The story of CloudNC: the beginning

Theo Saville
July 18, 2023
The story of CloudNC: the beginning

After nearly a decade of CloudNC, the company is moving forward quickly with new CAM software products, long in the making, that represent a paradigm shift in manufacturing productivity. 

But where did the idea for CloudNC come from in the first place, and why did we build this company to complete a very, very difficult mission - how to manufacture anything in a single click? 

Here’s how it all began, in my own words: 

Baby steps: learning design and finding metal

When I was a teenager, I was constantly in workshops, constantly designing things. I first got my hands on a manual lathe and on milling machines at 13 in school, where design technology (what’s called shop in the US) was my favourite class. I was a certified SolidWorks professional by 20, learning how to do 3D design - but it was very difficult to manufacture the things that I designed, especially when I got to university (at Warwick), and the complexity of designs and materials went up.

An early 3D design and render from Theo’s portfolio of a Hermes special edition Leica, made from analysing photos

So, over time I managed to wrangle my way into full access to the 3D printing department there, which meant I had access to all of these amazing 3D printers that I could just walk up to and use, without asking anybody for permission. FDM, Selective Laser Sintering, Digital Light Process, stereolithography and more. That meant I could design anything I could think of for engineering projects, and just go straight to manufacture in moments. It’s fair to say that I was in love with the 3D printing process for five or six years.

A 3D printing machine that you can figure out how to use, intuitively, in moments

I got very, very used to this process of being able to walk up to a machine and just make anything. I never read an instruction booklet for one of these machines - they were intuitive. Sometimes they broke, and we fixed them. But the point was I could design something and I could make it myself without any outside permission or interaction.

My team’s entry to the International Submarine Races in Maryland Washington. Much of the sub was 3D printed, but some had to be CNC machined. The printed parts holding much of the sub together could be made same-day. The machined parts for the drivetrain took weeks to get hold of.

Eventually though, I needed to start making metal components, which is when I started to encounter problems. There was one university module in particular, called “Introduction to CAD CAM”. It was designed to introduce you to CNC machining, and it involved designing and making a puzzle out of metal. Here’s what I drew up:

It’s still one of the items that I'm most proud of ever having designed. But as part of the competition, we had to make the components with a CNC machine, which is when I was first exposed to a popular CAM package which shall remain nameless - and I thought it was absolute trash.

Unintuitive software that looks like this makes me exceptionally unhappy

It didn’t do any of the things I was expecting it to do: it didn't suggest tools, it didn't suggest feeds and speeds, it was unintuitively laid out. It was impossible to work out without extensive training.

And that put in my head some questions: 

  • Why is this so hard? 
  • Why do these machines sit here doing literally nothing all day? 
  • Why are none of us able to use these machines? 
  • Why does it take me 12 weeks to get something through the workshop when all of the machines are idle the whole time?
  • And why do I need to give a 2D drawing on paper to the manufacturing engineers in the workshop? 

It all seemed mad, and in total contrast to the 3D printing process, where you could walk up to a machine, press a button and get a part, super easy. Why can't the machining process be more like that process? So that's where the idea came from.

First: Entrepreneurs

After I left university, I applied to the Entrepreneur First (EF) program with an idea for a new low-cost metal 3D printing technique, but over the course of the summer, I kept coming back to the idea of improving CNC machines. It seemed so much more impactful if solved: could we make better machines, with sensors that could feed back in realtime to the control software and adjust their cutting to improve productivity? Well, if you did that, there wasn’t any usable software that could do anything with that information anyway, so it’s a bad idea. 

So - why don’t you build better CAM software that makes it much easier for people to use CNC machines, and autocompletes as much of the task of making a component as possible?

At Entrepreneur First (a UK incubator that matches founders together) I pitched my idea on stage, and Chris Emery - who was also at Entrepreneur First, who I’d spoken to a couple of times previously - came up to me and said that it sounded great, and we should work together. 

We went for breakfast, and I took to him straight away: not only was he extremely smart and capable, but he had a very strong intuition for the problem. He’d also been working on a 3D printing start-up, but like me he’d grown disillusioned with his idea. 

So, we teamed up. I pretty much just started teaching him everything I knew about machining, especially about the theory, which I had a good understanding of. I kept building my knowledge there, and he started trying to create a machine learning-based approach that automated elements of the toolpaths. 

Our software engineers going through a rapid apprenticeship in machining at the advanced manufacturing research centre in Sheffield

Soon after, we spoke to some manufacturers and they told us that this was a real problem, so we knew we had something - and we started to think about raising money. We raised seed capital from family and friends, which enabled us to get our first couple of hires, while we built our deck and our pitch. At the end of our EF stint, I pitched to a room of investors, which kicked off the fundraising process, out of which we raised our first seed round from Episode 1 ventures. 

At that point, we thought that we were going to be able to solve the problem well enough to put out a first automated CAM product in around two years. We knew it would be hard, but the deeper we got into the problem, the more sub-problems we found, many of them NP-hard and requiring entirely new computer science to solve. Some of our investors thought that we were sandbagging - that really, we’d crack it in six months, and after that we’d be selling CAM software, and everything would be great. 

To put it mildly, It turned out that that wasn’t the case. 

Building blocks

We discovered that the problem we were trying to solve wasn’t simple - at all. My co-founder Chris has gone into detail as to why that is here, but in brief - accelerating precision manufacturing with software is really, really complicated because there are a) infinite combinations of solutions, and b) the physical environment of machining the software has to handle is incredibly challenging. 

As a result, we found that we had to build up a huge codebase of knowledge to solve all the disparate bits and pieces of the problem - which took much more time than we’d anticipated. And yet we always felt we were so close: but ultimately, we ended up being one year away from being where we wanted to be… for about six years. 

And on top of that insane level of difficulty, we also had to build a company. While Chris and the team were building toolpath algorithms, my job as CEO was in part to hire the people who could hire the software engineers we needed, find an office to work in, and take on the thousand jobs that scaling a company requires. Being a founder means, essentially, you do everything yourself, until you find someone who can do a task better than you can (and can raise the money to pay them to do so).

We also decided to build a full-scale factory, to get to know our industry in even more detail by producing parts at scale for customers from multiple industries and taking on many of the aspects of running a workshop to really get to know it inside out.

At first, we had a little workshop in Bermondsey, where we set up a Haas CNC machine. (Sidenote: when we bought it, it was the first time I heard of a ‘purchase order’. We really were figuring everything out for the first time). 

It took a couple of months to arrive (which I couldn’t believe) and then when it did come, I assembled it myself because I was too impatient to wait for the technician to come and put it together… and then we couldn’t turn it on, because it needed ‘activating’. 

But why did we need it? Well, we needed somewhere to test our solutions, and we figured no factory was going to allow us to use their production machines for experiments. Plus, we also needed somewhere that our developers could actually get to from the office - so we kept it as close to Old Street (where we were based at the time) as possible. 

However, we were separately realising that we probably weren’t going to be able to develop our CAM software in time for the next funding round, and we needed to demonstrate the progress and value of our tech another way. So by building our own factory, then not only would we have our own testing ground, but we’d also be able to measure the impact of our software on the margins of the industry and we’d be able to build software much faster. As a result, we found a facility in Chelmsford, a short train ride east of London, and that’s where we truly went into the business of machining. 

That’s the initial story of how CloudNC came to be. What happened next? Well, there were a few years of inertia as we built our software, raised funds, and turned the factory into a world-class facility with over 15 high-spec machines capable of machining components for leading aerospace, energy and automotive companies… but I’ll tell those tales in the next instalment! 

I’m also looking forward to writing a further part of this story in a couple of years’ time, when our software is being used by millions of people around the globe, and we’ve smashed through the unicorn ceiling to become something even bigger. If you’d like to be part of that journey - whether that’s from the investment side, or if you’d like to work at CloudNC - don’t hesitate to get in touch!

CAM Assist

Your AI co-pilot. Automate CAM in seconds.

Reduce your programming time by up to 80% from only $99/mo.