What are some of the ways in which you’re seeing emerging tech, whether it’s 3D printing, additive fabrication, or robotics, or the internet of things disrupting the way products are getting designed and the product design lifecycle?
Yes. We see a lot of really interesting things happening in the space, and I’ll give you just a little bit of an example or a story. When I was at iRobot where I spent 10 great years, one of the things we ended up doing is building the Roomba, and I’ll speak to this both mechanically and electrically. Mechanically, we wanted to get a working prototype to be able to understand how the robot behaved in unstructured environments, and for sales and marketing, and all that good stuff.
We would create the files, in our case, in Pro/E which is funny because not as many people use that anymore and build truly like $25,000 models of stereolithography or SLA, which was incredibly brittle, and there’s all sorts of examples of us like turning off the cliff detectors and having the robot just drive off the end of the table and shatter itself to pieces. Today, using either … You could pick MakerBot for FDM or Formlabs for SLA for much cheaper price. In fact, for a couple thousand bucks, you can actually buy your own machine and be able to create models that work even better than what we had 10 or 15 years ago at a fraction of the price at a much, much quicker iteration cycle.
Rather than having to wait say a week or two weeks to get your parts back, you can even have them back in the morning, and this lets you go much, much faster. Not only from the reduced iteration or the reduced build cycle, but also, when you’re paying $25,000 for a part, you want to make … Or for a robot, you want to make sure every last thing is perfect, so you spend a lot of time double-checking your work and that’s … It’s the old 80-20 rule, except you’re on the wrong side of it. Whereas if you can get a version in a day and it cost you a couple hundred bucks, you still want to be thorough, but you don’t need to dot every “I” and cross every “T”. That just lets you go … It basically democratize the process of getting a working prototype.
On the electrical side, again, at iRobot, when we wanted to build the first circuit board to say spin the wheel modules, we had to get down in the bare metal and designer on each bridge with flyback diodes and transistors, and you’ll figure out what component to pick and actually do the hardcore engineering. It took probably a month between designing it, sending the board out, getting the board back, and writing the code just to get a simple motor to spin.
Whereas today, literally in like 20 minutes, my 7-year-old son can grab an Arduino, copy and paste some sample code, and adjust the key variables, and he’s spinning motors. There’s just been like a really interesting obstruction from the complexity of how the thing actually works to much more of a, “Let’s focus on getting the product working and not worrying as much about the details.” I think that’s incredibly enabling for the prototype.
Now, when it actually comes to building the thing in volume where you’re writing much bigger checks and you’re working with statistically larger samples where quality is important, then you definitely better understand how that stuff works, but that’s something you can push off till later after you’ve proven out that there’s a product market fit for your prototype, you’ve got the right features, the right price point. Those are just two on the electrical and mechanical side that have been key enablers.
On the software side, there’s three main things we look at. One, everybody knows about GitHub for software where it’s easy to not only keep track of it, but to share your code. The same thing exists in electrical. We love a platform called “Averter”, which is the same way you can share reference designs, so that you don’t need to figure it all out from scratch. Then mechanically, GrabCAD I think does a great job of again being able to exchange a different model, so you don’t need to build it from the ground up. That’s some of the software enablers.
Then if you layer on top of that, in terms of the CAD world, I had mentioned that we use Pro/E, and then most companies that we work with switched over to SolidWorks, which is just much cheaper and much easier to use. Basically now, there’s another revolution going on where the folks … They’re just downstairs from us. They’re all this … The former SolidWorks founders have created a company called “Onshape”, and this is, as we see it, like the next generation of CAD, so that rather than paying … I forget what … How the work is, but let’s say it’s $5,000 a year with annual subscriptions of a couple thousand.
With Onshape, you can pay, let’s say, $100 to $200 a month for as many seats and per seat. Ramp up your seats, ramp them down, and it’s all through a web browser, so it doesn’t matter if you’re on a table, on a Mac, on a PC; and there’s just a lot more functionality built-in, so that’s makes it … Like most of entrepreneurs are going to have trouble writing a $5,000 check for SolidWorks, but to write like … Onshape is free. Then, if you want to have privacy, I believe they charge you for it, but the walls or barriers to enter in hardware getting from an idea to working prototype over the last 5 years have just gotten so much lower, and I think that’s some of like the key enablers that are driving the hardware renaissance.
You talked a little bit about how prototyping is becoming faster and more iterative in the hardware space. Once you proven out some of those ideas for your product, what are the next steps? I know Dragon Innovation is really involved in taking those initial ideas and helping people get to the small … Maybe the smaller quantity, and then even ramp up to much greater quantities. How does that happen?
Yeah. This is what we see as the next phase of the hardware renaissance or hardware revolution and that you can get to the prototype easier than ever before, but it’s still insanely hard with a lot of unknown unknowns going from one to many. That’s the area Dragon were specifically focusing on and see a tremendous opportunity.
Basically, what happens next is after you’ve got your working prototype, you feel good that it has the product market fit, and potentially, you’ve raised some money either through a successful crowd funding campaign, you’re part of a larger company doing a tangential product to your main product line, or you’ve raised some venture funds.
Then, the challenge is, how do you go from one to many? Typically, the next step is you have to pick a great contract manufacturer or factory to work with you. This is the most important thing that if you get this right, you can build a really strong foundation and create a successful company; but if you get it wrong, then it’s like death by a thousand cuts, and it’s very, very difficult to recover.
In looking out there, we haven’t found any good methods for how to select a great factory, which is why we’re building a suite of web apps to make that much more transparent and explain how the process works; and this is something we’ve been doing for years. Basically, what you’d want to do is go through a comprehensive RFQ or Request for Quote package, and you’d ideally want to have three to five factories bidding on it.
Now, the trick is, how do you find those great three to five factories? More often than not, you’re not going to find them doing a web search because it’s very difficult to know who’s good and who’s not good, and factories historically here are not very good at marketing or advertising. At Dragon, we’ve got a database of couple hundred factories we’ve worked with and are constantly expanding that. We down-select that to three to five factories. The reason you want three is that you have voting.
In a pricing line-by-line basis, you can see which factory is the outlier, either high or low. Whereas if you only have two, you don’t know … You don’t have a basis for comparison. You don’t know who’s off. Then typically, you want to keep five or under. Five lets one or two of them know bid, but still maintains your … The magic number of three. We find if you do more than five, you’re just not being selective, and you’re spreading your IP around further than you need to.
Then once you’ve basically picked the factories, then there’s a three-part process to creating the RFQ package. The first part is the typical text document which will explain the product, the company, the team, what’s done, what’s not done, and where you’re looking for help from the factory. This is incredibly important because the … Many of the factories that are the good ones are working with much bigger customers. As a startup, you have to show the factory the potential and the opportunity that they would … Why they should work with you versus somebody else from an opportunity-cost standpoint because they never make their money really until they’re shipping in volume, so that’s how you do that.
The second part is the bill of materials in a transparent and canonical format, so that you can line it up with other factory quotes, but you also have very, very good resolution into what … Pretty much every atom cost that’s going into your product or every molecule. Then, the third part is the schedule. We’ve got a proven process and tools that we built to create this, so it’s incredibly transparent, and there’s a lot of education that goes on.
Then once you’ve got that, you go visit the factories, develop … Start to figure out who’s the good one to work with, what’s the capability of the team, things like that. Then finally, you’ll come back and do the apples-to-apples comparison to understand the key cost drivers, and then how they line up based on your visit. Having gone through that process, then a company is in a great position to pick a factory.
We haven’t found anything that’s comprehensive out there that will do that in an open format because usually, what you find is middlemen that are connected with the factory and have a motivation to drive work into them, or people go on Alibaba or Global Sources, which can at times be okay for just finding a component, but it very rarely works well if you’re trying to find a factory that can do the final assembly.
Right. Is the same true of …? You talked a little bit about the Far East manufacturing. Is the same true for on-shore manufacturing as well for American factories? Is it the same process?
Yes. Yeah, the process would hold the … In the US … At Dragon, we’re always agnostic on where our customers build. The only thing we care is that they succeed. Because we build a lot of consumer electronics, China often makes sense; but if you’re doing lower volume, say under 5,000 units, as a rough guideline, the US makes tremendous sense. What we typically see if you contrast the US and China, in China, everything is very vertically integrated, so you’ve got the molding, the SMT for the circuit board, the quality testing, and the pack out all in one facility.
Whereas in the US, it tends to be more fragmented that you may work with a molding shop to do the injection molded parts, and then a different circuit board shop to put together your PCBAs, and then a different house to do the final assembly. You just structure the RFQ in a manner that’s conducive to that, but the process is exactly the same.
That’s interesting. We’ve talked a little bit about how a company can move from having this idea into prototype, and then maybe work with yourselves to get to the next level in production. What are you seeing on the larger company side of things in terms of companies that maybe already have an established product portfolio, and they’re seeing all this innovation happening at the grassroots level? How are they able to leverage what’s happening, whether it’s crowd-sourcing ideas or rapid prototyping that’s being creatively driven in smaller companies? How are larger companies fit into this product renaissance?
Yeah, so that’s really interesting. I’ve met with a bunch of them. In fact, I had a chance to meet with Jeff Immelt maybe a year or two ago. The biggest thing on his mind was, how do we get GE to go faster, like how do we get entrepreneur speed out of a company as big as GE? Just looking … For me, looking at a present or an initiative I did at iRobot maybe 10 or 15 years ago, I had a quote from Jack Welch saying the same thing like, “How do we get … In our big company body, how do we get the speed of an entrepreneur to innovate and stay on top of things?”
Across the board, every big company I’ve met with, their biggest concern is how do they innovate more quickly. It certainly a challenge because if you look at what it takes to move the needle for a big company versus a small one, it’s a tremendous amount of volume. When you do that, there’s a lot more risk that it’s very difficult to fail fast to succeed sooner. I think the opportunity for the big companies, and we’re seeing this quite a bit, is that they’re recognizing the importance of innovation and of going fast, and they’re finding ways to do smaller pilot programs where they can run 5,000 or 10,000 units to get a feel for the product market fit.
Then once they have that, then they have confidence that they can start scaling up, but these are volumes that they’re typically not used to building in, so it’s a totally different supply chain. The factory that you might pick to build 5 million, let’s say, ovens if you’re Sears is going to be different than the factory you’d pick if you wanted to build 5,000. It’s just the fundamental impedance mismatch between the two. At Dragon, we see huge opportunities to be able to help the big guys go faster and think much more like startups and entrepreneurs, but also be able … Once they get the signal that there is strength to that product line, then to be able to ramp up very, very quickly.
Interesting. I want to shift gears a little bit and talk about sustainability and environmental impact because one of the … I think one of the natural outcomes of there being a lot of new products imagined and then brought into the world is that there’s this entirety of the product lifecycle where you’re looking at it from birth to decline where you need to be concerned with what happens to the product after people are no longer using it regularly. What are the considerations for sustainability when we’re considering some of this emerging technology?
I almost see them as two important, but separate things. There certainly been some innovation around resins that are biodegradable and things like that. We have, at least in my experience, seen those enter the standard consumer electronic market; but where I think that the big opportunity is, is from a recycling standpoint to make products much easier to disassemble, and then separate into their core components, so pull out the circuit board, pull out the metal parts, and pull out the plastic parts, and then get each of those into their recycling streams.
To do that, it requires … If you think of the hierarchy of needs, the … Like usually for startups, step one is just get a product on the shelf that people want and start to get some volume. Once you’ve got some success or some health there, then spending the extra time to design in, because it does take more effort, the products for ease of disassembly is I think a huge opportunity. We’re starting very slowly, but we are starting to see a little bit of motion in that direction. I know MIT has some interesting things on that front, studying it.
Where I’d really love to see things go is a way to make the circuit boards easier to recover the components from. I’m sure you’ve seen all those like terrible videos of young kids desoldering components from a circuit board and heavy metals, and it’s just … It breaks your heart. Where I’d love to see it go is if there’s a way to design boards such that it was much easier to achieve that without having to refill the solder. I haven’t, at least in my experience, seen anything like that yet; but potentially, that’s the next frontier.
Yeah. Yeah, that makes a lot of sense. If you were giving advice to a designer/engineer who is just getting their feet wet just starting to learn about product design, what would your recommendations be for areas to focus on? I guess if you were in their shoes as this young product designer, where do you think a person’s focus should be?
Yeah, so that’s a great question. What we see and I think is pretty widely agreed upon is like design is such a critical component of consumer electronics. Usually, like my expertise is on the consumer electronic as opposed to other things. Finding a designer, or an engineer, or the combination of both of them that understands what it takes to have a great design, but also can lay on top of that, what it takes to make that design manufacturable because so often we see these beautiful designs that are just in no way, shape, or form manufacturable at the right cost.
Having somebody with the expertise to understand what does it mean to injection mold something and what are the constraints from draft angles, uniform wall thickness, ribs, parting lines, undercuts, shutoffs, and so on that can then create a design that still is incredibly compelling to the consumer, but also has a much easier transition without losing the key design features into manufacturing. I think somebody who figures that out would be very valuable.
Yeah, that’s … I have two boys myself, so I’m thinking about what the implications might be for their future careers even though they’re nowhere near that. I’ve noticed as I’ve talked to people about this topic that the language of agility around software and hardware seemed to be converging in some ways. What I mean by that is I’m a software guy, so I’m very familiar with scrum or with the practices around shipping software with an agile team, and I’m seeing parallels.
I don’t know if there’s any cross-pollination happening, but from your experience, are you seeing this convergence of … I guess they would be more of the project management techniques around moving rapidly to get a product into production?
Yeah, so we see … I think we do see the trend going in that direction. It certainly started with the very front of the product development timeline. I’d say from the napkin sketch up to the works-like-looks-like model. That has gotten incredibly compressed in … Just based on your first question, we’re talking about the $25,000 Roomba versus the $500 gizmo that you’re building that you can go very quickly to get something in a customer’s hands.
There’s a great saying from Jim Lynch, one of my colleagues at iRobot who was the co-inventor of Lego Mindstorms, that I’ve always remembered, and Jim always says like, “You got to build it to build it,” like you got to … You learn so much by those quick iteration cycles as opposed to try to make that perfect prototype with all the drafts and rounds, and have a tool ready.
I definitely see that on the beginning. It hasn’t caught on yet, but where I think there’s … At Dragon, certainly, we hope that we can play a key role in extending that for the second phase of the hardware revolution is on the manufacturing. The reason there is manufacturing … At least the nuts and bolts side really hasn’t changed that much. You’re still making these incredibly big, heavy steel molds to create injection molded parts. If you mess it up, steel is hard to change that there’s a larger penalty for error. You can’t just print a new part at least today.
What we’re actively working on is creating the software tools to enable you to move through the manufacturing process much more predictably with a lot better lighting or illumination on how the process works. Because of that, our feeling is that the customers want to know where they can take risks and where they shouldn’t take risks, and that will lead to efficiency. In terms of the agile and scrum, like I think that’s definitely happening from idea blank sheet of paper up to … Including crowd-funding to a large degree, and then up to working prototype. Yeah. The next frontier is going to be the manufacturing side.
Yeah, yeah. That’s going to be an interesting one to watch as additive fabrication for metal gets better and as … I don’t know. I’ve seen some pretty wild stuff on the additive fabrication side. I’m trying to keep up with that, but my bandwidth is limited in terms of time to research, but I know there’s interesting stuff coming.
Yeah. There definite is. Tooling typically, if you EDM it or CNC it, you’re looking at for a bigger part, six to eight weeks. Whereas with some sort of metal centering or 3D printing, you could do it much faster, and you can do shapes that … Or you can get cooling jackets in there in ways … In geometries that you just couldn’t machine, so it’s … I think it’s hugely enabling. The challenge is the surface finish on the 3D printed metal parts is not good enough yet, so you end up having to go and touch it with a tool, and polish it, and then by the time … Like again, it’s the 80-20 or the 90-10 problem that the last 10% takes 90% of the effort. Certainly, with time like as the resolutions get tighter and tighter, I am really excited about where 3D printing or tools will eventually end up.
Yeah, me too. Thanks so much for your time today, Scott.
Of course. This is awesome.