Lumafield is building the world’s only manufacturing intelligence platform powered by industrial CT scanning. We make systems that empower engineers to see inside their products, accelerate product development, and get manufacturing right faster.
When I talk to potential investors, employees, or members of the media I’m often asked the same questions: How did Lumafield get started? Where did the idea come from? Why did I start this company?
I’ve answered this a few different ways over the years, but I’ve decided to finally sit down and write the definitive story of starting the company so I can point everyone here if they want to learn more than I can share in a couple of minutes.
1987 to 2003: Growing up next to a fab
I was born in Arlington, TX and grew up a 10-minute walk from a National Semiconductor fabrication facility. Today that building houses an Amazon distribution center. There’s an entire essay to write about what that sentence means for the past, present, and future of American manufacturing, but we’ll set that aside for now.
At the time, that fab was just a landmark in my neighborhood, not a clue about where my life was headed or the inspiration for what I would work on for the rest of my career.
2003 to 2006: Discovering physics
My parents decided to homeschool me from kindergarten through 12th grade. I spent my school years reading books from the library, playing video games, and hanging out with friends. I wasn’t a stereotypical engineering nerd as a child. I liked Lego, but I followed the instructions. When I took standardized tests to assess my learning levels I’d usually score roughly at my grade level for reading and language and below it for mathematics. No one from the current startup ecosystem would have bet on me to become a CEO if they had observed me in the early 2000s.
I became interested in engineering and science in high school when I was taking chemistry. I read a popular science book on physics over Christmas break that year to get some much-needed extra credit in the class and was instantly fascinated. Over the next three years I read every book in the Arlington Public Library catalog on physics. When it was time to go to college I wanted to major in physics, but I was worried that I wouldn’t be able to get a job after college with a physics degree, so I chose mechanical engineering instead. After growing up on the south side of the poverty line my main goal in life was to have a stable, solid-paying job after school.
2006 to 2011: A soul-crushing internship and a new plan
I went to Baylor University for college. I chose Baylor because it 1) was a Christian school, 2) had an accredited engineering program, and 3) was a short drive from where my high school girlfriend was going to college in Dallas. In retrospect none of those three things mattered at all. Life has worked out and you can’t run the experiment twice, but I usually tell people my biggest regret was going to Baylor for college over other schools where I was admitted that had better engineering programs, like the University of Texas in Austin, or a more rigorous academic environment like Williams College in Massachusetts.
Regardless, while I was at Baylor I had a summer internship at Lockheed Martin where I discovered that having a stable, solidedu-paying job was actually a soul-crushing experience. I went back to school for my junior year extremely confused about what I was going to do with my life.
I ended up taking a class on technology entrepreneurship where I learned that it was possible to be a university professor, create new technology, and get other people—mostly the government—to fund your discoveries and help you turn them into companies. That sounded pretty good to me, so I set my sights on a PhD and a career in academia.
My grades, GRE scores, and relative class standing led the professors in my life to encourage me to set my sights low for graduate school. MIT, Stanford, and Caltech were scary places where Baylor grads smarter than me had failed. I was advised to choose one reach school in the top 10 and make sure that I went after a wide set of much safer options. I’m not sure who was more surprised when I was admitted to every school I applied to and got a NSF GRF—me or the people who had advised me. I picked my reach school, married my college girlfriend, and moved to the University of Illinois at Urbana–Champaign in August of 2011.
2011 to 2014: My first failed company and the manufacturing bug
UIUC was a bad fit for me. The professors felt distant, the administration was frustrating, the research was narrow, and my peers mostly had ambitions to work at legacy Fortune 500 companies or become professors at nameless schools in the Midwest. Everything about the experience pushed me toward the realization that I didn’t want the life I thought I was signing up for when I went to grad school.
The only good thing about my time in Illinois was that I started a small business. Oso Technologies’ first and only product was called Plant Link. It was an internet-of-things sensor that would send you an alert when you needed to water your houseplants. If your instincts tell you that this is a bad idea, and there’s no way it could lead to a successful company—congratulations—you have better instincts than I did in 2011.
However, Oso was my first exposure to building a real product outside of a lab. We filmed a Kickstarter video with a combination of working prototypes and ID models, sold just under $100k worth of product, and were suddenly on the hook to ship 1,000+ plant sensors to people around the world.
We needed to get them manufactured and we had no idea what we were doing. I remember touring a few small manufacturing houses that wanted our business. There wasn’t much to our product—a couple of PCBAs, some injection-molded enclosures, etc. You didn’t need to be Foxconn to make Plant Link. But I was shocked by how primitive the quality control processes were at every contract manufacturer we visited. At best, they had a person look at everything with their bare eyes and make a couple of notes on some paper. At worst, the entire thing was running open loop.
It was an odd introduction to the world that I spend all of my time in today. We persevered in the face of obstacles and ran the company straight into the ground. It was a messy wind-down that had more than one false ending and culminated in a sale of the company to Scotts Miracle-Gro for less than the amount of money we raised to get the company off the ground. You live and you learn.
In this case I mostly learned two things: 1) how not to run a company, and 2) that I wanted to go work at a real startup that was solving hard problems in manufacturing instead of making gadgets for Kickstarter.
2014 to 2018: Formlabs and the limits of 3D printing
If you wanted to work on manufacturing hardware in 2014 you basically needed a reason not to end up in 3D printing. I didn’t have a reason, and I was lucky enough to end up at Formlabs, where I kicked off the development of and ultimately launched the Fuse 1 SLS 3D printer. This isn’t a blog post about Formlabs, though—you can learn more about them from the movie, their YouTube channel, or their blog. Safe to say it was—and probably still is—the highest-potential 3D printing company in the country when I joined.
Unfortunately, four years into my time at Formlabs I had a falling out with my colleagues that basically boiled down to all of us being too stuborn to figure out a way to work together despite our differences. Fortunately, it turned out that 3D printing is not actually a manufacturing technology, so leaving Formlabs gave me the chance to hit reset on my career and get back to the thing I was passionate about when I first joined the company: how real products get made, at scale.
2018 to 2019: Wandering around and obsessing over production
I spent the next 12 months traveling around the country and the world thinking about what I wanted to do with the next decade of my life. By consulting with other startups I experimented with everything from autonomous airplanes to shoes to new 3D printing ideas. None of those things really scratched my itch to work on the problems I cared most about: how manufacturing actually works, why it’s so opaque, and why it still felt like the weakest link in the product development chain.
I would wake up in the middle of the night thinking about ways that carbon fiber layup processes, injection molding for foams, and heat-treating techniques for castings could all fail. I knew I was thinking about the right problems, but I didn't have a solution I wanted to go all in on yet.
2019 onwards: Why CT and why Lumafield
I was spent the summer of 2019 hanging out at MIT’s Center for Bits and Atoms where I was researching ways to make obscure display technologies more affordable. CBA had an incredible collection of tools including the first industrial CT scanner I had ever seen in person. I was intrigued by the idea of taking physical objects and turning them into super accurate digital representations. Almost like a reverse 3D printer.
That June I was invited was at a dinner that Onshape and Desktop Metal were hosting for hardware founders out in Burlington, MA. Ric Fulop and I were catching up after dinner when he told me that he wanted to build a CT scanner as a new product line for Desktop Metal. I said that was a bad idea—that CT scanning wasn’t close enough to 3D printing for it to be a product line, but that it might be a good idea for a standalone company.
He instantly agreed and asked me if I wanted to start that company. I was surprised and said I needed to do more research before I had an answer to that question.
Over the next couple of months I put CT scanning through the criteria list I had been using to evaluate potential ideas for the past year. Anything I was going to work on needed to check several boxes at the same time.
Industrial CT scanners are essentially X-ray machines that create 3D images of a part’s interior. Instead of just seeing the outside of a shoe sole, a turbine blade, or a battery, you can see every internal feature, every void, every defect without cutting anything open. The technology already existed, but it was slow, expensive, and locked up in specialized labs like CBA at MIT. Almost no one in day-to-day product development or manufacturing had practical access to it.
If CT could be made dramatically cheaper and simpler, it could move from a niche lab tool to something engineers and manufacturers use every day to understand and improve their products.
Here’s the checklist I was using at the time:
A solution to a big problem in manufacturing.
I wanted to work on something that mattered to how physical products get made, not just another prototyping technology like the Fuse 1, much less another gadget for the dumpster like PlantLink.
A hard technical problem.
If you have to pick between two things to work on and everything else is equal, pick the harder one. Harder problems attract smarter people, solving hard problems is probably more valuable, and solving hard problems likely means that you have at least some kind of moat around what you’ve created.
Hardware in a market that was held back by being too expensive.
This is where I directly ripped off the Formlabs playbook. Formlabs took the technology behind a 3D printer that cost tens of thousands of dollars and made it cost a few thousand. At Lumafield we wanted to take a technology that cost $1M and make it cost closer to $100k. An order of magnitude cost reduction seemed like table stakes.
Cloud software as the core value proposition for the company.
Cloud software as a service is the greatest business model in the history of capitalism. I wanted the heart of the company to be software, even if the tip of the spear was hardware.
The ability to produce a uniquely rich dataset.
I spent much of 2018 taking online classes on neural networks. It was clear to me that the most interesting AI companies would be the ones that could generate their own proprietary data instead of training on the same public datasets as everyone else. Creating unique datasets for each of our potential customers felt like an extremely
A consumption-based business model.
I wanted usage to be directly tied to value, and I wanted a model where our best customers would naturally grow with us over time. We ended up deciding not to charge per scan so this ultimately ended up not being true, but the potential was there in the beginning.
Something other people wanted to work on with me.
Andreas, Jonathan, Kevin, Scott, and Ric were on my literal short list of people I most wanted to work with after meeting them throughout my time at Formlabs. I knew that if I could get some of them to take a leap and work with me, I was probably onto something. Getting all of them to join in was ultimately a dream come true.
It turned out that CT scanning checked all of those boxes and more. I decided I was in. Ric and I had lunch at a sushi restaurant in Lexington, MA one day to hash out the details—equity, who was going to be CEO, etc.—and then we got to work recruiting our other co-founders, opening a bank account, incorporating the company, and calling VCs to raise money.
We had a legal entity by the end of July and our first term sheet in early August.
Since then, Lumafield has gone from a handful of people and a bank account to a company with multiple product lines, customers across a wide range of industries, and a growing team that’s still obsessed with the same problem I stumbled into at my first company: how to bring modern tools, software, and data to the invisible world inside the products that shape our lives.
As a kid, I would ride in my mom’s car past a semiconductor factory I didn’t understand. Today, I get to spend my time in factories helping the people who build products solve critical problems faster than ever before. That’s the work that Lumafield exists to do. That’s the work I’m excited to do every day when I get out of bed in the morning.
That’s how Lumafield started.
The rest of the story is still being written.