't get the result randomly, so I'm honest about that. But how is that going to affect in the real world?

Let's talk about the next couple years. 

Rick: Well, I mean, to me, the biggest mistake, uh, mistakes might not be the right word. The counterproductive thing going on is galvanizing AI under gen AI. It's the opposite. That's one type of AI. In our world, we need the collective set of AI capabilities, right?

Generative AI and AI as assistance and AI for augmented search. There's a huge value to those as long as you can train it on the right corpus of knowledge, which is not trivial. Then we've got what I always call meta sensing, right? That's where you're going to take AI and machine learning to take a video or a camera or a vibration sensor or something.

And turn it into some other insight, right? I'm using a camera, but as a quality or leveled attention, thermal analysis of how a part got heated. So you're using it in a different way. You're interested in, is the machine going to fail? Not what's the vibration. The analytics are already shifting from just, AI was used for Eureka kind of stuff, where data science would take a big data set, look for correlations, look for patterns, create models and say, Oh, When we run the machine 10 percent too fast, our scrap goes up 30 percent or when we're sourcing materials from this place and the weather's humid, we have lousy quality, whatever the case might be.

We've now shifted that from insights to basically prescriptive kind of stuff. I know a number of companies that are now actually closing the loop, right? They're changing settings to achieve some out, quality, throughput, energy, whatever that might be. All 

Natan: the way back to the control. 

Rick: Exactly. You know, it goes through this continuum, right?

Human in the loop. I have a chart I used to do at this lecture I would do every once in a while for a friend of mine. And it's like, human, machine, sense, decide, act. And it goes through a continuum and then full autonomy is obviously the machines, all three of those. With Gen AI, I think we're clearly still in the human in the loop, in most cases, or the decide and act 

Natan: part.

Well, taking the human in the loop is pretty terrifying if you have a full control loop, no? Because it's not like, hey, Gen AI hallucinates, type of statement. It's not just that. Even fuzzy algorithms can make mistakes. You can have security issues. You can have, you know, if you make the human go away, risks of catastrophic failures increases, no?

No? 

Rick: You should read this. I'll send you a link to this, this book. Uh, Ken Sanderson, a company called Compose, I've spoken with him, but he talked a lot about machine teaching, right? As opposed to machine learning. And it's actually more than a semantic difference. So the other flip side of not a lot of talent coming into manufacturing is a lot of talent's leaving, 

Natan: right?

Rick: How do we capture that kind of tribal knowledge on how things really work? It goes back to this. If it's 70 percent exceptions. How are we learning about that? How are we learning how to respond to those? 

Natan: Yeah. Well, we just need to sleep in the factory like Elon. No, that's the solution. You can Google that one.

That's easy. I can tell you that one. 

Rick: I just spend a little time there, wouldn't hurt. I mean, you know, you know my, we want things like OEE, right? It's helpful, but misapplied. 

Natan: Yeah. 

Rick: Literally spending two weeks or a week on the floor talking to the operators, you'd probably get all the answers to why you have quality and downtime issues.

Natan: Yeah. So it's not just JetAI and it's coming back in a way full circle. You're saying down to the control level, which typically is You know, people kind of talk about edge, you know, doing stuff at the edge, and that's a good segue to our last topic. My take, you know, we've been building and experimenting with AI and use cases with Frontline Copilot, and it will appear all across the stack.

You know, we've seen examples where you generate the line logic, and we're seeing the same, you know, people using our stuff to start experimenting with, you know, make the G code. You know, things that are very structural and make a lot of sense and, you know, gee, wouldn't it be nice if it kind of can live right next to the machine or integrated into the HMI, right, which Tulip is also playing from an OEM perspective, but then one of the biggest thing, and I still believe there's a huge human in the loop, not just because AI and machine learning, um, are fundamentally imperfect.

It is always a factor of like how much trust do you have? And even if you have like a classic machine learning algorithm, it can go out of tune, you know, ambient condition can change, parameters can change in like suddenly the model is just 95 percent and not 99. And guess what, that could generate like a huge problem.

So, Cool. I think the key is how do you stay close to the customer such that they can use this stuff for real. And like the concrete examples that we have is like, we build copilots into widgets. So you decide where to put it in the no code, low code flow. You take it and build it into the analytics. And so now all the contextual data, you don't have to ship it.

You can just ask your factory how it's doing. And I think what's interesting is to see How it changes how people, quote unquote, catch up, because we just talked about the knowledge worker, non knowledge worker sense, and they catch up and suddenly they're not left behind yet again. Because I think they were left behind from web 1.

big data, blah, blah, blah, you know. Let's not let them stay behind again. 

Rick: What people believe, as I do, that software is just encapsulated best practices and capabilities. Like, the more we as platform vendors bake in all the, you know, whatever the cool new meme of the day or technology, that should be transparent, right?

It should be like the functionalities available for people who are composed using Tulip, but they don't need to know that it's a graph database and a SLM using retrieval augmented, you know, who cares, right? They want a Solution. Exactly right. But still configurable. 

Natan: So what does all that has to do With the future architectures.

People put in architectures, they may think they're putting them for 10 years, but like, I think effective life of architecture is five years to first augmentation, and then it like goes from there, you know, that's where we are right now. 

Rick: And that kind of varies by like, you know, when something gets institutionalized in an OEM machine, historically, that might be there a really long time.

Like it just rarely gets touched. And that's actually, that's another big thing that you saw so much emphasis, you know, like applying AI and advanced analytics. To assets like the machines themselves. The real value, or maybe the bulk of the value is when we start doing it to processes, right? 

Natan: Which is between people and machines.

Rick: Exactly. 

Natan: And when I say machines, I also mean platforms. They're pretty complicated machines. 

Rick: Totally agree. I mean, both of the companies that I was involved in, we always talked about synthesis of people, systems, and the physical world, right? Though machines are just one manifestation of that. But my point is that.

There's still a lot to be done, like a number of the companies that do industrial AI that I work with. Sometimes you bring in data streams you didn't necessarily think were correlated. Obvious examples are things like ambient weather. Well, turns out that the way the cookies bake is affected by the humidity of the outside air that day or, you know, whatever.

So you need ubiquitous access to data, right? The cloud helps us with all of the, you know, the massive compute we need to train models. But increasingly, it's being smart about where you partition things to execute at the edge, you know, close control loops, high availability. If you're truly opening and closing valves and changing settings, It's probably a good idea to be, you know, somewhat autonomous in the way you can do that.

Natan: Autonomous, simulated, have like a AI buddy or whatever we call that, that watches over you. 

Rick: Perfect segue that the human in the loop may shift to something in the loop. Something in the loop. Right? And so real time simulation, physics, you know, first principles type models. Yep. If you look at what some of the, the Alan Kuhn and some of these other people have been talking about recently.

It's like we learn a lot when we apply algorithms to validate and measure the goodness of the output from these models. And then once we understand that, we can kind of like, you know, it's going to be this virtuous loop of you're going to train the simulations with AI output, and you're going to train the AI with simulated.

It's kind of wild. 

Natan: So Rick, at the risk of creating yet another trend in buzzwords that then we'll both have to hate later on LinkedIn, you can pick your flavor or blueprint of cloud architecture. And we've shown at Microsoft, AWS, all the usual suspects to show edge cloud, you know, how those interact. And how that lives with the platform.

You can start implementing MES, CMMS, all sorts of like three, four letter acronyms. But I think that the next gen architectures for real, for automation, they have to be action driven. So they have to be something like, if you're really serious about empowering the people who do the work and all that kind of stuff and how that in turn collectively becomes the change of how the company works, then it's all about like how quickly you can respond to something that's happening in real time.

Breach. And I think that type of architecture is called action driven architecture or action based architectures. And there, there's a lot to continue and build. And you know when we're going to talk about that? 

Rick: Operations Calling. 

Natan: In Operations Calling, which is our event that's coming up October 8 and 9, 2024, here at Assembly Square in Somerville, which is Boston, for those of you who don't know.

We're going to continue this conversation and see if we can come up with some more concrete examples of how to think, create, design those action based architecture. Yeah, I'm looking forward, man. Thanks for joining today. It's been awesome. 

Rick: I'm sure we'll talk before then, but I'm looking forward to seeing you up there and uh, I think we'll see a lot of customer insights there and that's the kind of content that I think is immensely powerful.

Natan: That's the main feedback we got last year that it was very much about community conversation. We learned a ton. That's what we're doing it two days this year and I'm really excited. So really looking forward. 

Announcer: Thanks Rick. 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 slash 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.