#1 Theo Saville of Cloud NC on Transforming Manufacturing (Full Episode)

Yes, it's a constant struggle that has lasted the last eight months and will probably continue for the remainder of the company's life. But we've just been essentially making components to prove that we can on our equipment, because what we're all about is automating this particular piece of equipment.

Sure. So I would say that there is this machine called the CNC milling machine, and very few people have ever seen these or heard of them before, and yet they are responsible for the manufacture of more than a hundred billion dollars worth of components every single year. They make everything from iPhone bodies to jet engine turbine blades, injection molds for plastic... they cut blocks of metal into useful shapes with spinning custom tools. And the issue is that they should be automatic, but they are not today. If I want to use one, it's not like walking up to a 3D printer and just plugging in a 3D file and pressing go. What I must actually do is hand my 3D file to a very highly skilled human operator called a machinist, who will then use their years of experience to figure out how to cut that block of metal into your final component. This can take anything from 20 minutes to more than a hundred hours. And then once that program has been generated and given to the machine, what we typically find is the components are being made less than half as fast as they theoretically could be. It's very hard to figure out how to cut a block of metal into a shape when you have a trillion ways of doing it. So that means... CloudNC automates this task. Just give us a 3D file, tell us the machine and tools you have, and then we run a vast artificially intelligent mathematical optimization trying to find the fastest way of making that part. This takes minutes rather than hours, no skill versus years of experience. And when you get your outputs, the program that the machine will use to cut the part is much, much faster than what I would be able to come up with just by thinking about it.

Sure. Well, if in ten years time we achieve everything that we want, and that's a big if, the way the world looks is that manufacturing is as easy as using Amazon. I should be able to design a component, upload it to CloudNC, get automatic design for manufacturability analysis, and I should be able to get either one or ten thousand components of that in one to ten days, and I just don't need to worry about it. I never need to talk to a factory. It is smooth, it's simple, like using Amazon. Now, getting from what we have today to that is just immensely challenging. There are so many technological and market barriers in the way. If I actually wanted to buy that off the factory today, I'd probably have to email a hundred factories to find maybe seven that would be interested and able to do it, and then it would be a six week lead time. So we're a very long way from that today.

So this is a component with a vice clamp on it. Again, although this looks simpler, it's actually much harder than that. It's got things that need to be cut on almost every single side.

That one I think is called Sriracha.

It's getting there. I think we've done 40 or 50 parts that we've actually given names to so far. But I think this one, Tabasco, is the original. It is the best.

I think that was the thing that felt unnatural. So when I came across this concept and when I was asked to find EF, it was just the most obvious thing to do. It was a one second decision process to apply and join.

Yes. So I didn't have a list. If anything, I really didn't know. I didn't have a clue who I was looking for. All I knew was that I had this problem. I didn't even really know how hard it would be technically to solve. And I just thought, hey, I'll pitch and I'll see who's interested. And so we did. And over that six months, I think we still have some of the Facebook conversations where I pitch him and we start talking about it, and we both think that this is much easier than it's ultimately going to be. So when the day came, the process of finding a co-founder took all of 15 seconds, and we've been working together ever since.

Yes. We've had a slightly easier journey than most, I think.

We overlap a little bit in the middle. So the roles were instantly obvious, and that has worked out extremely well for us. I think it's very important that you have that broad coverage in a co-founding pair: someone with a lot of business acumen and someone with incredible computer science skills, and some knowledge of the market in the middle. Because if you have that, you're constantly looking at all directions at all times, and it is very hard to be taken by surprise by some huge technical risk or some huge business challenge. And that will make your startup life an awful lot easier.

Back at that first Facebook conversation, Chris and I had, we thought that it was going to take two working years to get to the point where we will now be in eight months. So we thought it would just take the two of us that little bit of time. Actually, it's taken a team of 14 software engineers two and a half years, and we're not done yet. So the technical challenge is absolutely immense, but that is mostly solved now. But the other challenge we faced at the beginning was it was very difficult to convince factories to talk to us because we had this totally crazy idea. As far as most factories are concerned, the concept that this job could be automated is almost ludicrous because it is so complex. Before we found some people who were willing to take a chance on us and back us, that was probably one of the biggest early challenges. In fact, we've only just hit the point where it is relatively simple to get introductions and be taken seriously by the manufacturing industry, by virtue of being able to actually cut metal and having a working prototype. That was a very long journey.

I think the main thing was just being able to ask a question which isn't answered there, and just being able to ask directly. That was a great opportunity.

Unfortunately not. He's a very busy man. Maybe some other time when CloudNC is worth many billions of dollars, hopefully.

It is incredibly difficult. It is one of the most important things that you will do as a founder, possibly even the most important thing, assuming that you can gain funding. We've hired 14 software engineers so far, which falls mostly into the domain of my co-founder. And I think we interviewed people for maybe three or four months before we finally made that first hire. We were getting a little bit of pressure from our VCs and from our advisors, quite rightly saying, 'Your standards are too high.' And we were very lucky that Chris didn't give in to my requests of 'Come on, hire someone, please, anyone at this point.' But no, we stuck to it, and then we made our first hire, and it turned out incredibly well. And through that very painful process of finding that first hire, we learned how to hire incredibly quickly. The other thing I'd say is it is very, very difficult to hire for a startup if you don't have enough money and if you don't have interesting work. In order to make the absolute top quality hires you need to work in a deep tech company, you need to be able to pay really ideally above market average for at least software engineers, and really for everyone actually. But you need to be able to pay above market, even if it's just slightly, and you need incredibly interesting work. We were very lucky that we raised a large enough seed round to pay slightly above market, and also the work is absolutely fascinating algorithmic stuff. It's really, really good stuff to work on. When we tried to hire for roles that weren't that interesting, then suddenly we found that we were going nowhere. So our recruiting capability as an organization took that workload off myself and my co-founder. And that's something that you must be seeking to do as a founder constantly: it is to automate yourself. You must run around all of these company things that are going to be in total chaos and leave behind organization and automation as you go. And that is your constant job.

What probably surprised us was firstly how incredibly difficult it was to find anyone, to convince anyone to come for a job interview with us. It was really hard. And then once we got good, how easy it was to find a certain class of software engineer that loves working on our algorithms, and how difficult it was to find the people that we thought would be easy to find. You can expect to be surprised by just how much more difficult it is to find certain people than you were expecting.

I think the main thing is that we've had a very easy journey as a startup. We haven't had that 'oh' moment that is talked about constantly. We haven't had any of them so far, touch wood. And that's mainly because of what I talked about: Chris and I both have that 360 degree vision on the business, and we pair that with constantly talking to our advisors. We are very acutely aware that we are first-time founders, we've never done this before, and frankly we have no idea what we're doing regarding most of this stuff. But whatever it is, we have an incredible network of advisors who we can go to anytime, day or night, about anything, and we talk to them on almost a daily basis. So I think that's one of your biggest roles as a CEO: to bring the best advice into the business.

I think the main thing I've learned is that I love creating some kind of repeatable process. That's just not interesting at all. Creating a process that is a lot of fun but that has an endpoint, then you need to move on to the next thing. And I think if you're an employee, you're not going to have that constant stream of incredibly complex problems where you have to relearn everything you know. As a founder, you need to figure out how to do everything to a basic level. You need a basis in tech, for us it's technology and manufacturing and sales and marketing and law and PR and accounting and finance, and so on and so on. Not to forget fundraising as well. You have to learn all of these things from scratch, and that is a very exciting journey.