There’s been a lot of talk about a hardware renaissance, with Paul Graham, the White House and the New York Times among its proponents, and Fortune raining on the parade. We believe that the hardware renaissance is a very real phenomenon, as demonstrated by the Maker movement, the explosion of 3D printers and the newfound ability of startups to raise capital on crowdfunding sites like Kickstarter and Indiegogo – it seems like every other project on there is hardware (and at least half are either 3D printers, coffee paraphenalia or drones!) On the other hand, Fortune has a point-getting hardware right is hard. The gap between idea, design and low-fi prototype has gotten smaller, but there is still a chasm between that prototype and mass production.

We spoke with Scott Miller, CEO at Dragon Innovation and partner at Bolt.io, who is focused on helping startups cross that chasm. Scott knows hardware development and scaling firsthand: after studying engineering at Dartmouth and MIT, he worked with Disney Imagineering and iRobot, where he took the Roomba from prototype to high volume production. He created Dragon Innovation as a way to help startups make the jump to mass production in China, helping them select and work with contract manufacturers. They’re at the cutting edge of the hardware renaissance, taking it past the hype and into reality. Interview below:

Enter the Dragon

Boris: “What are the pitfalls you’ve seen with hardware entrepreneurs going to a working prototype and the first production run?”

Scott: “What we always require for the Dragon Certified Program is that the entrepreneurs need a working prototype, CAD files and a bill of materials, so that we make sure they’ve got that de-risked before they crowdfund, because the real challenge with crowdfunding…we think of it like a chainsaw. So it’s an amazing power tool that if used properly can give you tremendous efficiency, but if you don’t know what you’re doing, you’ll cut off your leg, and bad things will happen.

For us it’s really important that people have a prototype that works ahead of time. We’ll always use an 80% complete point. It doesn’t need to work very long and it doesn’t need to work very well, but the key areas of technical risk should be substantially reduced so much more than what we call a “looks-like model,” which is an industrial design with a gray coat of paint that you can make a compelling video with. If you do that, you can get in a whole lot of trouble in that building things is super-hard, and if you haven’t proven it out ahead of time, you shouldn’t be crowd-funding until you’ve been able to show to yourself that the product you’re building actually works.”

B: “Most guys don’t have those tools in house…you have a 3D printer available and a laser cutter available in your local hackerspace, but that’s it. How do you see more creative entrepreneurs bridging that gap between a CAD design and capacity? ”

S: “Resources like Tech Shop and Artisan’s Asylum in Boston are great. It’s a really small monthly membership that gives you access to a wide range of tools. And for those who have a little more money, to buy a Makerbot or something like that isn’t terribly expensive, and you get tremendous value from it.

What we’re focused on at Dragon is helping people build successful companies, and those would be targeted towards, typically, 5,000 to 5,000,000 pieces. We find that the entrepreneurs who are well positioned to do that have been very entrepreneurial as it were in terms of figuring out where they can get access to these tools, whether a tech shop or even some schools. There’s great resources like GrabCAD where you can download and share files, GitHubfor software and Upverter for electrical. Today you need to do so much less work than you did back in the day at iRobot, where we were, for example, designing our own H-Bridges and working in bare metal. Now you can buy, for 20 bucks, an Arduino on Adafruit and be spinning motors in no time.”

B: “That brings me around to my next question. The hardware renaissance is bringing things back around to where they were 100 years ago, where you didn’t necessarily need a master’s in engineering to create innovative hardware. We’re seeing things like Cubii, a phenomenal piece of exercise equipment made through Kickstarter, where none of the team have an engineering background, right?”

S: “Yes! There’s a whole spectrum of the product development timeline-going from the idea, validating the product-market fit, building a prototype, crowdfunding or raising some external finance, design for manufacture and assembly, manufacturing, pre- and post-production and so on. And where I think where things have changed, for non-engineers who do not have a formal background, they can get involved much, much earlier when they recognize a gap in the market, and can very quickly build up a working prototype. But as you do get to the right of this timeline, where you start building real products, then it is essential to have people who have real experience and network to make sure you can navigate those waters safely. Otherwise, failure points we’ve seen are that people will pick a factory that’s not well-suited for what they’re doing. They’ll rush into tooling too quickly, because the factory doesn’t necessarily have any real motivation to make sure that they’ve gone through the full quality control. Or at least the factories that you’d probably find on Google wouldn’t. And they end up with a lot of product that just doesn’t meet the quality spec, or basically doesn’t work, and they’ve wasted all the money that they’ve raised.”

B: “So you provide that layer, between the factories and the entrepreneurs. Because you only have one chance as a startup, you can’t really blow it on a sub-par contractor. One of our partners here in Iowa which makes molds, a lot of their work involves fixing cheap Chinese molds which the customer thought they were saving money by buying, and which end up cracking and the startup is stuck, because they had nothing but a price comparison.”

S: “That’s exactly right. And there is a danger zone between…in the early stages of development, you can fool yourself by getting one working and thinking, “oh, I’ve got an Arduino working, great…how hard is that to scale up?” And that’s the danger area because you can get way in over your head before you realize it.”