​

Natan: This is super cool 'cause like you can bring all these operations people to Assembly row to a main stage that is in a Ford factory. We couldn't ask for more from the gods of operations symbolism to bless us with. 

Narrator: 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. 

Natan: Hey Maddie. Hey, Natan. Can you believe it is the first episode of season six of Augmented ops? 

Madilynn: No. But happy birthday, Augmented ops, 

Natan: happy birthday, and a new season. And it's also what, 48 hours? Not even after Ops calling number three, just ended ops 

Madilynn: calling three.

Yeah, 

Natan: it was a great event. We had a lot of fun. 

Madilynn: Yeah. This event has been like. My end, so many of the Tulip team members around the world, really like our, like life and main project over the summer. We like shift in, in June to like ops calling land where it just steady grows and grows.

And I think being on site and having everyone, come to our HQ and come to Assembly Row and hang out with us and share their perspective. I think it's just an incredible experience and there's always so much that we learn and all these new people that you get to meet and it's a great experience.

Natan: Yeah. So for folks tuning in who do not know anything about operations calling, let me give you the 1 0 1, like in 90 seconds. I'll try. So we're a community driven. Open ecosystem company who have been forever. And we actually, the genesis of this was a, an event we've done pre COVID. Yes. That was called Agile Manufacturing.

And that was to e emphasize the fact that we wanted industry to transform and adopt agile tools and stuff like that. And fast forward to three years ago and we grew and. It's simple. We wanted a music festival, so you know, who doesn't like music festivals? So in Boston there's a really famous one called Boston Calling.

So three years ago we set it up as a homage to the city that we're living and working in Boston. And we called it operations calling 'cause there's also a calling, right? What is the calling, Maddie? What is the calling? 

Madilynn: Our calling at least is to help support operations and Manufacturing.

And we wanted this festival it's 'cause it is a festival. We have multiple stages like. Yeah. One of the things about Boston calling that I think is really cool is they always have a local stage where it's like local musicians and then they bring in a bunch of cool local restaurants, like Tasty Burger and Roxies and that's the food.

And so we had that vibe too, like maybe you don't always know when you're there, but like all the food and assembly, like we had, legal Seafoods and the Smoke Shop and it was like, yeah, it was Assembly Row local, but also made to be fun and. Engineer, nerdy for operations, like to celebrate operations.

Natan: Yeah. And for folks who don't know, our HQ is located in this really hip neighborhood called Assembly Row that if you would visited a, even a few years ago, a decade ago, it would be something completely different. So now it's really transformed to, cool shoppings and condos and office spaces and so on.

And we live in an old Ford factory. That became a mall that we are now reviving to become a center of Manufacturing and operation tech. And we live there with our good partner and friends at DM G Maori. Yeah. And very lucky to have them and many other partners join the event. But it's poetic that we live in assembly row and when you walk around assembly row, you see all those cranes and weird artifacts from an era where the Ford supply chain was around.

This factory. So then we bring a bunch of actually we brought what was it like around 800 people joined us in the event this time. I 

Madilynn: think the exact number was 7 92. So Okay. Basically 800. 

Natan: Basically 800. No, but we like our data. Yeah. To be accurate. But this is super cool 'cause like you can bring all these operations people to.

Assembly row to a main stage that is in a Ford factory. So we couldn't ask for more from the gods of operations symbolism to bless us with. So it was great and I think we had a lot. To say and share. Yeah. And we're gonna put it all out there, but what did we show? What was the main thing from your perspective?

Madilynn: Maybe before I hop into it, I think we had 20 customers speak at Yeah. Ops calling. So like the majority of content was from. Companies and their operations teams. Yeah. And I don't wanna speak for those folks because they spoke for themselves and we're gonna put that online. Totally. And it's totally, it's compelling.

But from the Tulip standpoint, the theme this year was AI for operations and we had some pretty big launches on the floor. So we announced the availability of our agent builder. So people are talking a lot about agents and if you ask the five. What does this really do?

Questions. It's really just here is a pre-configured module that does a deterministic task that you can't change and you get to pay us extra for in a monolithic software package. At least that's what I've seen across most of the industry. And what we're bringing is the same ideas of composability as well as like configurability starting points to tulip's platform.

So build agents and our agent builder. Or download agents. I think we're gonna have eight available in the library and configure and change those agents and to just like further prove that, this isn't some made up functionality. We had an agent. AI Builders Challenge at Tulip. So we had 15 of our, end users come in day zero.

They spent all day with the Tulip team and each other. They were in teams and they built their own agents. Yeah. Using the agent builder. And then on day two they shared and talked about the ones that they came up with and their experience building them. So it was like, this isn't vaporware. We have people.

Doing it. And it's not even guided really. It's like there's some instruction, but they're working on this with other people so that's like that piece. And those will be ready for download. And the library people are getting access to the beta now. 

Natan: And the builder experience is a really good example to explain sort of the first principles of operations calling that have been refined over the three years we've been doing it.

But I think they're settling really nicely and. There are obviously great user conferences out there, and there are places where people learn a lot about products and meet the company they're working with and find a lot of users, but a lot of them are just complete propaganda shows that.

You get, I was calling me just like propaganda and a sad sandwich, more or less. Yeah. We were going for, we maxed out all the restaurants in Assembly Row we did, and the adjacent neighborhood. So I guess there was good food variety. Yes. But when people come and build and then.

It's not that we've done this before, or the product was like fully tested six months in, in the field. It was like, Hey, we're really doing this now. We need you to tell us what is this good for? That's a first principle, so it's kind of show. Don't tell. And then we extended that to teach and learn.

Yeah. Which is always bidirectional. So that, that is one. Yeah. Obviously the other one that has been with us since the first one, which really catalyzed the open ecosystem and the way we like to define it, because every platform has an ecosystem. Yeah. And a lot of time is, oh, we have an ecosystem so everybody can make more money.

Our ecosystem, it's really not about that. It's really about, we have an ecosystem because we cannot predict, define, dictate how customers are creating value from a solution that really fits what they need. And like it's not about do I like this barcode or that torque gun? It's about whatever works and if it's like.

Reducing the cost of system integration, complexity, time, and leaves the team who does that with the tools to do it. Customer wins, the ecosystem wins, the business follows. Yeah. So that is operation driven ecosystem. Agreed. And so there was like this rule, I don't think we ever written, the rule is always the debate of like unwritten rules are better than written rules.

I don't know. 'cause I don't know that we wrote all these rules, 

Madilynn: if it's an unwritten rule, it's a value, right? 

Natan: It's a value. Yeah. Or an operating principle. Yeah. Anyway, it was very simple. It's if you do not show up with a working demo, don't bother asking to join. There's no here's your 10 by 10 booth and bring your, whatever the, your screen and yeah, I don't show your videos.

Yeah, no slides, no videos show a working thing. So we had like robot, all sorts of robots running around. Yeah, it was great. Like cog Knight was there showing their industrial agent talking to Tulip agents and UR was there showing a MR action in a pharmaceutical environments, so many.

Madilynn: There were 41 partners. Yeah. Across the tech, the pavilion, and the hardware wall. And it would've been great to include more, but at a certain point, the office Yeah. Is only so big. Yeah. 

Natan: We gotta talk about our house party in a second. That's because we said we're gonna have a section talking about the vibe.

Madilynn: Yeah. 

Natan: But the other experience that, I think was really different and represent this teach and learn principle is that we wanted people to come. Yeah, to meet great people and have enough time to talk to each other and hear from customers with no filters and all that stuff, but we also wanted them to be able.

To teach themself and learn something new. And in this sort of AI wash era that we're all experiencing, and in industrial operation, I think it's even more pronounced. Yeah. Maybe we'll include the AI panel I did with Alex from AWS and Tom CTO Zebra. And it was about AI and operation, like in the wild, and there was a interaction with the crowd and at some point we asked who here is actually using ai not in a.

Proof of concept, not as the individual, but as an organization actively using it on a production line, more or less. I'm paraphrasing, but that was the question. I didn't expect a lot of hands going up, but I didn't expect zero. Oh, and this was like a full room at the row, so I wanna say it's 60 people sitting there or something like that.

Yeah. So not even one hand went up. So it was so telling, and this is why we wanted to show concrete use cases for AI learning and what we call the AI password. So maybe you want to tell what that is. 

Madilynn: The AI passport is like a new concept we brought this year. So we wanted to go through like. All the different ways that you could use AI in Tulip.

And you know this, like there's generative AI and agent ai, which are like two types of AI that we're talking about a lot, like all over the public consciousness. Also, like a lot of new use cases that we're introducing between the builder and also like existing use cases in Tulip with AI composer and translation features.

Prompt actions and automations, but we wanted to show people what that would look like in specific value oriented use cases. So we had a whole section. And then also, the rule Celebrate operations. Make it fun. If you went through all of the different, stations, you learned a lot about ai and then you also got like a literal.

Passport holder and a passport. Yeah. Cover vending machine. So 

Natan: that was connected to Tulip? 

Madilynn: Yeah. And I think the standout one in the passport was what we're calling like the factory playback. 

Natan: Yeah. Not. Different than our customers. Tulip is also going through its own internal AI transformation, and I don't think we're that different than many organizations, but in a very structured way we've started to implement internal tools that allow us to query all our enterprise data and we are incorporating into our marketing, which of course Maddie, you can share how we're doing stuff and, yeah.

In our r and d process and it helps us have faster cycles, deeper to a degree. 'cause like you can get a lot of research done and a lot of coding done, but it doesn't relieve from the human aspect. That kind of takes all that and figures out how to get it to a bar and quality that is.

Something that we would be proud to put out there, whatever is the word, product, coding, marketing process, whatever. But one thing for sure, it allows us to, to experiment faster. So one of those product experiences that we've been working on for a while. So Tulip was born with a computer vision stack.

At that time, it wasn't that important to say that it's ai. Because I guess computer vision was just enough, it was a computer vision pipeline that we're really proud, ran in the cloud, had a hybrid kind of approach, ran an edge, cameras off the shelf. You can define areas of interest.

It can detect. People and their actions. And it can support things like PIC light or handle models, whether internal or external from great partners like AWS or lending AI and do a quality check, things like that. So we had it forever and it's, yeah. Shockingly using this really old tech called Deep Learning that actually works and many other techs.

But, so we always had the ability to hook up a bunch of cameras to Tulip and if you think about Tulip completely differently as if it's basically a rapid development environment for process and for the, for the operation people, the folks on the shop floor to create a digital workflow, right?

Yeah. And if they do that. If you believe in rapid development, then it means you don't need just the development environment. We call those IDs. And so today, of course, it's cursor and the like, and you also need the profiler and a debugger. And so factory playback is a profiler and a debugger for operational environment.

So you can hook up cameras, on a timeline interface. You can scroll just like the way you would scroll when you're editing a video. And find what happened. So for example, for use cases like drop in production and you want to go and debug the process, you can literally go and see because the video on the time horizon is correlated with all the inputs that are coming from Tulip.

Whether it's, stuff coming from the database you integrated or. The raw data flowing through the UNS or the Tulip events that are defined in the apps or what have you. Yeah. But it also shows you what to do. So instead of, teams needing to go, this happened, let's do a retro and see what happened.

And then people need to imagine what happened. They literally can go back and play the tape and. Of course we're in early stages of this and none of this is like something we're shipping like tomorrow morning. But it's this building block approach to how to make AI interactions very composable and thinking about AI as yet another composable element of your production system that with time will become Yeah, not only more connected and capable of doing more than just say core composable MES functionalities.

'cause, the other principle we were talking about, of course, is composable X. Yeah. Where you now can span to use cases that are not just oh, I'm just in the MES box, or I'm in the QMS box, or I'm in the LIMS box, and so on. You can combine those things and that's where, composability and AI come together in the future of this playback.

It doesn't matter whether you're doing it in industrial, putting together discreet. In cutting a discreet process or if you're doing a hybrid discreet pharmaceutical process, everybody needs that. Yeah, everybody needs that. Everybody needs to be able to introspect what happened. And I think what's exciting is that especially for regulated industries like aerospace, defense, medical device, biopharma, as soon as you regulated.

It'll dramatically change future audits to a point where people can rely on proofing to audit teams, how they perform their process based on video and metadata correlated output. And that, yeah, that is what nerds like us are excited about. 

Madilynn: I think there's there's two things that you mentioned there that I think are like worth pushing.

Further, right? Like 

Natan: Yeah, 

Madilynn: We think about composability in two ways, right? We think about composable architecture and we think about composable solutions, and both of those things are integrated and supported. And I mean we, we did the launch provision, I wanna say like within my first six months at Tulip, like AI in different flavors has always been in the Tulip platform.

And it's, one of those things where customers that use Tulip can go from paper to ai. In the same kind of breadth that people are going from totally like paper to PDF on digital. And when we say ai, we're talking about all kinds of ai. We're talking about multimodal creation, right? From video to app, just like we're doing with PDF to app.

And when we say that we mean interactive workflow where you can further integrate other AI and connective capabilities, it's like how do we take the most powerful technology available? For manufacturers, and we integrate it together in a platform and we extend it with an ecosystem and we make composition and orchestration as simple as possible for the like, problem solvers on the front line.

And that's our mission, right? And all these launches we had, that's like a. Ops Moto, that's AI composer for video. That's agent builder. That's the, teaser we're doing for factory debugger. It's like we know these challenges that exist across operations. How do we help these companies in this like critical industry that we love, overcome these challenges?

And really give their people superpowers and I think like it was gratifying and a lot of fun to get to nerd out on the tech and also the implication Yeah. Of that at this conference. 

Natan: Yeah, I think it will be awesome. I'm pretty sure a lot of folks from the builders are gonna, were gonna share something on what they actually built at some point, because, we mentioned composable X and of course our library is like where that set of product functionality and content live.

But it's also a moment where the content is not just remaining in the content department content. In our ecosystem, in our library is becoming fast context. And what I mean by that? And I think all those terms like agentic and context are being thrown around. If they're just said without definition what they do, turn this podcast off immediately kind of situation.

So we don't want to be in that camp. So let me explain what this means in plain English, it means that, for example, if you. Want an agent to run on Tulip to help you with the validation of your GXP process, you would download context. It's wrapped up in this thing called Validation Agent. It shows up in your agent manager within Tulip.

It looks like the now quickly becoming. Extremely well-known, familiar interfaces that where you have a chat interface or an agent definition on one of the foundational model. And we love them all. OpenAI, Claude, and Tropic, and so on and so forth. And we are getting used to this interface.

So you get this interface. You define the agent, but you have a starting point because the hard part is like how to do prompt engineering. And so we gave people a great starting point that allowed them to interact with, the Tulip tables, our data models the connectors and so on. And of course, deeply integrated tour MCP.

That is getting richer and richer with access to more and more tools that help the agent do stuff. And then you can combine it within automations and workflow. So suddenly, like your agent can show up and say, okay, I'm here and I'm gonna do work within a, just a well-defined kind of workflow that you control.

So that's what I mean by context. Then you can own it, you can refine it and eventually curate it and combine it into a set of those superpowers that you're talking about. Maddie, they become. Encapsulated in those agents that people can be running along, and then one engineer can do the work of three or maybe five, and then we solve the labor productivity equation, which is what it's all about.

Madilynn: Ultimately. There's a lot of, hand wring concern, maybe concern trolling. I don't know if you're, if I'm allowed to say that, but there are a consistent amount of unfilled jobs in Manufacturing and especially as we like try to invest more in advanced Manufacturing. There's a lot of high skilled roles that folks haven't been able to fill.

And so being able to. Have one engineer, be able to accomplish more and also just make work generally across Manufacturing better. For everyone who works there is really critical to industry. And I think one of the things that we're seeing beyond like the individual solutioning is like a rethinking of solution.

Yeah. And maybe that's like our spreadsheet example and also like our challenge on composable X versus composable. MES connected worker. Insert your favorite, least favorite enemy. TLA here. Third letter acronym. 

Natan: We, we love all our enemies. We love them. The thing is that the current architecture has gotten to such a degree of deficiency that this sort of.

Gen AI Madness has brought to light somewhat like as a catalyst. It's not that they were not deficient before. 

Madilynn: Right. 

Natan: We came up with frontline operation when, I'm trying to, 

Madilynn: in 2021, we're also 21. We're working from home, so we're like, we're making a diagram maybe in a Miro board that's like frontline operations and then like the list of all the TLAs, it's what if there was a better way and it was just integrated solution.

Natan: Yeah. Composable. Even though we live it and we understand it technically, and we understand it operationally, we always talked about those two. Yeah. Aspects of the term. It's still pretty abstract. Yeah. For a lot of folks it's like what the hell you mean? Com compo what?

So where you say look. You don't need to buy like A-Q-M-S-A-C-M-S-C-M-M-S MES, lims, LES, and then decide which portion of, what part of that are you gonna implement, where and which department owns, and then spend the time, money to integrate it. Those days are gone. And anyone who's using that architecture, and by the way, that architecture is really bad to do anything.

AI agent. Yep. Anything. It's actually, it's the data. The data is not, it's like not good, bad Arne bad architecture. Yeah bad architecture. When you saw the stories about scale, this was actually a really interesting phenomenon, because I saw in this event, like really mature customers, like DMG and Stanley back and Decker, and they have dozen, dozens of sites each and many thousands of stations. They're like, let's go. We have been doing this, of course we're putting AI in, no problem. And they're like, great, we have the data, we have this. And the new customers are like, how do we get faster their situation so we can actually do those for them.

The starting point is with ai. Almost a day zero. Yeah. And so they wanna go faster and demanding the speed so they can get to that state of, it was like super interesting to see. And I dunno, we'll probably include some of those links below, but yeah, 

Madilynn: for sure. 

Natan: But one cannot just simply turn on the AI and it'll work.

Madilynn: The AI has a lot of tech that it, and also context data that it, it needs in order to really support the scale. Maybe some closing thoughts for us. Obviously we can sit here and talk about operations calling for hours. I think one of the things that I'm feeling most energized about, and it's mostly I think, ops calling's a great catalyst for it, but it's really just all the time spent in community and like getting to talk to people and hear from people.

It seems like there are so many. Smart and capable engineers and operations leaders and, write tools and focus on the problem and the productivity crisis or the way that we re industrialize Manufacturing like we can do it. And it felt like more of a, an energy and a commitment to doing.

That I think was really hopeful for me and left me feeling excited about being in industry. I don't know how you felt and Natan, maybe that's a little too dreamy from me.

Natan: There's like stuff I keep saying to the team and it's usually the stuff I hear in my head all the time.

So it's not like special. It's like one of them, you can't fake product market fit. So I keep hearing this in my head all the time, because that's stuff that people who build companies really care about, right? We really wanna build a product that makes difference and valuable and solves a real problem and all that good stuff, and.

I think the enthusiasm from customers and having them wanting to basically become design partners on this journey with us because we're not pretending we figured it out, yeah. We've learned a lot and there's enough to put out there and tools to put in people's hand and all that kinda stuff, but the real figuring it out is coming and it's coming through.

Use. There was this example of people built, one agent that does warranty assessments and testing out like how you should handle different condition around warranty. And, the workshop was three hours. They refined it and looked it again and all that kind of stuff, and they gave us like preliminary numbers, like half a million dollar a month potential.

OI on something like a single agent. 'cause humans are not good at catching and correlating tons of data. And the reason they know, and it's so easy to calculate is like when you are paying the warranty, there's some ledger that records that. So I just think this is like so powerful and this is like one agent, one engineer, three, four hours going fast.

So what happens if 10 engineers do this for a week. So all this like between really important deep conversation and a lot of noise in agent space on digital work. And what does it do? I think it boils down to tools at the hands of the people doing the work. And just trust them. Just trust them.

Madilynn: Yeah. This is a tangible example and it's also that's tedious work and there's like an opportunity cost of having to have, people do that work. And now. Those folks can do other things. So there's like a lot of a value in like kind of the shift around where people spend their time that's unlocked by, this like digital workforce.

So I'm excited. 

Natan: Me too. I'm excited and a bit tired, but I guess like at the Friday of we, we, it's allowed, we can be. Exhausted a little bit. 

Madilynn: Exactly. And we're already like having a lot of people ask us like, what are we gonna do for operations calling 2026? So we you 

Natan: know, we have to ask the community because there's like this, maybe this is a good cliffhanger to end this episode on, and then we put a survey out.

What do you say? I think what we supportive here's the situation. Okay. We oversold, like we, we had to put the weight, like we had to stop. The registration 'cause yeah, that's it. We could not accommodate people, buy more, people, buy more food, 

Madilynn: put up another tent. 

Natan: But this is back to our house party.

Like we already talked about the vibe, yeah. It was a music festival, but like a house party. Yeah. I don't know. We find a bunch of people to our home. Hope they don't break shit. Yeah. Put a dj, not get caught by the parents. Lots of open 

Madilynn: bars. Yeah. 

Natan: Lots of open bars and it's just much fun and cool. We like the house party. Yeah. But the problem is like we need to either get a bigger house or we need to figure out how to change it. So what are the options? The option is like we go rent an industrial space somewhere near us that is not a venue, but is like a factory in the making or an old factory that is now sitting idle.

So we can have more room, but then we have to rebuild. But it's not our hou, but it's like the proxy for our house. Or we have to keep it small and have like multiple smaller events. I don't know. Yeah. 

Madilynn: Or just I guess cap it and then fewer people get to come, which we really don't wanna do.

That's let us know. We're let us know we're gonna, we gonna start planning, we're gonna put 

Natan: some survey and we're gonna start planning. And it's maybe also time to thank everyone that like, not only on the team, that kind of worked so hard to make it happen. Yeah. And what an unbelievable performance the Tulip team has shown.

It's just so fucking great, but also to everyone who came, the partners and the customers and the conversations. So yeah, just thank you simply yeah, we're happy and appreciative. 

Madilynn: Totally. You can have a perfect party and if the guests don't make it fun, then like it's a shitty party. Yeah. So it's a fun party because of this community, and we're grateful.

Natan: Yeah. 

Madilynn: Yeah. And I hope we can see everyone next year. Thanks. Thanks for having me on Natan. 

Natan: Yeah. Thanks Maddie, and thanks everybody and see you on this great new season of Augmented Ops Season six. Here we go. 

Narrator: 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 dot 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.