What are some of the ways that you’re seeing emerging tech disrupt the product design lifecycle?
I think the most disruptive element of the product design life cycle is crowd funding, at this point. Just being able to prove that this market exists, before you spend all the resources getting your manufacturing line up and running. We’re seeing like a huge, huge huge drop in the barrier to entry, from crowd funding, because it’s just becoming the status quo, it’s like, “If I have an idea I can build it quickly and then I can get it out there” and things like Kickstarter do a really good job of helping you market it first and getting people to look at it.
Instead of having the traditional way, do some R&D to what people want, do some prototyping show up your pieces, figure out what parts you want go get quotes right for all those parts from people from China, find someone who can build it in China. Instead of all these infrastructure that you build a couple of million of them, hopefully you have some contracts with large retailers that will be willing to feature your product and then you put it out there and see how it goes. That’s just so much risk and so much investment. Nowadays there are so many really, really, good prototyping technology like Tessel, Arduino, Raspberry Pi, Picoboard, that it takes very little resources to get up and running and to buy different sensors and peripherals from Sparkfun and Adafruit.
If you want to build a prototype, your first degree of approximation, and get feedback from friends and once that’s good enough you should obviously spend some time figuring out about how much it’s going to cost you before you bring it to market through Kickstarter, but that time to market is drastically reduced. I’ve seen three different websites that are offering to make finding fabricators and suppliers easier more like Yelp for manufacturing. What comes to mind is tindie.com/biz I think and they’re sort of leading this effort. There are a couple of others I forget the name of and the fact that that whole area of product development is becoming less opaque is really critical because now as we’re building Tessel we came from a web development background and we just want it to be able to make hardware, at the same sort of iteration speed that we make software.
Obviously it’s not going to be entirely possible because there’s shipping physical goods involved in that, but we can, there’s a lot of room for improvement. What we found was that we could get pretty far just using off-the-shelf parts, but then as soon as you jump into the world of custom boards and building these things yourselves, it’s totally opaque because a lot of people want to keep that information secret because it’s a competitive advantage to know where you can buy the cheapest boards from, where you can find the cheapest factories.
Even factories in the Far East will treat Westerners different from people from China, so the whole process of how do I find a brokerage, how to get me a better price, all of this has been left opaque by the monoliths who already do part of manufacturing. We’re seeing, as this hardware renaissance approaches, we’re seeing a lot of people wanting to shine some light in those spaces and making them a lot more accessible.
There’s still a long way to go to make it easy for people to know how to say “Okay, I started prototyping this on a Tessel or whatever, now I need to make ten thousand of them. How do I make sure that this is cost-effective and the best design?” It’s a long way to go but there’s definitely some improvements being made, and then that type of a machine we’ve also been doing some a little bit of research into as a man who knows offer really well, how can I use software to highlight what sort of the optimization paths are for when I’m done with hardware? If I’ve written a program that’s 3 Megabytes in size, and I’m using a development kit that has 32 MB of RAM flash, well, can I move to a 4MB chip, and if I can, what are the cheapest sources I can find online for that chip, and how much are available on quantity, what does the end of life look like for that chip?
Using more software to do some of the mundane tasks of finding the right part that you’re looking for, and eventually being able to do things like stitch together schematics so that people don’t have to spend the time doing these super mundane but tedious tasks of connecting one pin from one part to another pin and another part, when it’s already outlined in your head and in their codes, essentially. We’re doing some research into that as well but it’s very, very far from completion or being available for use by other people.
Right. That’s interesting because one of the themes that is cropping up as I’m doing interviews with designers and engineers in this space is that there is some convergence between the attitudes and methods of software development that are being sort of cross-pollinated with hardware and device design engineering. You’re getting some of the agility that you would expect from software and sort of a product that you can iterate on, combined with the hardware elements which traditionally had not iterated quite as much, and their product life cycles are usually much longer. Clearly you’re seeing some of that yourself?
Yes, I think it’s slowly been building up for awhile now. If you look at hardware cards themselves, ten, fifteen years ago a lot of it was just a bunch of passives, capacitors, conductors, resistors, of course there were ICs and things like that, but lot of them were just built up from these fundamental building blocks. But now we’re getting things like, essentially the equivalent of a software module where I just buy this accelerometer and I put it on my board and route a couple of lines to it but those are essentially the interface, just like software is supposed to be an interface. It’s been converging for its being as close to software as possible, and I think that it’s really starting to accelerate now.
Yeah, that’s completely fascinating to me. Are you aware of any … if you’re a software developer there are some preferred methods for attacking problems, especially if you’re an Agilista, right? You might practice Scrum or some flavor of that. Are there similar movements that seem afoot in the hardware world, or the hardware-software hybrid world?
You’re talking about project management techniques let’s take into account hardware-software hybrid. I don’t think there are any formal processes. Some of it we’ve been doing in-house, that we brought on course a couple of months ago, is essentially designing the hardware with our design tools, and submitting it to be manufactured, and while it’s being manufactured, we write all the software tests that need to be made for the operation of the board. When the students and the board of faculty meet a week later we’ve all been working hard on making sure that it can be tested, and it can be tested right away, proving whether or not it works. If it doesn’t work then we can iterate very quickly, make changes in one to two days and send it back out. That’s something we’ve been doing but I haven’t quite seen some huge movement around project management techniques around hardware designs.
I would think, I mean, because that’s one of the, I mean, depending on your point of view that that could be one of the strengths of software development is that you have these various techniques and tactics for organizing the way you’re addressing pieces of the problem, as opposed to the waterfall approach that made software development tedious and take a long time. I almost feel like, something like that, I mean, I have no idea what I’m talking about here, but I feel that something like that is coming, or at least people must be thinking along those lines, if they’re trying on software development methodologies.
Yeah. Another thing that’s common, or becoming more common I think, I don’t really think this has been the status quo, forever I’m not sure because I only started doing this about three or four years ago, it’s buying a development board that is fairly close to what you want your custom part would be. Back when we started building Tessel, it was based on a specific microcontroller, LPC1830. What we did was we found a development kit from the same manufacturer of that microcontroller with that microcontroller and external memory and accelerometer, and essentially it was more than we needed, but it was a very good replication of something we buy off-the-shelf and start developing the software while the custom hardware was being made in tandem using that development board as sort of the reference design. At that stage we’re able to continue development of the software which is really important as we work out kinks in our custom hardware.
That’s a good metaphor. And I’ll send you a link to a blog that I wrote that eventually has all the tips that I’ve learned from how to move as fast as you can in a hardware company when you come from a software background. It mostly just covers testing and development, using references like sort of the things that we’re talking about here.
I want to dig a little bit more on that. You mentioned that you’ve been doing this portion of your career for about three years. Are there a lot of engineers who are being attracted to this space based on the fact that it is easier to produce products now … In other words, are you a representative of a much larger group, like an engineering surge, into the product design space?
Yeah, absolutely. I think that we’re just starting to see the beginnings of this movement. I think there’s a couple of factors leading into it. Number one, software is extremely, extremely saturated right now, so any new idea you come up with for a software product it’s going to be made already. Probably made three or four times it’s frustrating to people, I mean it was frustrating to me. Back when I was in college I’d have a lot of list of things I want to do and they were made, and I got so tired of being constrained to either the web or the phone that I wanted to learn how to build my own hardware. I started taking classes in that. It’s very, very, freeing to say that “Okay I have an idea, and it doesn’t matter what form it takes, like I don’t have to build for the web, or I don’t have to build for the phone. I could just make sure that I build the best possible user experience, whatever form that takes, and deliver that to the market.”
Being able to have that freedom of expression and deliver a product quickly is pretty important. The other factor that’s driving it that we talked about already is the fact that the barrier to entry has just dropped enormously, the fact that anyone can pick up a microcontroller and sensors and get going right away and it’s becoming easier and easier to connect to the internet. So we’re seeing, in terms of IOT, there’s a bunch of what we call “layers to the stack” that are being filled. Like the hardware layer, the networking layer, the communication layer, the application layer and there’s different companies that have been jumping up to fill in these layers, so that any startup that jumps up in the hardware space doesn’t necessarily have to fill alternate roles.
As a hardware startup these days, you don’t just have to build hardware, you have to have, hardware-ready you’ve got to have a global app, you’ve got to have a website, you’ve got to have a back-end. There’s always different pieces to the puzzle which seem really intimidating at first, but the fact that all these tools have jumped up and made it really easy to integrate them together means that you just have to start with what’s like your main box, or what you’re starting with, and you can put these pieces together right away.
Maybe eventually you optimize out these specific services so that you can pay less or have it better tuned to your needs, but it’s just become so easy to get it all together. And arguably the only thing missing from IoT stack right now is a security layer, which nobody really seems to be keen on jumping on.
Wow. That’s really interesting because, I mean there’s … This might be a separate topic altogether, but from what I can tell there aren’t a lot of open standards in the same way for IoT as there were for the web at its inception, right? In terms of, at least, I don’t know if this is one of the layers of your stack, but the communication. There’s definitely competing methods for communication with IoT, is that a limiting factor right now, or do you think that’s going to shake out and there will be some dominant standards going forward whether they’re open or not?
Sorry just want to make sure I understand your question, you’re asking about, like when the web started seems like there was a bunch of agreed-upon standards, but right with IoT there doesn’t seem to be?
Yeah. I wasn’t really around when the web was starting, but I can definitely agree that there are a lot of competing standards right now. In terms of, I think what’s going to win out is what is going to be open. That’s a big leap of where all the hardware and software have been open sourced. The thing that everybody is using that’s free to use and gets audited often for security issues is a thing that’s going to be open-sourced, and I think that’s what’s going to be ultimately adopted. I think it will take some time for sure but, and figuring out how to operate between devices is, if a device has an open source API, it’s going to be a lot easier to connect to it. Once more devices have this open-source APIs with a really heavy-duty security halo on top of it which hopefully is also open-source, I think there’s a lot more collaboration between devices.
Right. That’s, part of the promise is that there’s the device-level product but then there’s also the network, really as a product in and of itself. Probably a little silly but your home network with all the pieces that play in it, or on a grander scale, you have your smart cities or connected hospitals , connected health rooms, and things like that, where devices are able to talk to each other and transmit data etcetera. I’d like to talk to you a little about scale, because it seems to be part of your, I mean it is part of your business model with Tessel and it also speaks to the difficulty of taking a product from prototype to getting some marketed option, and once you pilot it and then getting into the thousands and tens of thousands in terms of production, and getting that kind of growth. How do you see, or maybe let’s start with Tessel, and how that fits into a scale up to the production process.
Yeah, absolutely. So what we found with our customers is that they, very often, after they prototype something on Tessel, wanted to go to production and really weren’t sure how. Because it isn’t an okay process, it’s what I was taught out before. So we’re trying to make it a more automated process and something that’s more like a service, like you have software services all the time, or a website, to build something that was something, something similar to that. What we found with existing products was that they just weren’t designed for that scaling. As an example with Raspberry Pi I mean it was designed to be a learning tool. And it’s made it really high volume, but if you want to, if you prototype something in that and you want to go to production, you’re really going to have a very, very hard time sourcing the Broadcom chip, because Broadcom generally only sells in really high, tens or hundreds of thousands of units. They’re not really going to listen to a small person starting their company.
Similarly VGA board nobody really knows how but it has extremely low prices on all their parts so if you wanted to base a design out of VGA board you might end up being more expensive than the original VGA board but with less functionality. Actually we see a lot of people embed VGA board right into their products, even though it’s relatively expensive at forty dollars. We wanted to help people sort of take that next step, and its interesting because if you look at Kickstarter or any other company of that size usually on average the first run of any product is going to be a couple hundred to a couple thousands maybe up to ten thousand. Hopefully as the company grows their manufacturing sites will get larger but as is generally well, the first batches are for the first early adopters, and the interesting thing about that volume is that you’re surfing this weird place where it’s very possible that it makes less sense to build your own custom hardware rather than using an off-the-shelf part.
For example with Tessel, if we’re building in tens of thousands, and you want to build something in quantity one thousand, if you use off-the-shelf parts you’re going to get into the economies of scale of ten thousand rather than you buying pieces yourself at one thousand. It’s this weird middle ground that hasn’t really been served very well of “How do we apply the economies of scale from something like an off-the-shelf part but then allow some pseudo-customization per whatever the specific needs of that customer is?”
With Tessel 2 one thing that we are trying out is, if people do you use modules that they plug into the ten-pin ports on Tessel, maybe we can just automatically integrate those modules into a single PCB so you don’t have the ten pin header anymore, looks more professional, it doesn’t wiggle or shakes and things like that.
It looks like a professional offering. Or if they’re not using the Ethernet or USB ports, some of the ten-pin module ports, let’s just take those ports off and save the money on their bill of materials. That’s relatively easy to do. We’re trying to find these creative ways to do pseudo-customization possible at this median-level scale for people who are trying to build products.
Right. That seems that a market niche that needs to be served. How’s that going?
It’s hard to say because there are definitely many people who are interested in it, about thirty percent of the people who’ve built with Tessel 2 are interested in building a product. I think the fact that we are explicitly saying “Hey we’re not just a development board, we’re a path from development into production,” has caught a lot of people’s eyes.
That being said it’s about a hundred thirty, a hundred fifty people who are trying it out. I don’t know how large the market is yet, maybe once we go through the process with a few different people to prove that the technology is viable and it’s some sketchy, weird website, where you don’t know what you’re going to get.
Maybe we’ll get more people interested, I think it’s been inconclusive the results whether or not it’s something that will catch on, but once we try it on and give it more time we might be able to say “Yes, this is something that a lot would want,” or “No, they do not like this, they’d just rather have the ten-pin headers and everything populated,” then don’t want to go through that process.
Right. I want to shift gears a little bit because we’ve been talking a lot about, sort of the prototyping and sort of the early stages of getting a product to market and finding an audience for it and then getting that to production. I’m wondering what your perspective is on large companies that really do need to innovate and see all these activity at the grassroots level where people are creating new things really quickly with products like Tessel or are 3D-printing their cases for things, and creating all these incredible, innovative stuff, but at the same time, you’ve got your GEs and your Samsungs of the world, that could very much take advantage of some of those new ideas that are being brought to life at the grassroots level. In your experience so far, have you seen … what’s the interaction between these smaller, very innovative startups, and larger companies that definitely need to innovate, but may not have the cultural capacity, or are looking to acquire startups.
There are two sorts of interactions between these large companies and startups like us that I’ve seen. The first is that, all of these campaigns like Microsoft, SAP, have software services that they’re trying to sell, and they see companies like us as “the new cool thing.” They’re actually some of our biggest customers because they buy Tessels to attract people to their Hackathons and because it’s the fastest way to connect data up into their product services. That’s one thing. They definitely have an eye out for like what developers are trending towards and trying to integrate those tools into the other services that they’re already offering. The other thing is, companies who want to make a compelling service or product and don’t have it already set up and we see a little of that as well. It’s really up to these big companies who’s mantra is “If it’s not built here it’s not great.”
All the time they’ll just spend millions of dollars investing in their engineers building everything from scratch, and after a year they get frustrated and they’re like, “Well you guys have accomplished very little and why don’t we just build it off something that’s existing and they think, more so recently that the core was saying to all these big companies come to us and say “All right well, we want to use Tessel, how will we use it internally and build out on top of it?”
There’s been a massive, massive uptake from big companies using smaller setup tools to build out their products, yet, I think a lot of that is trepidation of the fact that “Well if this company gets bought, what happens to what we’re in the middle of,” and being open-source certainly appeases some of those fears but I think it still exists. I’m really hoping that more and more big companies will start to use these tools, they’re specifically developed to help anyone move faster, not just the individual entrepreneurs who move fast and beat these larger guys to market but anyone can use this to build faster by iterating faster, you become a more innovative company.