 Hey folks, this is Joshua Montgomery, the CEO of MyCraft. Why we started MyCraft and where the name and the idea for the overall thing came from. And maybe a little bit about the rest of the company's history. So why did we start MyCraft? So I am a service provider in Lawrence, Kansas called Wicked Broadband. It's one of the few internet services nationwide that has fully embraced the concept of net neutrality. The idea that you shouldn't have prioritized traffic on a network, that everybody should have equal access to the network, that you shouldn't be able to charge websites, speeds in their traffic. I've been doing that for a really long time. About three, four years ago I started Makerspace in my local community with the idea that we could help other entrepreneurs to achieve their dreams. And that Makerspace includes a co-working space, a complete Makerspace with a wood shop and a metal shop and PCB fabrication facilities, 3D printers, laser cutters, all those fun tools, as well as a data center for the ISP where we switch traffic for thousands and thousands of members of our local community. And when I was building this space, I was inspired by a Makerspace in Kansas City run by Dave Dalton called Hammerspace, where Dave had built a voice assistant into his space called Iris, where he could sit behind the cash register and hit buttons and it would make announcements and he could type stuff into the computer and make it say stuff in his Makerspace. And I saw that and I said, you know, we can do one better. I bet we can do Jarvis from Ironman. And so I went out and did some homework to see if we could take technology like Siri and customize it to our needs. At the time, Amazon Echo was in closed beta, so we had no idea that the Echo was a product or a service that was available. Google Assistant didn't allow you to customize any of these experiences. And the open source projects that were out there, Jasper is one of them, and then Stephen Hickson out of, I think it's Georgia Tech, had a little voice assistant that he'd been playing with in his spare time. But neither one of those technologies really met our needs. And as I did the homework into the tech, I realized that all of the underlying technology stacks were beginning to reach the point where we could build a voice assistant that had a user experience that was indistinguishable from speaking to a human. And let me reiterate that. Speech recognition is now achieving accuracy ratings that are higher than a human can achieve. Natural language understanding engines based on machine learning and based on known entities rules are reaching a level of accuracy for intent that make it extremely easy for the computer to understand what the user wants. And speech synthesis is now reaching the point where the speech that's rendered by the computer is of high enough quality that people have trouble determining that it's a computer rendering the speech. And those are all trends that have accelerated since 2015 when I had this concept, but at the time it was very clear that that's where the technology was headed. And I've always been a huge fan, a huge fan of Robert Heinlein. My dad started me reading books like Farmer in the Sky and Tunnel in the Sky, kind of Heinlein's Junior novels when I was a very young kid. And I've always been a huge fan of his three major novels, Stranger in a Strange Land is one of them, Starship Troopers, not the movie, the book, Fantastic. And then the third book, The Moon is a Harsh Mistress, where he explores political philosophy and the idea of revolution and self-direction. And in that The Moon is a Harsh Mistress, there's a character called Mycroft. And Mycroft is a voice assistant, a computer system that controls all of the controls on the lunar colony. So it controls stuff like atmosphere and temperature. It controls the telecommunication system. It controls the transportation system. And it controls the induction catapult that they use for transporting goods and services off Luna. And the character Mycroft in this book does a number of amazing things. Number one, it speaks English. Number two, it speaks English through the phone systems. You can call Mycroft and have a conversation. On in the book, Mycroft is even able to create a CGI image of itself and be viewable via video chat and video conference, which as far as I know, and I could be wrong, but as far as I know, it's the first instance in science fiction with regard to photorealistic rendering. And so I've always been, and I had the concept that the voice assistant we would build could be similar to Mycroft. And Mycroft in the book is a character that is fully self-aware. It is equivalent to a human intelligence above a human intelligence in a lot of ways. And throughout interact with Mycroft on the phone, they're unaware that they're, and so it passes the Turing test. And so we decided to build this voice assistant. We decided to build this AI entity that inhabit our space. And we decided to name it after Mycroft for obvious reasons. And Mycroft, by the way, I should point out as a protagonist in the book. In case this is in science fiction, is not some kind of nebulous threat to humanity, but is actually an asset that does good things throughout the novel. So we decided to build this technology, and I have a really, really deep commitment to openness. I believe that ideas can be owned. The concept of intellectual property to me is anathema. And I believe that computer software, because the cost of replicating the software is effectively free, should be as close to free as possible. But that doesn't mean that you can't build successful businesses around it. It just means that the budget should be open and that people should have access to it. So think of free as an access, not free as in no cost, right? So we decided to start the Mycroft project as an open source community. And the piece of the market that we wanted was there who wanted to use this type of technology, but for whatever reason didn't want to be locked into, at the time, the Apple ecosystem. And we've now had two more ecosystems stand up, one from Google and one from Amazon. And that the technology would be open so that people could use it, customize it, change it, hack it, do whatever that they wanted to do with the technology. So we began working on this, because it's a makerspace, there's Raspberry Pi's everywhere. And so we took a Raspberry Pi and we started working in Raspbian. And building originally based on Stephen Hickson's code was my implementation. And building really simple exact match skills. So at one point we had it set up so that if you ended any sentence with the word Deadpool, it would spin up the YouTube on the Chromecast on the TV in the conference room and play the Deadpool trailer. Which of other little hacks and gimmicks built into the original technology stack that would allow me to do a demo of what it would eventually do without building a lot of the underlying core technologies. And in working on it, we found ourselves with a Raspberry Pi on our desk along with a keyboard and a microphone and a mouse and a power strip and a bunch of cables interconnecting everything and an ethernet cable was this giant Medusa mess of cables inhabiting our desks. And so I went back into the back room and I developed a little hockey puck size, not size, hockey puck shape. It was actually bigger than a hockey puck about eight inches across. No, six inches across because otherwise it wouldn't fit on the 3D printer. So six inch across disk where we could pack all of these electronics into the disk and it was set up to be powered over ethernet, had a little off-the-shelf speaker and an off-the-shelf microphone in it. And we were working, I reduced it to that and put it all in this little case so that I could work on it on my desk without having a big mess. And at the time two things were happening. Number one, we had a lot of entrepreneurs coming into the makerspace and saying, I want to start a company by doing a Kickstarter and we'd never done a Kickstarter. So we didn't really know how to do that or what that would entail. And at the same time I was looking at this device on my desk and thinking, you know, I wonder if anybody wants to buy a smart speaker. And today three years later, you know, that seems like a dumb question. They're going to sell 56 million smart speakers this year for a total market size of more than $6 billion. But in 2015, it was still an open question. You know, for many people, what is a smart speaker? Like it sits there and you talk like, I used to have all these super smart VCs that were, you know, would sit down with me in a room and I'd explain to them what we were doing, you know, you'd talk to it and it takes action on your behalf. It's kind of an augmented reality audio overlay over the room that allows you to access internet services using nothing but your voice. And they'd look at me and they'd be, oh, nobody's ever going to want to buy that. Like I'm Mr. Super Smart VC and, you know, we want to back SAS companies that build an API that helps with middleware and some vertical application space. Sorry, thank you for your time. And so, you know, we had vision beyond that. But as entrepreneurs who are more experienced maybe than other entrepreneurs, we wanted to validate our assumption. And so we took this puck and we decided this hockey puck size thing and we decided to go to Kickstarter. And being an entrepreneur, I went out and did my homework. So I went around the community and said anybody run a Kickstarter before and one of my ISP customers, Earl Shweppy, who had been a instructor in computer science going all the way back to the 1950s. Like World War II vet worked for the National Security Agency in the late 40s and early 50s when Russia was the giant threat, went on to teach at the University of Kansas for a good 30 years. And when we were doing underground fiber work at one of our sites, Earl used to come out and come chat us up. He's an older gentleman. And Earl mentioned, he said, my grandson, Derek Shweppy, had done a successful Kickstarter. So we connected with Derek, who's now our, as many of you know, our Chief Design Officer. And Derek is a design instructor at the University of Kansas. And so we took the idea to Derek, who'd run the largest Kickstarter ever to come out of Kansas. He was like a 55, $56,000 Kickstarter for an IoT switch. And we said, hey, Derek, can you give us some advice and some input? And we ended up bringing him onto the team as our designer. And between me and Derek, we went through rapid iterations on the little hockey puck thing until we ended up with something that's effectively Mark 1. So a disk space speaker that's big enough to put a Raspberry Pi 3 in with a pretty reasonably high quality speaker. We learned a lot about speaker volume that will be applied to the Mark 2 and the Mark 1A, which will be out next year, to improve the sound quality. And I wanted to make the technology anthropomorphic. You know, one of the things that happens when I interact with people and when you interact with people is facial expressions and body language really make a big difference in how things are perceived and how you interact with technology. And what we didn't want to build was an anonymous black cube that sat in the corner and represented the Borg, effectively, right? Represented the giant data collectives that they're building out here in lovely Silicon Valley. We wanted to build something that had personality and we wanted to be inspired by the very best of science fiction. So characters like ET, characters like Eva from WALL-E, characters like WALL-E himself, like Johnny Five, characters that are perceived by the public and perceived by technologists as positive and funny and humorous, not something that's Borg-like and anonymous and doesn't have any personality. So we developed the Mark 1, which has the two cute little eyes and the little line of LEDs between it. We took it to Kickstarter and lo and behold, we ran the biggest Kickstarter ever to come out of the state of Kansas. And once we did that, we realized that if, you know, more than a thousand backers around the world were willing to take a bet on this anonymous group of people in Lawrence, flipping Kansas, that they could build a smart speaker and ship it and there was enough demand for the technology there, there probably was and there turns out to be a much, much, much larger market for the technology. So we hired some folks to come in and help build the tech. One of our early hires was Sean Fitzgerald, who was one of our Kickstarter backers and Sean was the first technical hire on the Siri team. He was the second engineer assigned to the Amazon Echo Project at Amazon. And so he had built this type of technology twice and brought that experience to the team. And Sean had already been working on a library for natural language understanding or intent parsing called ADAPT, which is how we do a lot of the intent parsing in Minecraft now. We also now have a machine learning one called Pedacious. And actually ADAPT is being pulled into Mozilla's IOT efforts as Mozilla's intent parser and NLU engine. So, you know, Sean's work is still growing even today. And Sean, you know, looked at the code that I had built based on Stephen Hickson's code and he kind of chuckled to himself and shrugged his shoulders and archived it and started over, which is probably a good decision being as I'm an aerospace engineer and not a software development professional. And we brought a couple of other really smart people on Jonathan Dorlands who led our back-end team for a couple years, as well as Mateus Lima, who still works on our back-end team and Augusto Montanero, both of whom are out of Fort Elyse in Brazil. And one of the reasons we made international hires is we couldn't afford full-time developers here in the United States, so we reached out to find people who were highly qualified overseas and were inside our budget and found some really inspired and hard-working folks in Brazil, which is where Mateus and Augusto and Jonathan all came from. And we began building this technology. And as part of Kickstarter effort, one of the things we didn't know, but we certainly knew before we ran the second one is that Kickstarter generates a ton of earned media, meaning that the magazines like Forbes, Popular Science, in our case, Inc. Magazine, Entrepreneur, and others cover Kickstarter campaigns that are going well. And the combination of our thing being cute and today being in the voice space, which is now a huge fast-moving space, really drove a lot of interest. And one of the magazines that covered us was the Silicon Prairie News, which I think it may be defunct. I'm not 100% certain, but it was an online publication that covered the startup scene in Omaha and Kansas City right there in the center of the country. And John Fine, who now runs Firebrand Venture Capital out of Kansas City, and at the time was the managing director of the Sprint Accelerator, which is powered by Techstars, saw that article and had a mandate from the local community to find some companies to join the Sprint Accelerator that were local. The prior two classes of the Sprint Accelerator didn't include a lot of companies that had come out of the Kansas City area. And there were a lot of folks in Kansas City that really believed that if Sprint was running an accelerator and it was in the local community, that there should be some local companies involved. And so John reached out and said, hey, would you like to come and join Techstars? And I hung up on him politely. I went to my team and said, hey, these Techstars guys are calling us. And it seems like a giant distraction. Like we're busy executing, right? And my team went out and did their homework and they said, Josh, like Techstars is one of the top accelerators globally. And they're going to give us $120,000 and provide us with introductions to all these really great people. You should really reconsider. And we had a conversation about it and we decided as a team, it's actually one of the few times in the entire project that I've been overruled by the team. I really didn't want to do it. And my team came back and said, hey, we need to make this happen. And so we made a decision as a team to do it. And like any good team, once the decisions made, we executed. And so our team started spending every Wednesday at the Sprint accelerator, getting to know the Techstars team. We kind of cyber-stocked them a little bit. Like they would show up to do presentations in the local community. And magically, there'd be a member of the Minecraft team there to clap and cheer and egg them on. We really made a big effort to get to know those folks. We went and won a startup weekend competition in Kansas City just to show them that we could execute. And we got invited to join the Techstars team in November of 2016. And that's really where we made a decision to not be an open-source project that's run out of a basement in Lawrence, Kansas, a basement data center in Lawrence, Kansas, but to be a venture-backed technology startup like MySQL or Mongo or dozens of other open-source companies that were venture-backed and have done quite well. So we went through Techstars and we got introductions to all kinds of really fabulous people both in the local community and globally. And we wrapped up Techstars in May of that year. Really, the big rock, the big thing that happened during Techstars is, number one, we met Steve Penrod, who is now our CTO and our co-founder. And number two, we executed an agreement with the Ubuntu team to package Minecraft for Ubuntu, something that's still in process two years later. And then we went out and we raised some pretty significant money. After Techstars, we ended up closing a $350,000 angel round which gave us the money to start hiring additional folks here in the United States and globally to really begin moving the project forward. After that, we took an investment from Jaguar Land Rover who had a vested interest in improving voice technology for the car. And based on the strength of the project and the velocity of the sales growth, we got invited to join 500 startups. We also began shipping our original 3D printed POC Mark I devices. So I think we ended up making about 50 of those and we actually 3D printed the enclosures. We hand soldered and hand fixed the cases. They were kind of a mess, honestly. Folks, if you have one of those sitting on your desk and you'd like to exchange it for production Mark I, please contact me and I will make that happen. And so those shipped really early and started getting us feedback on what this thing would look like at the end of the day. And shortly after we wrapped up Techstars in May of last year, we ended up shipping the other, it's almost 2,000 Mark I's to folks all over the world. And that really began to drive momentum into the project. We opened a investment round in actually late 2016. We opened a series seed and took our started taking money then and we performed a rolling close all the way through 2017 and into early 2018. And for those of you who don't know, a rolling close is where you set the terms of an investment, but instead of getting all the commitments from everybody and circling them all up into this one high stakes, like it's almost like a mortgage closing. What you do is you create the documents and you take money as people are interested. So although we closed a series seed that's actually just a hair over $2 million, we took the first check in that series seed in November of 2018. So that $2 million, although it seems like a lot of money, it didn't all just show up five months ago. It started showing up nearly two years ago and we've been using that money to grow the company throughout that entire time. And so the series seed went fairly well and we ended up with a bunch of institutional investors as part of that. Folks who invest in seed stage companies with the idea that they can get significant multiples on their investment by getting in early. We got money from Social Starts, which is a fantastic venture capital firm out of the Bay Area that makes early investments in the seed round and they also have a series A fund that helps on down the line. We took an investment from Deep Space Ventures out of Dallas, Dallas, Texas. That connection came through through my Air Force service. I'm a reserve officer and met some folks in the venture capital community who specialize in working with members of the military and that's how I connected with Stephen Hayes who's an Air Force Academy graduate and runs the Deep Space Ventures team. We took money from T2 Capital out of Bogota, Colombia. We took money from Kika Capital out of Tokyo. We took money from Technexus Venture Collaborative up in Chicago and a variety of angels and early stage investors from all over the world. We have investors as far away as the Greek islands going east here in the United States and as far west as Singapore. So we have investors from all over the world and throughout this investment process, the software team was busy executing. So when we originally built the Minecraft stack, what we ended up doing was using as best we could existing open source technologies to solve the various problems that need to be solved within the voice assistant. We also used the CmU technology to help wakeward spotting. So spotting the phrase, hey, Minecraft, and activating the listener. Initially, we used PocketSphinx which was an open source package developed by CmU. It's not a community developed technology. It was a technology that was developed inside CmU and they periodically tossed the source code over the wall for other people to use. We took that, cleaned it up, did some work, partnered with Vocality out of Boston to help a community manager at Ubuntu donated the samples from which the original voice was built and we did the best that we could with the speech synthesis using that piece of tech, using, oh, that's the, sorry, that's the speech rendering engine. The PocketSphinx is what did the wakeward spotting. And so the wakeward spotting software was spotty. I mean, we're doing it in software, not in hardware. We're not using fancy microphone arrays and on chip processing the way that some of the big companies in the valley were using. For intent parsing, we used ADAPT which originally was known entity rules-based. You know, Microsoft Core was really what tied together our original skills framework which was spotty to say the least to begin with and then speech rendering was using Mimic which was actually Festival Light, I apologize, which was also developed at CmU and once again is tossed over the fence periodically there as a open source package. And so, you know, what the software team has been doing for the last two years is either advancing the state of the art of those technologies or replacing them with more modern approaches. So, you know, PocketSphinx, the original technology wasn't very accurate and as a result, it got a lot of inadvertent activations or more importantly it missed a lot of activations especially if it was a kid or a female, it had a lot of trouble. So we developed a new one based on machine learning called precise because it's this precise wakeward spotter and we started collecting data from all of these devices where users had explicitly opted in to share their data and started using that to build a machine learning model. You can actually help with precise by going to home.microft.ai, creating account logging in, hit tagging, hit the precise link and you can actually help us to determine whether somebody said hey, Minecraft or not which improves the accuracy of the engine. The precise wakeward spotter is much, much, much more accurate than the original and so that's some work that we've been doing as a software team. In terms of the NLU engine, ADAPT was a Node entity rules based engine so it was really good at things like turn the lights on in the kitchen, lights are an object on as a toggle position, kitchen's a location. It knows based on its data set that that's probably an IoT call, it calls the IoT skill and it sends the appropriate command. It sends the correct hook where there are no known entities and so our machine learning team developed a technology called Pedacious which is able to absorb large quantities of data related to intent and make determinations based on a machine learning approach. Minecraft Core has obviously evolved significantly in terms of its ability to deal with skills, its ability to abstract the various different functions in the technology stack, pairing data access back to our back end and then of course our speech synthesis engine. Mimic is still the production engine. We have a new version of that called Mimic 2 which is based on a machine learning model. We wanted to have those voices wrapped in April but we actually ran into a problem with data acquisition. We couldn't get the data set built as quickly as we would have liked to have for dumb reasons related to not having access to an appropriate studio and sound stage and a couple other pieces. In Resolve we now have a complete data set and as far as I know that should be out by the end of June in production. The first voice that will be available there using that engine is most likely to be an American male. However we will continue to expand into other accents and other languages over time. And so the software development team has also been super super busy and the future looks bright. We ran the second Kickstarter early this year and we ended up taking that between Kickstarter and Indiegogo to more than $450,000 in pre-orders. Those will ship on December 4th of this year at 5pm central time. So those are going to ship on December 4th at 5pm and they should be available for Christmas which I'm super excited about. And we've had a lot of engagements with some big corporations. Jaguar Land Rover made an investment in a company in early 2017 and our team spent a significant amount of time in Portland working with their enterprise team, their technology development team to do an integration into the F type. There's some video of Minecraft running in the F types dash. Obviously development plans at automakers are super secret so that's as much as I can really say about that. There's been a number of other enterprise companies that have reached out to us. We executed an agreement with Hewitt Packard towards the end of last year and had an excellent experience working with that team there. And there are a lot of other opportunities within the enterprise space and a lot of companies that are reaching out to us because while we were doing all this and trying to achieve our vision on AI that runs anywhere and interacts exactly like a person it turns out that smart speakers are wanted that those VCs who told me nobody would ever buy them were wrong. The way VCs are often often wrong. Jucero anybody? And so we were off executing and trying to achieve our vision which is an AI that runs anywhere and interacts exactly like a person. When you talk to it you can't tell you're talking to a computer that is what we are building. And we wanted to run anywhere. Watches, mobile, desktop smart speakers, automotive any aircraft anywhere people want to run it. That's what we're building as a community so we've been off doing that. And in the meantime smart speakers have become the fastest growing technology in the history of mankind. Let me say that again. Smart speakers and voice technology are the fastest growing technology in the history of mankind they are growing faster than fire. They are growing faster than the last fastest growing piece of technology which was smartphones. And so it's really important to realize that us as a company we've always had this goal irrespective of what the market was saying. It's just been validated by tens of millions of people out there who think that this technology is important and if they're doing business with us they think it's important also for that technology to be open and available and private that they want to have control of their data, they want to have control of their their online presence they want it to do what they want it to do they want to have access to user agency and so those are the folks that are engaged with us. We've got a huge and growing community surrounding the technology. We have a number of corporates that are engaged with us at a variety of different levels we've got a great smart speaker on the market for makers and hackers the Mark 1 is not for the general public it is for makers and hackers. It was designed that way. However the Mark 2 is on its way and will deliver on December 4th at 5 p.m. central that first shipment of Mark 2's will go out. And that one is being designed for the general public. We also have a Mark 3 on the horizon which is a voice assistant with a much larger screen and a bigger amp that's designed to go in a kitchen environment primarily although it would probably be fine in a desk or bedroom. The Mark 3 will ship with 3 of our Mark 1 A's the Mark 1 A is basically the Mark 1 that changed to be consumer ready and we plan to ship the Mark 1 A and the Mark 3 towards the end of 2019 in a Black Friday sale. So we're actively engaged with a number of big retailers with the concept that on Black Friday in 2019 here in the United States and definitely globally will have availability of a smart speaker that comes in 3 formats the small hockey puck sized one, the Mark 1 A which we're aiming to price below $60 the Mark 2 which is a larger speaker with great sound quality that has a video camera and a screen on it so it can do a lot of interactions using that visual interaction and then a Mark 3 with a much bigger screen for doing things like recipes and surfing the web and giving access to the internet using a voice technology and the Mark 3 will come with 3 Mark 1 so you can place them around your home. In the meantime we have a lot of community members and you can actually get in our chat you can join our chat group at chat.microft.ai you can see that there are a number of developers out there that are engaged with Enterprise and sometime in the next 6 to 12 months you can expect us to launch a channel partner program to allow those guys to sell our enterprise back end into some of these bigger enterprises and so it's really exciting to be able to share the history of Mycroft, why we named it Mycroft, a little bit about the the traction and the history of the company we have 17 full-time employees here at Mycroft, we have employees as far east as Stockholm and as far west as Melbourne, Australia we have a fantastic team that is really committed to making this happen a community of thousands of developers who are contributing code to the platform every single day we have a big library of skills that are rapidly improving and we have a plan to deliver a voice assistant to production in February of 19 you should understand this if you want to know about Mycroft we've been in Alpha for a long time we went to Beta in February of this year Beta will last a full year and so the user experience that folks are getting from technologies like Alexa and Google Assistant is on the horizon and we'll ship in February of 19 when we go to production our goal as a company when we go to production is for reviewers to say Mycroft doesn't have all the bells and whistles that we can get with Alexa or Google but for users who care about privacy and user agency it's an excellent replacement to these technologies that are sending your data to Silicon Valley giants you should buy one that's really the review we're looking for in year one and the teams working very hard to make that happen in the meantime we're raising money and that's going extremely well our first day of raising money through Regulation CF was impressive to everybody here at the team it's actually growing faster than our last Kickstarter which was already going pretty fast and so we're really excited to share our vision and share our company with the community that's helping us to build it so thank you very much for watching this podcast about where Mycroft came from and a little bit of history about the company maybe not a podcast, maybe it's supposed to be an AMA I'm looking forward to connecting with everybody in the future I'm here in Silicon Valley if anybody ever wants to get together for coffee and we'll be traveling extensively throughout next year so regardless of where you are there's a good chance that we can meet in person if you have an interest so thank you very much for your time today and please come out and support Mycroft thank you