Inside Nominal, the Intelligence Platform for Hardware Engineering | Lightwork

0:00Bryce, welcome to Lightwork. Thank you so much. It's a pleasure to be here. Happy Series C Day. Thank you. And today you guys are launching something really big. We sure are. We are in a really fun place where Nominal is a company that gets to serve a discipline that's just been really forgotten about in Silicon Valley. You know, if you walk around any place that is not San Francisco and New York and ask, what do you do? And they say engineer. What they mean in many other places is a mechanical, electrical, aerospace, chemical engineer. or someone that is devoted to making bridges and getting water to your house, or building airplanes, or rockets, planes, trains, cars.

0:36We get the gift of serving them. And now we get the gift of helping them adopt AI. We've been in a really fun place where we've been building this company for four years, but a lot of it just keeps coming back to, how do we help these particular people get frontier technology? And it used to look very different. It was, how do we bring them best in class? Data infrastructure and controls infrastructure. to really just like anything that's happening more broadly in the world of software and figure out how do you make it like purpose built for their life. And now the thing they've been asking of us is like, hey, can you help us adopt agents?

1:09You know, can you help us know how and when to use them and when it's safe? And as you can imagine, there's a lot of consequence and stress and, you know, things in the job. And it's just an interesting moment. Before we get into how you all fit into this physical AI layer and everything happening in that world. I want to get back to what even nominal is in the first place for someone who might not have had the privilege of hearing about it yet. Yeah, totally. Yeah. So I think nominal builds software for hardware teams. We come in and we both, you know, deliver product into their life. A lot of that product is built to help them understand the systems they're producing.

1:42So I think if you're building it or you're testing it or you're operating it out in the field, you know, we help them actually have just like a deep comprehensive view of, is it safe? Can we go forward? Can we go faster? What should we do next? You know, just the life of a hardware engineer is just asking questions. It's a fun company where like we built it really starting in the aerospace industry for years. But a lot of that was about just what the founders knew and how we did it. But really, we now get to serve people in advanced energy and all sorts of manufacturing use cases and automotive and just any complex machine in the world.

2:13Really, the teams behind all of the complex machines in the world is like awesome nominal customers. And my understanding is you all started with like analyzing test data and now you all are going way beyond that. Like, tell us a little bit more about the origin story and how it's evolved over time. You know, we had to solve our customers problems through the lens of just like basic bottleneck theory. You know, it's like, what's the worst part of their day and how do we come alongside them in that? And it started with just how do they analyze the data they're producing? Hardware companies are really, really, really good at measuring stuff.

2:42They can know how hot, how cold, how much it's vibrating, whatever is going on. And like, they're quite good at starting down the path of answering the questions they have. But the thing that, you know, many struggle with is, well, what do we do with it? And how do we make decisions together? Think like, if you're building a satellite, no one knows everything about that satellite. Everyone has a discipline. They know a lot about the battery or they know a lot about the solar panels. They know a lot about the camera. And what Nominal did for years was like, how do we help each of them answer the questions about their system?

3:13And then how do we help them work together to know that all of it's going to work? So think we went through like our analysis era. And then that very naturally worked its way into like, OK, people are really good at like going through this process of building these reports and compliance and plans and things. It's like that naturally moved into like, well, how do we help them like in real time when the plane is in the air or the satellites in orbit or the car is on the road? Like how do we help them in like a think like a mission control context, real time context? So really for the last two years, we've been just deep into real-time data infrastructure, but think like there's so much weird complexities in the physical world where we don't just get to like sign up for an AWS account and start building.

3:51It's like I have a 15-person infrastructure and deployability team where like we're figuring out how do we buy compute hardware that we can deploy waterproof onto a Navy ship. You know, like so and we have to figure out how do we make those systems like mesh talk to each other and do these like hard things. But it's like because we're a purpose built company, you know, like we build for this industry. So it's like, well, what problems do they have? Well, the stakes are so much higher, right? Because you're building for these things that the world depends on. Like, how do you think about building software for this hardware customer in a way that's so different than what everybody else is doing?

4:28A very consequential place to be. If you're going to help them make these decisions, you have to take it really, really seriously. So I think we do help them test their hardware, but like we test our software really intensely. And it is not just like, cool, we can like put it in a CI and roll out 100 changes a day. It's like, hey, what is it? What does great look like? And how do we feel really, really, really confident? Like this is going to be the right answer to this problem. But I think that starts with just like how you design the company. Our customers would say that their like intellectual property is the data we manage for them.

5:01You know, like if you're building any of these things, like so much of what's really, really unique about them is these like really fine tuned things they do. How do you get a little bit more out of this car? How do you make sure this satellite can do this one interesting thing in space that no one else can really understand? You know, we have Design Nominal to be, you know, secure by default. We have all of these really, really interesting parts of how we hire and how we design and how we practice and test and plan. But it's because we take the consequential nature of our customer really seriously.

5:29And like that's why they trust us. You know, like that's why they want to partner with us and why they feel very, maybe for lack of a better word, just like cared for. Well, it's a consequential day. $240 million Series C. What is the next like 12 to 18 months look like as you guys continue to grow with this new round of funding? We went from analysis world to like operations real time world. And then really, before we talk about the next 12, I can tell you, like the last 12 months have been really interesting where like we really believe we were going to serve like core infrastructure to these models that they were buying and lots of interesting things.

6:03But think like we are shoulder to shoulder with them as they're trying to figure it out. You know, we realize that like, oh, we are this interface for them. Like they just think about us as their technology partner. So what do we do? So we've been building this thing we call the analyst, which is our first in-platform agent. I think the next 12 months is going to be a lot about, you know, how do we help this industry adopt agents in a way that they can trust it? And they know in these consequential moments it's telling them the right thing. And way more importantly, they can audit what it's telling them.

6:31So say I am testing a new robot or deploying a new robot or drone. What is life pre-nominal or what is life using nominal versus, you know, a random regular AI model? So, you know, imagine we were going to go build a new, what should we pick? A boat. We're going to go build a new boat. It's going to be a commercial boat. It's a wake boat. You know, we're going to go on a wake boat today, but we're going to build a new one. It's going to be battery operated. So think like those engineers have to think about, okay, how do I make sure when I'm on the water that the batteries regulate to temperature correctly and when they overheat serve particular things and there are different types of motors that's making things happen.

7:05So imagine a world where we were out on this boat in the field and the motor just went out on us. Yeah. We would start asking the question to say, okay, why? Why is this motor not working? And so we'd start with like, well, what happened right before it went out? Was it hot? Did it go faster? Was it vibrating? You know, what was going on? But we're in a weird place where we have no internet connection. Okay, so your problems begin. Amplifying the number of problems. And we're stuck in the middle of the lake. And so think your whole world ends up being, okay, well, now we've got to solve the problem to get out of the lake.

7:42I've got to go home for dinner. and I've got to like figure out what else can we get out of the boat? Like, can we get it operating again? Can we keep testing today? Think our customers are like resource managers. You have a very finite amount of things in the physical world that you have to learn from. And so they're always thinking about like, what can I get out of that boat? Okay, so imagine we also, we realized we brought the Starlink with us. We do have internet. Back on life. Okay, so half of our problems are solved. That's where models come into it. So think like we actually do see a lot of our customers literally just like flip on a Starlink.

8:16And for a long time, they would like take all of this data that was being produced out of various systems. So think the boat was producing data, but also we probably instrumented that boat. We put other sensors on it that measured extra pressure, temperatures, voltages, vibrations. And like, you know, same thing happens in aviation, every other piece. We would take what the boat's telling us, what all the instruments are telling us, and we'd start trying to feed that into models. That will either take forever to upload, problem one because even if you have an internet connection it probably sucks and then when you get there the models will just start like telling you stuff about like just this like little scope of data you know i probably was only able to give it the last 10 minutes think 10 minutes of data on some of on this boat terabytes of data as you can imagine like maybe maybe and it could weed through those terabytes and start to tell us what's wrong but with nominal like every day builds on the last day.

9:07So when they come in and we've streamed all this data in and they are like real time, they're engineers looking at it in nominal. The moment the motor goes out, they can just start asking questions and say, hey, what happened? And they don't just start looking at the last 10 minutes of data. We can look at all the data in that boat's history. We can look at the with analysts and with just our analysis products and our real time mission control products and think like it all works together. It's like you can't just have agents spitting crap at you. You need to understand what are they telling you?

9:36They need to give you these deep visualizations. They need to do it in a way where, you know, maybe I can call someone back at headquarters and they're like, you know, they're diving into it with me. So I both get that collaborative aspect, but I also just get this like lineage. I get to actually, you know, think the answer for a lot of these weird problems is like, oh, someone yesterday actually like wrote a comment in a nominal procedure. and that comment said like turned on motor didn't turn on flip switch turned on second time you know and you're like okay what happened right there and you know they get this magic moment we're like now with analysts and we'll know how to like fold time so it can just be like oh well like i saw it a day ago here's what happened now i'm gonna put those right on top of each other like they're exactly the same okay we have really close to root cause or at least we are starting on the path to root cause that's just things that like you can't just log into clod and go do yeah um you can't right and think it all just keeps going they can go back into the designs of the boat and not at all they can go through when was the change made to that motor like you just you need the whole understanding of it which now i can say is like a context problem because it is either in the heads of these specialists on decades of paperwork about this thing or in these like crazy real-time decisions people made in the accumulation of those crazy real-time decisions.

10:55So my understanding is obviously AI is as good as data underneath. And like specifically with this space is there's a lot of very messy data that's extremely fragmented. And so under the hood, what have you all done to be better at measuring some of this as opposed to just guessing? You know, think our customers have a couple different data problems. They have the context problem, which is how do we actually understand just like the context in which this data was produced? How hot was it outside? Where was I on earth? You don't just care about the temperature. You care about the temperature and context of the environment and the engineers on staff and that kind of thing.

11:28So think like there's a context problem. How do I wrangle all of that understanding of how this was produced? And then it is once it is produced and once you have to actually manage it, how am I doing that in a way where I can, you know, recall it and compress it and mesh it, you know, solve all these connectivity problems. So we have been building just best in class data infrastructure from day one. And some of it we take out of like, how does Vercel work for front ends? But some of it we have to like literally create. No one has ever had to solve a problem that looks exactly like this. So we get to take like, here's best in class what's happening and then build our own version of particular things just because of the constraints of this world and bring them right on top of each other.

12:09And that is really valuable thing to these companies. And that's what lets them kind of go through this three-step process with Nominal where they like, you know, first get to the point where they can just like answer basic questions about their machine. And then they kind of like graduate into like, now I have like a watchdog. Like now I'm like spending my attention in better places. I'm not staring at squiggly lines. I'm like, can I get an extra hour out of the boat today? And I think what's been fun is now watching people who have done those first two things, right. Go say, okay, what can I do with a like proactive agent?

12:37Well, the agent's like sending them a plan in the morning. That's like, Hey, it's all your plan yesterday. Like, have you thought about trying this, this, and this? And it just becomes this magical partner for them. But it's because it leans on them for the expertise. It's all about the engineer. And that kind of creates this really nice harmony. They have a new coworker and it works really well. Say I'm a big tech exec at an aerospace or defense company. How do you pitch a business case? What's the pitch? Literally on our pre-seed deck, the headline of it was how do we help hardware teams accelerate with confidence.

13:12Those are the two keywords. So I think what we bring to these organizations is we help them move a lot faster. And you're like, well, why are they not moving faster today? Sometimes it's by design. For that executive, like, think in our world, in the industrial sector, many of the executives, like, came out of the trenches. You know, like, they were the best engineer. And then they were the best engineering manager. And then they were the best chief engineer. And now they're managing a program that we're all going to count on to keep us safe every day or something of the sort. So they like think they're technical.

13:43They know the problems. They know it forward and backwards. But they're also like have scar tissue. Of course, they got left in the middle of that lake. They got left in the middle of the lake. Or they like got deposed, you know, like because someone on their team made a call that they checked a box on and they signed their name. And it freaking matters. That's the reality. They are always in this place of thinking about how am I making sure I set these gates to make sure people are not being willy-nilly. working together or building real consensus or doing all of these things. And so acceleration for us is like, yes, we're getting really good at helping people like take a product that's a prototype and get it into production and then get more out of the production cycle.

14:17But think that's like one big loop. But like how you accelerate a big loop is actually a bunch of small loops. How do you help one when it's on the end of the manufacturing line? How are we going to feel confident it's ready for the lake? Keep with our boat. Like when it pulls off the manufacturing line, what's going to make us confident that if I put this thing on the lake, These people are not going to get stranded. That is a really important problem. If we accelerate or automate that step, great. Where do we go next? And it's like, well, once we get to the lake, how do we check out that it's ready to go on the water?

14:45Did it, did anything happen in transport? You know, like, so you just, you work through this as an engineering team, but all of that accumulates into, hey, well, like the next boat, it got that benefit of the automation and then the benefit of the acceleration. And I didn't have to send an engineer to the field. That engineer got to go design something. Like if you're an executive and you're watching this accumulate around you, it is so absurdly clear. But the really, really awesome thing is like the reason they buy it, the reason they do it in the first place is because we care about this confidence.

15:15It is not just about I'm going to automate everything. It's like I'm going to automate it in a way that you can always audit it. You can have all of these on off switches. You can choose like and this is like confidence. And how do we help engineers be confident in themselves? confident in each other, confident in the machine, confident in their boss, their boss be confident in them. It's like design principles. So as we think about how we design front ends and what the agent tells you whenever it gives you an answer, like we literally have like pieces of the prompt that structure the response such that like it front loads very particular information and like that is confidence.

15:50So purpose builtness and we help them move a lot faster when you're doing work like building a boat, like it matters. It absolutely does. And you're giving people the confidence to continue to accelerate, which makes so much sense. Talk to us a little bit about the team that you guys have and maybe what you're looking for. We are always just thinking about any job we open, like how do we get the best person for it? And I know it's like, yeah, you read that on a blog post, but like at nominal, what that looks like is sometimes it's like, bring the person with the expertise who has been stranded on a boat, you know, and go find that person in the universe and make sure they come on.

16:26But other times it's like, well, who has ever managed 10 terabytes of data that was produced in the last 10 minutes that had a hundred million dollars on the line? Maybe someone at a high frequency trading fund. We bring people from industry, but we also bring people from all sorts of different places and teach them and bring them in. If you start at nominal, any job, most of them end up their first couple of weeks, like on site with customers. We serve people in the physical world and we build where they build. So we literally go to them and spend time with them. And if you're a recruiter, you will actually probably have also just went and seen how a boat is built one day.

17:01Like, we get the gift of this physicalness. It's very tangible. You're in the middle of it with them. And you're like, I get what we do. And I get why it's important. So if we can find this like best in class talent from outside, sometimes it's an AI researcher, sometimes it's great data infrastructure people, sometimes the best recruiter on the planet, they don't have to know, I don't know, thermodynamics. to be successful at nominal, but I am going to teach them why things get hot and hopefully inspire them a little bit about the people we get to serve, because it all kind of comes back to like when we win, it's people that devote themselves to the mission, you know, that come back in.

17:36My dad was an engineer at Harley-Davidson. You know, like my grandpa did something or I have a cousin that had this. Sometimes it's that personal family thing. Sometimes it's just they're like, I'm so tired of like working in an industry where like I'm just trying to like inch out where nothing is tangible. Smallest little thing. it just like we don't need the 100th vector database you know like they're pretty good but it's just like it's so cool to just like see it physically manifest and work on three years from now what is something that you think will be enabled or exist in the world because of nominal i joke that most of our customers are actually like building things that someone thought of in the 60s 70s or 80s and so it's like i think the really fun part about the next 10 years that people don't realize is like, it's gonna come true.

18:22Like, like these things that have been weird promises for a long time. Honestly, it's kind of like what self driving has felt like the last three years where like, it was always kind of five years away. You had always feel felt a little out of out of hand. And now it's everywhere, right? And in every city. And so you feel like we're gonna hit there with, you know, some of these other Jetsons like things. It's like, I'm gonna be able to, I live in Austin, Texas, I'm gonna be able to fly here a lot faster in five years and that kind of feels to many like 20 years away. I mean it's inspiring that you feel that as an entrepreneur right like I feel like we go New York to San Francisco and I just always feel like it's going to take forever but the idea that it is not one day.

19:02Totally. I absolutely love like what is happening in the current moment in San Francisco other things but think like I spend way more of my time in Nashville and Atlanta and lots of other places out there where like there is like no doomerism right now it's a lot of people who are like extremely optimistic and it's because they're devoting themselves to it it's like we're gonna help people fly faster or like it's gonna be really cool when i get to go to my like niece's birthday party on a random saturday because i can like take this flying taxi thing over there and like it's like that's aviation world energy world i mean quite literally like we are on a like nuclear renaissance like people are solving every problem in the supply chain to deliver a limitless energy the size of a 18 wheeler, like pretty sweet.

19:48And those are our customers and they're going out and building it and you go spend time with them. And you were like, I am very inspired, you know, because you see it so up close, you know, you see person who has three degrees from MIT. That's like the best plasma physicist in the world next to the technician who's like trying to figure out how they're going to produce the hundredth, one of these things. And they are like in lockstep. We have a core value that's empower their mission. And it is like, our customers are literally on a mission. Like, yeah, you don't have to really like question what they're doing.

20:15It's like, oh, this is like 500 people and their job is to make this thing go five times the speed of sound. And everyone's on board. It's like, we're going to do whatever it takes to get this thing to go five times the speed of sound or go to Mars or do whatever it has to do. Fun because it's infectious. It's such a visual depiction of why you should go work at nominal. We talked a lot about what's happening in aviation and energy and everybody's excited about the frontier right now. I think another big topic is obviously re-industrialization. Yes. And I think everyone's on board. We're excited about it.

20:45But I feel like there's still a lot of open questions about how you actually scale all this production. Totally. Totally. Yeah. If you like go on Twitter right now, it's like build your prototype in California. Build your manufacturing center in Austin, that kind of thing. You know, it is a really interesting moment. Think we can talk about like the venture backed world specifically. Like we are in a place where like lots and lots of like venture backed companies have just blood, sweat and tears earned their way to these like prototype systems that they're like ready to take to market. They are really confident if they built a thousand of them, like it's going to work and they're going to do a job really well in the world.

21:18And so we have been just like in this moment with them to say, okay, well, like, what does that look like? What does it look like to just start doing things at scale? How do we start to scale the knowledge we have? How do we start to just get more and see more and do more? And it starts with like your production manufacturing facility, you know, you have to ask yourself the basic questions of like, well, like what, what am I going to subscribe to? Or am I going to do six Sigma? So that like, are we going to assemble this thing in an assembly line? Are we going to do it in pods? So we come with our customers.

21:45And actually we have like, because of the like industry experts we have inside of nominal now think, yes, we do help them go like use our technology. My like favorite cases are like when I go to customer onsites and they're like, by the way, like Charlie taught me that we should have actually been doing end of line manufacturing validation that looks like this, like we actually just get to go out and like instill best practice. And part of it is like the like the longitude and like latitude, we get to see everything at like, we get the gift of not just seeing what is happening in aviation, but it's like, well, what's happening in automotive?

22:19What's happening in energy? What's happening here? And like, they all have something to learn from each other. In automotive, if you are about to bring a car to market, it's a like, super consequential moment. It is not just about like, Hey, well, it's like this car might work. It's like, no, no, this car must work. And like, it's going to go into a customer's hands. It's going to be serviced by not me, the master engineer. It's going to be serviced by a technician somewhere in the world who got a checklist on what to do. Like, um, these are like really important and also just really high dollar consequential things that are happening.

22:51And so they start to solve those problems in their manufacturing line. Um, and they prove at every step that that battery turns on every time or something happens. And like, we've been taking these best practices, we see an automotive and we're like, Hey, actually like aerospace hasn't had to build at scale in a really long time. Um, for a long time, aerospace was like, I'm going to build the most complex thing that has ever existed, but it flies five times the speed of sound. And now they're asking themselves the question, well, how do I build a hundred of those really fast? Um, they're changing the way they operate.

23:20We're coming in and saying, well, like, Hey, actually, like, these are some useful things you should be taking out of automotive. Here's some really useful things. Like you actually don't have compliance that looks like this energy thing, but like the reason they have it is for this reason. We can start connecting these dots for them and institute best practice. And they love us because we are a really good technology vendor and a partner. We are with them in making sure they're going to build a hundred of those airplanes and every airplane is going to work. And then even better, like when those airplanes are in the field, they're going to be maintained by nominal and people in mission control are going to be using nominal.

23:51And oh, by the way, like their design team is going to be running simulations and using nominal. Like we want to fundamentally change the way design happens, building and production happens and the way, you know, you can operate these things out in the field. And like, that is how the future is going to like change around us. That's the only way that we're going to get faster airplanes and cars and trains and whatever else, like that'll improve the world. But like, that is, that is what is coming. Like that is what our next decade is going to look like. What maybe the last two or three years in AI has looked like.

24:21I think that like warp drive feeling is about to happen in the physical world around us. That's what physical AI is going to look like. And it is not the fact that like, this cute robot that's the same height as me can like, go out and do these things for me. It's like, no, no, we're just like, nominal gets to do physical AI by serving engineers, by serving physical discipline engineers with AI. And I think that has a much bigger impact in the average person's life in the world. Yeah, you make us so excited and energized I'm so excited about the future. Good. You should be. I should be. Yeah, exactly.

24:55It is amazing. And if you have an uncle or a cousin or your dad or your neighbor that is a physical discipline engineer, like, go give them a hug and say thank you. Because it was a really uncool job to do for 20 years. The only cool thing you could do is go be a software engineer and have nap pods and stuff or then an AI researcher. Like, yeah, these people are doing it because they care and they want to fix things and they want to joke like my we do like, you know, customer personas and all this stuff. And I'm like, well, if you really just want to understand our customer, it's like the person you invite to a dinner party that like goes out and like fixes your car in your garage and then like make sure the sound system is working and like offers you like 17 different e-bikes you should buy in order.

25:38And like this is like a really special person. Yes. The one takeaway I want you to have is just like go thank that person. because they're really, really important for everything that's to come. Everything. Yeah. To all the people who are fixing and building our world. Yeah. Yeah. Thank you so much. This has been awesome. Thank you so much.