Insights for Success from a World-Changing AI Startup

Intro 00:01
Welcome to Startup Success, the podcast for startup founders and investors. Here you’ll find stories of success from others in the trenches as they work to scale some of the fastest growing startups in the world, stories that will help you in your own journey. Startup Success starts now.

Kate 00:18
Welcome to startup success. Today. I have Joseph Krause in studio. Joseph is the co-founder and CEO of Radical AI. And Radical AI has raised 55 million in their seed plus round, which tells you they’re some up to something very exciting. They’re working on materials development through R&D, AI, Joseph’s gonna get into it. I’m so excited to have you here. Welcome Joseph.

Joseph Krause 00:47
Thanks so much, Kate. Thank you for having me. Big fan about what you’ve been working on, and excited to dive into all things Radical AI.

Kate 00:53
Yes, I am too. Your background is so different from a typical founder. Can we talk about that for a moment? And if you can just kind of walk us through it and how you got here, that would be, I think, really helpful to set the stage.

Joseph Krause 01:08
Absolutely. I love talking about the background, not because of me, because I think it really demonstrates one of my favorite quotes in the world, which is Steve Jobs’ quote around you can never connect the dots looking forward, you can only connect the dots looking backwards. And my path to founding Radical AI is a true reflection of that quote. So if we go way back, we started out, I was in graduate school at Rice University, and I was working on discovering new advanced materials for the computing industry, semiconductor specifically. And I was going through this process, and was getting super frustrated with academic research, not because academic research is poor, but because I really was focused on how I could make an impact. I wanted to take the stuff I was making in the lab and start to commercialize these things. And so I ended up looking for opportunities to be able to do this. And I got a fellowship at the Army Research Lab, which is this corporate research lab. It’s for the US Army. It’s right outside of Washington, DC inMaryland, and there’s 1000s of scientists and graduate students there working on conducting fundamental R&D, but with the objective of trying to commercialize these things in the future. (Oh, interesting.) And so I kind of took, like, one step closer to okay, how do we start to look at pushing some of this novel research out naturally as the entrepreneur I now am today, it still wasn’t enough. And so I started thinking, Okay, if it’s not going to be here, I have to start a company. I need to actually go commercialize my research. But I had no idea how to do that, and I had read every book and took every class online, but I really thought, Look, I know I’m going to be inside tech. Let me see if I can find someone that I could learn from, that I could understand the way you can actually build a company, as well as the ways to not build a company, and then go forward from there. And so I had cold emailed a few people in New York. Kevin Ryan being one of them, he responded right away. And I actually told Kevin, if you’re not investing in material science, you’re not investing in the future. And he was like, That’s a bold claim. I’ve been doing this for 30 years, and I’m quite good at it. I said, No, but I’d love to come show you what I mean. And so to Kevin’s credit, he said, Come to New York. Let’s do a six month internship, and we’ll see how it goes. And that was, man, coming up on four years now since that time. So I took a leave of absence and joined AlleyCorp, his venture firm, and was there for about two and a half years before rolling out and starting Radical AI. And so there, I did all material science and deep tech investing, from semiconductors to 3D printing and everything in between. And it was a really nice segue into actually forming and starting a company.

Kate 03:44
That’s incredible. So did he kind of, did you guide him on some of the investments based on your background?

Joseph Krause 03:50
Yeah. So Kevin is very good at letting experts be experts in what they are, and really giving kind of a guiding perspective. I’ll give you a perfect example. Whenever we were looking at investing in things, he would always ask, what’s 10 years in the future going to look like? Let’s be thinking where everyone is going, not where everyone is today. But inside materials, it’s up to you, Joseph, to be able to index the market, understand what companies are forming, and really decide, like, where are the real opportunities that we can go after, and where should we go after? And he was very good at allowing a lot of learning opportunities and really myself being hands-on, while also giving a lot of mentorship, a lot of direction, and still being deeply involved in diving into a space and understanding when we should make an investment and in cases when we should not. And so it was a very good relationship between being able to take risks and also have a lot of mentorship in that process.

Kate 04:45
That’s incredible. What a great experience. So you must have seen a lot, but you didn’t see what you went on then to found. (Absolutely) You saw a hole. So explain to us the hole that you saw.

Joseph Krause 04:59
Yeah. So the hole that we saw. So one of my co-founders, Jorge Colindres, who is one of the smartest people I’ve ever met in my life, him and I sat next to each other at AllyCorp. And Jorge is a startup person through and through. He had been on the founding engineering team of two different startups. They both exited, and then he went and got his MBA. And then ended up going to AllyCorp, and was an investor there for about five years, and he was much more on the software, dev tools, started looking into AI. So I come into the office one day, and Jorge is there highlighting research publications. Now you have to understand, as a material scientist, normal, as a software engineer, not that normal. So I was like what are you doing? And he’s like, Well, you know, I’ve been looking into AI. Obviously everyone in the VC world and tech world is talking about it. I don’t understand why if the technology is this strong, all of the applications we’re seeing are really low hanging fruit – email, marketing, et cetera. Why is no one trying to cure cancer? That’s exactly what he said to me. And I was like, well, curing cancer is a hard problem. That’s a good one. But you know what about material science as well? There’s a lot of industries that are impacted by material science: automotive, aerospace, manufacturing, defense, climate, energy, semiconductors, electronics. The most important industries in the world are all a direct result from materials R&D. So Jorge and I spent the next two months reading hundreds of research papers and understanding like, where is AI inside material science going to lead? And we bumped into our third co-founder, Dr Gerd Ceder. He’s a Samsung distinguished professor at Berkeley. Has built autonomous labs in the past. He has one right now, and we really had a great conversation with him around where we thought the future of science was going to go. And us three teamed up, and we launched Radical from there.

Kate 06:57
Wow, and you didn’t see you know anyone doing anything groundbreaking in this area?

Joseph Krause 07:04
So yes and no. The yes part is yes, there was work in this area. We had read papers from big technology, Microsoft, Google, Meta, they had teams focused on it. We had seen some other startups in the past doing what in our space was called informatics, or much more on just the AI side. The no part of the question is what we do today, which is a fully vertically integrated company. So we go from this AI discovery lens through to actually making and testing materials in a robotic lab with the goals of eventually manufacturing and selling those materials at scale. And so in that place, we had not seen anyone building a company with AI and autonomy at the center that was really trying to be a future materials company like Dow Chemical or BASF, and that was kind of the unlock we had. Was we don’t want to just stop at the AI and software level. (I see.) We don’t want to just do high throughput experimentation. We want to build a flywheel at both of these things that can not only generate and discover new materials, but can allow us to scale them with the data as well. And that was the big differentiator that we went out to build.

Kate 08:12
Yeah, that’s a big differentiator. So that’s like delving into Radical AI, that’s what you’re doing. I mean, kind of walk us through in a little more detail, if you wouldn’t mind.

Joseph Krause 08:24
Yeah, of course, absolutely. So I talked about how many industries are impacted by materials. When we thought about materials, and again, I was a material scientist, we have a lot of material scientists in the company, we kind of see two big problems in the discovery sense. Number one, incredibly long timelines and high cost. We usually see north of 10 plus years, upwards of 100 million if you really want to go from that first discovery to full commercialization of a novel material. And so the approach to not only designing and developing but then scaling novel materials is very long and cost complex. And then the second problem was the fragmentation. When we looked inside material science on one paradigm, you had academia right, which is US universities, other universities really driving our understanding of science, but not much focus on the commercialization side. (Okay, I see.) And then the other paradigm, you had corporate R and D. And this R and D, when we talked to the research teams, was very much focused on optimization. How do I drive one, two, 5% improvements in the things I’m working on today, but not much novel discovery. And so in the middle was this wide, open, wide space to do fundamental, really novel discovery, entirely focused on commercial markets that require novel materials. That was where in the material space we really wanted to make an impact, and kind of is why we actually set out to build the vertical approach I just talked about, and where we see us going in the future.

Kate 09:59
That’s so neat, because, like you said, that’s a real area where people weren’t, you know, doing anything and it’s also a huge undertaking. So, I mean, were you a little nervous about leaving AllyCorp? No, you were just fired up and ready to tackle this?

Joseph Krause 10:19
Yeah, absolutely. We had just known, you know, Jorge and myself had sat down early on when we started doing research in this, and then when Gerd joined us as the third co-founder, us three looked at each other and said, if we don’t do this, no one’s going to do this. And when we went out and raised our pre seed round from AlleyCorp, we had said to Kevin and the team, look, this is a binary based approach. This is either $100 a billion company that’s one of the most important in the world, or the technology does not work and it will not drive enough value in the scientific discovery process of novel materials. And we were okay with that binary based approach, and so we didn’t honestly think twice about it. We knew that this was what we were meant to do, and just went after it.

Kate 11:04
It does sound like the way you described the three of you and your backgrounds that you were meant to do it. Like I can see where you all bring something very helpful to the table.

Joseph Krause 11:16
Probably one of the most exciting parts of building a company is sitting in a room with really, really talented and smart people and solving problems that actually make a huge fundamental impact on humanity. You know, a lot of people want to solve hard problems, but what is so unique about materials is that, again, no matter what industry you’re in, it’s going to source back to novel material discovery. And so whether you’re talking about inventing new workout clothes and trying to do cool things there, or you are going into aerospace and defense and some of the more complex material systems that exist today, everything comes back to that same central point. And so for us, we get so excited about solving the problems that I identified, because we think the impact for the human race is really one of the largest it can be when you’re building a company, and that is why we are so passionate and excited about what we are going after. You know, I’m sure we’ll get into this in more detail, but we tell everyone that we interview that wants to come work here. If you’re looking for a job, this is not the place for you. You are so talented, there’s a bunch of other places that you can go work. If you want to work on a mission, then you should come work at Radical AI. And that’s been a big approach that we’ve taken to building the company today.

Kate 12:30
That’s pretty motivating. So before we get into that, I have to ask you to go down this road a little bit more for our listeners, because we talk so much about software on the show. So go a little bit more into why materials are so important. I think people don’t really understand all the verticals you touch and all the things that you’re doing, right?

Joseph Krause 12:53
Absolutely. So I’ll walk through an example of how materials can be so impactful. So, you know, a really exciting thing I always like to talk about is the transistor – an obvious one. Most people are aware of semiconductors and the impact they make today, but if you think back to what they’ve really enabled from the invention of the transistor and then the rollout to actually Silicon Valley, where companies like Intel and others were formed, and there was really a big push from both the government and private enterprise to push this technology. And from that, of course, came the internet, and from that, of course, mobile phones, and from that, of course, applications and other package and ship software. And then from that something like Waymo today, right, where we have always self driving cars that can drive around cities. And when they made the transistor, I don’t think at least, they weren’t thinking about autonomous driving cars. If they were, they had serious foresight into the future. And so this is how materials can make such an impact, right? The transistor, this invention around silicon, and really the way that we use those materials in conjunction to generate current that is the bedrock of modern semiconductor technology today, has led to everything all the way up to and including something like Waymo, and all the devices inside that powered by those same semiconductors. And the reason I tell that story is because how far removed those two ends of the spectrum are. When you are working in autonomous cars, or you might even, like me or my parents, riding in a Waymo. You might not think back to the transistor technology, but something that fundamental from material sense can really impact it all the way up the chain. And so when we think about material science, really the built world around us is material science. The computers that we use today, the clothes that we wear, to the sports equipment that we might use, or the cars that we drive, all of those things come from novel material inventions that enable the technologies to move past current limitations. And we’re seeing this in space today and nuclear energy today, and other spaces where we’re at the forefront of those industries, and we’re seeing novel materials come out for them. So if I had to put it in a sentence, you know, what materials provide are the building blocks to actually enable new technology, and that’s the part that we’re trying to discover at Radical AI.

Kate 15:21
Okay, I love that. And I don’t think people realize how hard it is to, you know, figure out those materials right, and bring them to commercialization. So tell us, how is AI helping you with this? Right?

Joseph Krause 15:35
Yes, absolutely. So whenever you start looking into a material space, the first thing you try to do is index what you should work on, right? And so what I’ll do is I’ll actually come into a space as a scientist, I’ll read a bunch of papers in that space, scientific publications, what has the field done in the past, what can I learn from those? I’ll probably search the web and read some textbooks and study some more about the underlying physics, the underlying chemistry, the underlying material science that actually enable the materials I’m working on. And then I’ll make a new hypothesis, and I’ll take that hypothesis into a research lab, I’ll make the material, and then I’ll test the material and compare that whole process, and then restart again, right? And go back to the top and use those results to continue to move through until I reach that angle. So let’s move into an AI driven materials world now. In an AI driven world, the first three steps of that process, reading publications, reading textbook and searching the web and making new hypotheses are all done instantaneously. Our AI engine can read millions of scientific publications instantaneously. It can actually look at and understand textbooks and patents and other publications or indexing searches on the web, and then it can generate these new hypotheses. And this is really the output of our AI engine. So this is called inverse design. And inverse design is where you take an end property from a system, like a really strong material for a jet turbine or something similar, and you inversely design the material from that. So you work backwards from commercial into discovery. And that’s how you predict novel materials. And this is what an AI scientist would do today. But that’s not where Radical AI stops. The next part of that was making that material, testing that material, and then analyzing it. And we had AI and autonomy in that process as well. So once our AI generates these new hypotheses, it sends them into our self-driving lab, which is a fully robotic lab that has no humans involved that can make these materials. It will analyze or test these materials, and then it will look at the data from that analysis and draw conclusions about what material to make next. All of that process is human driven today, and so AI can sit on kind of both ends of the hypothesis generation side and then the testing and analysis side. And these things together form what we call is a material flywheel. And we think it is this flywheel that gets at that timeline and knowledge of discovery I mentioned at the top of this call.

Kate 18:23
For sure, that’s incredible. Are you kind of blown away? Is it even faster and more efficient than you thought initially?

Joseph Krause 18:33
Yes, it’s quite incredible. I’m blown away by what it can do, but we’re still very early in its infancy as well, which is great, and the reason why, for materials and more on the technical side, is the lack of data. We do not have the data sets exactly that exist in other spaces, like bio, or even way bigger than that, like what the language models use today when you’re building ChatGPT or similar type software. And in material science, we actually have two forms of data. We have what we call computational data, and these are simulations that come from theory that are really predictions or assumptions on where novel materials can exist and are stable. And then we have experimental data, the stuff that comes from a research lab, the real things you do as a scientist. And that experimental data is not really captured as a scientist today. No, we might scribble it down in a lab notebook. You know, I might tell Kate, who works with me in the lab, but I’m not going to tell someone else what we’re doing in a lab that doesn’t work there. And so you never build this corpus of data on all of the experiments you did that probably led to the thing that worked. You know, 90% of the work that we do as scientists is trial and error. That’s what the scientific process is. And so what our self-driving labs enable is us to capture all of that data. We are capturing the temperature, the pressure, the oxidation, the concentration, even something like the humidity in the room, because maybe that impacts the material, and feeding that back into this AI engine. And that data set is very, very small today. It’s a big piece of why we’re building self-driving labs as well. And I think as that data set continues to grow, we will be more amazed to your question, at the predictions and the understanding that our AI scientists will have of certain fields in the future, it is certainly going to be an incredible time in science.

Kate 20:28
Yeah, that’s so cool. I mean, I’m just kind of thinking about all the data you’re going to collect. That’s amazing. I mean, when you think back to how this started, right, with people like taking notes and tracking it and then where you’re gonna go with that, that’s incredible.

Joseph Krause 20:49
And there’s really a cool piece that we like to we don’t get to talk about much, which is intuition. And if you think back to some of the great inventors or scientists of the day, what they really had, after a long career, was intuition. Look, I have done 20 years of experiments so I know what to recommend or guess as a new hypothesis. So as you start to replicate that in the orders of weeks or months, you start to build pathways to understand intuition. Why did we make that decision? What were the defining mechanisms or characteristics of that material that led to that decision? And then can we learn those mechanisms or those pathways and then use them to generate new materials? We’re not there yet, but we’re very excited about that future, because if you can capture intuition, in our opinion, you can capture intelligence. That is the true intelligence of a scientist, and if we can replicate that in some form, then we can truly get better at making predictions and better at analyzing those predictions. And that, we think, is the new paradigm that this technology is really going to unlock. We go from a human driven approach to really an AI and autonomy driven approach, and it is this approach that will remove materials as the biggest bottleneck to our most important industry. So we’re very excited about the way the future will unfold.

Kate 22:13
That is so incredible. I can’t believe it. You talked about this, and now I want to delve into this, that you’re on this mission, and that’s what you look for when you hire. Tell us a little bit about Radical AI, like, what, where it stands today in terms of employees and your culture? Because, I mean, what you’re doing, people have to be passionate about it, I would think.

Joseph Krause 22:37
Correct. Absolutely. And culture is one of, if not the most important thing that we think about. We think people build technology, and people are the most important part of the company that you’re building. And so to attract and really teach and build a culture or build a good organization, all comes back to culture. And so we are really ridiculously crazy about our culture. We are not afraid that we like our culture. We’re not afraid to promote our culture. And candidly, although this may sound harsh, we’re not really afraid to vet against that culture either. There are some people who just are not a fit for that culture, and we do remove them or end up not hiring them directly because of that. So why? Well, the reason is, what we’re doing is very hard, right? To some people, borderline impossible. And to us, that’s the entire reason for doing it, is the tremendous impact and unlock you enable if you were able to crack the code. And so when we think about the people we want to bring into an organization, yes, they have to be technically skilled. Yes, they have to have good experience, or be willing to get a lot of good experience in the systems that we’re specifically building. And yes, they have to be really bought into making an impact, but they also have to just subscribe to the way that we do things. We move ridiculously fast at the company. We have a 51% rule, which we stole, I believe, from SpaceX, which is when you are 51% at decision, you need to make that decision and iterate on that decision, because delaying a decision is more harmful than making the wrong one and learning from that. We think a lot about interdisciplinary approach. You know, if we have five different buckets inside the company, on the technical side: machine learning, software, materials, robotics and mechanical engineering. They all need to work in an interdisciplinary fashion. So all of those things boil down to a lot of our core principles, and the way that we like to talk about culture is a first principle, ambitious approach. We need people who are not afraid to ask why of the hardest problems and processes in the world, and that is the biggest thing that defines Radical’s culture. I don’t care where you come from, what you did before, what you do in the company today, or where your idea is originating from, you should ask why, when we make a decision, because that drives true first principle thinking, and from that true innovation on actually thinking about the right way to approach problems. And so for us, the culture is the most important thing that we think about, and we think as founders, Gerd, Jorge and myself always talk about this, it is the founder’s job to uphold this culture. We always talk about ourselves like a bowling lane with the bumpers up on the side. That’s the founder’s job. Are you not going to have everyone be directly in the middle of that lane? That’s not reality. But you can certainly put the rails up on what the culture means and make sure everyone inside the company either stays within that or is just not a fit moving forward. And we’re quite serious about enforcing that.

Kate 25:54
Think you’re on to something. I mean, I’ve been doing this show for a long time, and all the founders who have been very successful have been so passionate and double down on the culture, right? Absolutely, I think that’s the piece and what you’re doing with everything else. And you add that culture piece into it, I think you’ve got a pretty winning equation there. (Absolutely) We’re coming up on time. This has been fascinating. We always wrap up with just other advice you can share with startup founders listening. I mean, the way you walked through your company and culture was so helpful, but any other parting words of wisdom?

Joseph Krause 26:32
Yeah, good question. Bias to action. Bias to action is the most important thing in the world. It’s so obvious. It’s so cliche. You would be amazed at how many people don’t have a bias to action. And in startups particularly, but in a lot of areas of life as well, everything comes down to taking action. You know, we actually have a funny joke inside the company, because a few weeks ago, I had sent an email banning the word “strategy “for a certain period of time at Radical AI because, you know, so many people like to strategize on the approach that we’re going to take. But again, everything comes down to iteration. You’re never, maybe not never. You’re rarely going to get it right the first time. And so the faster you can execute, the faster you can learn, and then the faster you can update your process from that learning. And I think the thing that founders can do to really make waves in their industry and make an impact is have a ridiculous bias to action. Do things immediately when they come to your desk. Outreach to anyone that could help you build the company. Be a resource to people, so that you can help them build their company, and just have this innate desire and really ability to take direct meaningful action when you have an idea that you want to pursue. I think that’s what entrepreneurship truly is in a nutshell, is having a new idea, vetting it and understanding the approach you want to take, and then executing on that. Ideas are worthless. Execution is everything. There is a reason that saying is cliche. It is because it is true. And so if there was one thing, one trait, I would give a founder, it would be a ridiculous bias to action.

Kate 28:18
I love it. It’s so true. And then the power of the iteration like you said. There’s so many people that hang out in these large companies and worry about strategy and the rollout. But that’s not going to make it in startup world, especially in your space.

Joseph Krause 28:34
I’m gonna get in trouble by someone after this for talking about banning strategy. But I think what you say is so true. You really have to think about iteration speed. I always love the picture, I don’t know if you’ve seen it, of the SpaceX raptor engine, where they had this, like v1 v2 and v3. To me, at a larger scale, but that’s like a perfect depiction of evolution and innovation. We start here, we execute, we get results. We move to here, we execute, we get results. We can move to here, we execute, we get results. And if you would have tried to start at the third engine, you would not have done it. You’d be still strategizing, on the board, how to make it look more pretty. But by just starting and iterating. Of course, you end up at that place, and I think they do it so well as a company. We deeply believe in that approach as well. And so I just think bias to action leads to that and to your point, the more revolutions you can have, the more shots on goal, as the famous saying is, the better chance you have of one of the pucks going into the net.

Kate 29:36
Absolutely. Where can people listening find out more about Radical AI?

Joseph Krause 29:42
Absolutely. So first of all, feel free to reach out. I’m a big believer in just talking to people that you want to talk to. I’m sure you can find our information somewhere, or can guess it. Two, our website radical-ai.com. You can go there and find our careers page as well as a little bit about what we’re doing. And social media is as well. We are on X, we’re on LinkedIn. Feel free to interact with us there, ask us questions there, and we’ll be sure to get back to you. And if there are any people listening that really want to work on some of the hardest problems in the world, in an environment where you’re going to fail daily, but succeed greatly, please reach out to me directly. We are aggressively hiring right now, and we need the smartest, most ambitious people in the world to come help us build this company. So please feel free to reach out.

Kate 30:27
Incredible. Love it. You really are changing the world. You must love waking up every day. I’m glad I got to talk to you early on. You know within a few years when you’re too hard to reach because you’ve done all these things.

Joseph Krause 30:40
I’m looking forward to the few years. I’m gonna reach back out to you in a few years, we’re gonna have a check-in conversation.
Kate 30:47
Promise? (I promise) I’m going to hold you to that. (Please do) What a treat to have you on. Thank you for your time.

Joseph Krause 30:53
Thank you for having me on. Thanks for everything you do.

Intro 30:56
You’ve been listening to Startup Success. To make sure you don’t miss out on future episodes, subscribe to the show in your favorite podcast player. Like what you hear? Tap the number of stars you think the show deserves in Apple Podcasts. For more tools and resources for your own startup success, check out burklandassociates.com. Thank you so much for listening. Until next time.