The VoiceFirst Roundtable - Episode 4
Host: Bradley Metrock (CEO, Score Publishing)
Duration: 31 minutes, 04 seconds
Bradley Metrock: [00:00:00] In the fourth episode of The VoiceFirst Roundtable I interviewed Leor Grebler, CEO and co-founder of UCIC, a company that works with hardware makers to integrate voice. In this interview, you'll hear him discuss his 2012 Kickstarter for the Ubi, which was a product that was way ahead of its time, how that shaped his voice-first vision from that point forward. There's a little bit of echo on my microphone as a result of recording this podcast remotely, so bear with that, and that will go away for the next episode. Thanks very much for listening and enjoy.
Bradley Metrock: [00:00:43] Hi and welcome to Episode 4 of The VoiceFirst Roundtable. This podcast is dedicated to examining all issues of the emerging field of voice technology. Our sponsors for today's podcast are Fourthcast which turns your podcast into an Alexa Skill. Get started at Fourthcast.com today. And that's spelled F-O-U-R-T-H-C-A-S-T. Our other sponsor for this podcast is The Alexa Conference. The Alexa Conference is the annual gathering of Alexa developers and enthusiasts. Learn more and get registered at AlexaConference.com. We are so pleased to have as our special guest today, Leor Grebler. Am I pronouncing that right?
Leor Grebler: [00:01:31] You got it right.
Bradley Metrock: [00:01:32] Perfect. Thank you so much for joining us Leor, we greatly appreciate it.
Leor Grebler: [00:01:36] Bradley it's a pleasure to speak with you.
Bradley Metrock: [00:01:39] So I want to start with....almost five years ago, August 2012, when you and your colleagues rolled out a Kickstarter for the Ubi. Let's just start right off the bat with you explaining what that is.
Leor Grebler: [00:02:01] So the Ubi was short for ubiquitous computer. It was a Wi-Fi connected, voice-operated computer, that essentially you could plug in, it would connect to your home's Wi-Fi network, and you could talk to it. You could ask it to play music. You could look up weather information. You could send email and text messages. You could even make phone calls. You could also control home automation devices through your phone. So yeah, it sounds familiar maybe to some things that are on the market today.
Bradley Metrock: [00:02:37] It does. And the form factor, it looks like if you took a Echo dot and you turned it into a square, and maybe flattened it a hair and then added a plug to the back of it. That's what this thing looks like. And I want to take a moment and read....I've got the Kickstarter pulled up in front of me. I want to read the list of features that you guys launched this thing with, voice-enabled Internet search, speaker phone, indicator light, home speaker system with sound piping, virtual assistant voice memos, alarm clock, intercom system, baby monitor, noise pollution monitor, and controlling the climate of your home. And a partridge in a pear tree.
Leor Grebler: [00:03:27] It slices and dices.
Bradley Metrock: [00:03:28] So I want to emphasize that. Because this was not yesterday that you rolled this thing out. This was in August 2012, and in contrast, Apple just had an event just a few months ago where they showed off the Homepod and there's a slide that I have burned in my memory of Phil Schiller explaining the Homepod, and their list of features was actually worse than yours from August 2012. So what you've done is incredibly, incredibly impressive.
Leor Grebler: [00:04:01] It sounded good and we kept on adding more features to our feature list as we went through our Kickstarter. Of course, that was a rookie move and we should have actually had fewer features listed, but yeah, we could do a lot with it. We weren't sure where this technology would go, and we weren't sure how people were going to use it. What we knew is that the supporting technologies were finally getting to a point where if we put them together, we could create something that would be really unique and the result is, we start to think about all these different ideas on what you could do with this type of product. And that's why you ended up with that large feature list.
Bradley Metrock: [00:04:50] You did all the Kickstarter stuff right. You had a phenomenal video which I watched earlier today and you had all the tiers for your Kickstarter very thoughtfully planned out, and you had a goal of $36,000, and you ended up with nearly $230,000. So it was a big success obviously, it was a very forward thinking product, to say the least. Walk us through sort of what happened. Obviously you delivered on that. Talk to me and the audience about how that led you to where you are now. Tell us the story of how Ubi lead to UCIC.
Leor Grebler: [00:05:37] So yes. So actually the story starts a little bit before our Kickstarter project back in 2012. And you mentioned a few things with Kickstarter, and I won't take this conversation in the direction of Kickstarter, but a lot of the things were really intentional. 2012 was kind of the heyday of Kickstarter, where you could you could put something more or less amateur up, and it would still get quite a bit of traction. Things were still really new and exciting on Kickstarter. Now you have to have a lot of marketing and a professional video in order to make your Kickstarter successful. So we actually were hacking Kickstarter quite a bit. We had analyzed the top 15 to 20 technology Kickstarters. We looked at a bunch of different variables for all of them, and we said, "Okay. At the very least, we're going to design a Kickstarter that's going to be successful." And it wasn't our first. We had actually....your listeners should never look up Salsa Clock just strike it from your memory. Don't look it up. But one of my partners, we had done a Kickstarter for that. That was just basically two guys talking about salsa dancing in a video. And at the end of our Kickstarter campaign, we ended up....I think were trying to raise $5000. We ended up with $150 and there were two backers. There was myself and my partner …
Bradley Metrock: [00:07:01] And your mom wants her money back.
Leor Grebler: [00:07:02] Yeah, but my mother wouldn't even contribute to that Kickstarter. So she wanted nothing to do with it. But no, it was a big failure. But we actually learned a lot from that in order to go ahead and prepare our next Kickstarter which we knew we would have to have a real tangible product. We had to have certain wordings, certain pricing tiers, you had to create demand at the beginning for it. So a lot of things that we did to hack it. But before we even talk about Kickstarter, and just the process of raising money, the idea of the Ubi actually came quite a bit before our campaign, where we were....myself the other co-founders, we actually have our backgrounds in engineering. We worked for a company that provided research and teaching at university. It's a really cool company. And we got to go all over the place and see what was coming. And I know you're from Nashville, right? So I had a chance to visit Vanderbilt and visit the Haptics Labs in Vanderbilt, where they were working, and we were trying to see if we could help out with some of the projects around in Haptics, and in that lab. So we got to see all these cutting edge things and when we would come back to town we would go out for lunch and we kind of see this disconnect where people were really....all the people we were working with you know as part of our day jobs, were working on these cutting edge technologies that were going to revolutionize human computer interaction.
Leor Grebler: But when we looked around us, people were heads down on their devices. They weren't even paying attention to the person sitting across from them. And they were distracted and we saw this this progression where people would become more disconnected from the world around them, more disconnected from other people. And maybe the way that they were interacting with technology was the problem, where you are heads down looking at a screen that consumed all of your attention. What if you could get technologies to fade into the background and chime in when you need it? And we started to think about all sorts of ideas. Originally we were thinking of just a lamp that would glow different colors to alert you of different things. This has been done like dozens of times by different Kickstarter projects. We thought about it, then we kind of paired it down. What if we were to work on just a plug, that you could plug in, and it could be Internet connected. This is before Wemo, this is before millions of devices that are similar to that, that are just Wi-Fi connected the plugs. So we started to work on that on that as a project. And we actually had gone ahead, we made a Kickstarter video for it. We had done all sorts of research on it.
Leor Grebler: We were actually going to go forward with that, when all of a sudden, another company, another startup actually had launched a similar project and we knew that we had lost our first mover advantage. So we scrapped that idea. And eventually we went back to the drawing board and said, "What else can we do with this that would really kind of make an impact and go in that direction of getting technology to fit the background?" And we said, "Okay. This has a Wi-Fi connection. What if we were to add a microphone to it so you could control it?" I said, "Okay, that's pretty cool. It's like the clapper or something like that, where you could do you could do a little more." And then we said, "What if we added a speaker to it? Then what else could you do with it?" And the applications at that point mushroomed. And we realized we had something that could be really cool, and really innovative, and it was with that that we knew that we could go on Kickstarter, and it would be a remarkable product. People would be literally willing to remark about it and the idea would spread. So that's how the at least the Ubi got started. And we put it on Kickstarter, and of course know with all of our work from learning from previous failures that we were able to get a successful Kickstarter. We were able to work with other IoT projects that were on there to get lots of people coming to our campaign. We got a lot of press, and it was a big success for a Kickstarter project and then towards the end the project we started to have this like horrible realization that we actually had to build this thing. And that was really frightening, that there were now thousands of people who had ordered this product and we had to get going. Now at that time, 2012, you didn't have much. I mean you had Siri, but Google had just come out with....I guess they were calling it Google Now at that point and we said, "OK. Do you know what? This is going to be a piece of cake. We're just going to get a little Android computer and we are going to stick it inside of this box we're going to add a speaker and microphone to it. And we are going to basically just leverage all of Google Now's capability where it started out with these voice responses back when you were to search, and it didn't have the trigger word. But if you press the microphone button, and asked, "What's the weather?" it would actually speak it back to you at that point which was really cool.
Leor Grebler: So we said, "OK, we're going to get an Android computer, we're going to put it inside. We're going to add a trigger in front of it and we're done. We're going to deliver this. We're going to be an early delivery Kickstarter project!" And we very quickly found out that wouldn't work. We couldn't....the device was a headless device. There was no monitor for it. So we had to figure out where to simulate a button push after a trigger word. And whenever there was a change to anything from Google it would break that process. And also, there was no predictability about what would come back as a voice response from Google versus a text response. And if it was a text response, how would we actually deal with it? So there was no API for voice interaction. And it really dawned on us, like oh what have we got ourselves into. We realized we had to start to put together all of the components on our own to make voice interaction hands-free, ambient voice interaction work. And it started off with you know when you look at the spectrum, the chain of voice interaction, there are a couple of components. So the first is a wake-up word. So whether ....ours was, "Ok Ubi" but you have "OK Google", you have "Alexa", you have all these other wake-up words. Oh, and "Hey Siri". After that, you have speech-to-text.
Leor Grebler: So you have different services that will do speech-to-text. Now there are lots of them, but there were only a couple that were on the market a few years ago. And then after you get back the text, you have to do natural language understanding. What is the user's intent? You actually have to do NLU. NLU is just one way of doing it, but you have to understand the intent. Then you have to do the integration, and then you have to do text-to-speech give some type of acknowledgment back to the user. So we kind of broke down the interaction into those steps and we had to start working on every component of them. And then beyond that we had an additional problem, and that problem was far-field, and as soon as you get beyond like arm's length, of course, you get a big drop in the signal. The signal degrades. I think it's a square law, or a power law. So if you double the distance, you get four times less signal when it comes to voice. So with that, we have to find a way of augmenting the the voice signal without augmenting the noise as well. And that was the challenge. Today, there are maybe five or six companies on the market that offer off-the-shelf digital signal processing for far-field. They have far-field microphone kits. I think there's going to be more that are going to be on the market even by the end of this week. Lots that are there. But back in 2012, that wasn't the case. You had a few solution providers who are doing things for automotive. But not really for far-field...
Bradley Metrock: [00:16:02] And that's where I've got a question for you. So here you are doing all this stuff, and it's not like you're doing it secretly. You're doing it in the most public way possible with the Kickstarter. You're figuring out all of these problems where....and you're creating this device where if someone rolled out the Ubi tomorrow, Leor, with your feature list, and the same video that you produce with just somebody else in it. That's a commercially viable product today.
Leor Grebler: [00:16:43] Yes.
Bradley Metrock: [00:16:44] And here you are doing all this stuff back in 2012, into 2013. And nobody....I mean, yes, you had twelve hundred backers, and I bet it would be fascinating to look at who those backers were. But you should have been acquired within the first two seconds of you posting the Kickstarter. So why....actually, maybe you can explain some of the communications you no doubt got from some of the big tech companies and how your interactions with them have been doing that, amd since then.
Leor Grebler: [00:17:23] Yes. I wish I could talk more. I got some great salacious stories to tell you about.
Bradley Metrock: [00:17:31] I have no doubt. But you're under NDA. Don't let us ruin that for you.
Leor Grebler: [00:17:38] Why not? So let's just say that we spoke to many of the big companies early on and I think it was a combination of multiple factors that we clearly weren't acquired by any of them. But one of the things is that acquisitions....we were doing something very public, and so an acquisition will be very public. And if that were the case, it would probably signal what any one of the companies were doing long in advance of a release.
Bradley Metrock: [00:18:13] That's true.
Leor Grebler: [00:18:15] So that's the explanation I tell myself over why we weren't acquired.
Bradley Metrock: [00:18:22] That's a good one. I have not thought of that.
Leor Grebler: [00:18:25] So even if you look at....the Google Home didn't come out until this year, Apple Homepod was only a couple of weeks ago, and you know the the first Echo device was you know more than two years after our Kickstarter campaign. So that's still a long ways off from 2012. So if they were to have come in and swooped us up, then it might say, "Oh OK, these guys are working on something and would have lit the fuse under under them to maybe get a inferior product out before it was ready." So I think I think that's kind of the reason why at least one of the reasons why. The other is that there is still a lot of things that needed to be solved, and they were content to kind of look at and see what this small company was doing before deciding to jump in. And as well, the other technologies still needed to evolve. So like I was talking about far-field, when we were doing our work on on far-field, we were getting quotes something like $2 million plus to do an implementation of a far field algorithm on a on a DSP chip, and the bill of materials cost would have been like 10 to 20 dollars at high volume per device, which is just enormous. Especially for a startup we would have had to raise you know a significant amount of funding in order to be able to pull that off. And even then, there would be no guarantee of performance, and no guarantee of market. Plus we would have to also create our own channel for marketing the products so there were a lot of challenges for what we were facing.
Bradley Metrock: [00:20:22] Well sure, and I don't think anybody had the vision for how all this was going to play out, and a lot of credit goes to Amazon....all of these companies are doing work and they're working furiously at creating their own voice assistant and all that, and that entails machine learning and all the rest of it. But Amazon deserves unique credit for being out there marketing it. You couldn't watch an NFL game this past season without seeing ads for Amazon Echo.
Leor Grebler: [00:20:57] Oh yeah.
Bradley Metrock: [00:20:57] And that did something. It formed the market. If you just imagine this bunch of clay in front of you, and Amazon sitting there shaping the market and preparing it and showing people the use cases, showing people that this is not a gimmick, showing people that this is the future, and so I think they get a whole lot of credit, but you're right. Back then, if Amazon had that vision, they certainly couldn't have shared it with anybody. I doubt that they did. And no one knew how all this stuff was going to go.
Leor Grebler: [00:21:31] Exactly. And to your point, the education that was required for the market was a very high bar, because this was a new interface altogether. There's so many little things that could make a voice experience not work well, that it requires a few initial attempts, even how you say the wake word, how quickly you can talk to the thing, when it's finished speaking. We saw all of this from the early interaction with the Ubi that you know there's a certain cadence that you use when talking to devices. If you don't have that in place, it won't work. So Amazon had to educate the market and the amount of effort and marketing dollars that went into that is just enormous when you look back, my prediction, my estimate, and this is purely unscientific, is that it's probably over a billion dollars that that's gone into development and marketing of Alexa. And thousands of person-years in development. There are thousands of people doing Alexa work now. So I still remember that infamous day, I think it was November 6, 2014, when I get this e-mail from my brother saying, "Uh-oh.", with a link to the Echo. And now my Echo just lit up...
Bradley Metrock: [00:23:04] Congratulations, we're 20 some odd minutes in, and you've now achieved the rite of passage of all VoiceFirst FM podcasts, activating a voice assistant.
Leor Grebler: [00:23:14] Exactly so. So with all of that they were the first thing on the Amazon.com website for the week following their launch. So I think there might have been something like half a million pre-orders but the amount that you would have to spend to get your own product on the front page of Amazon for a week is probably in the tens of millions of dollars.
Bradley Metrock: [00:23:44] If they even let you.
Leor Grebler: [00:23:45] Yeah. And so we saw really quickly like that.
Leor Grebler: [00:23:50] Now looking back you know things are bittersweet, by that point, I was fairly bitter about the about the Echo coming out, and clearly, at that point, we were actually selling Ubi. We had launched from beta, although the product still had issues to overcome. We had launched out of beta at that point, and we were selling on Amazon the Ubi. And so our kind of claim to fame was, if you want something like the Echo, and you want to buy it on Amazon, the only thing that you could get is an Ubi. So we were kind of hanging our hat on that, but now all the the effort that went into educating the market really has helped us, and as we transformed away from you know being our own consumer hardware product company, to helping other companies turn their products into voice interactive products.
Bradley Metrock: [00:24:51] That's part of what I think is so interesting about you, Leor, and what you've been through, is a cautionary tale of entrepreneurship in the voice space, and this is just for tech in general, but for voice specifically there's a lot of venture capital money and investor money that wants to come into the space. And it's like anything else. You have to tread lightly around the four horsemen of tech lest you get stopped, and that's not exactly what happened to you, and you're too smart. Obviously you're in a great place now, but there's a lot of stuff coming up today, that you just look at it, and you're like what are you thinking exactly, because all it takes is Apple or Amazon or Google or Microsoft to decide that they want to do this thing, and your revenue goes directly to zero.
Leor Grebler: [00:25:45] Yeah. So you kind of have to learn to dance with the giants around you. Like, a lot of our work was around timing. Of course, we came out with a product a really early way ahead of where the market is, before someone had educated the market and when we were going out and doing our own fundraising for it, we were met with countless objections. People saying things like, "I'm not really sure that voice is the future" or "Why would I tell a device to turn on the lights when I could just go up the light switch and flick the light switch?" or "Why isn't this going to be on a on a mobile device to use it?". And now I just want to go and like throw an Echo through their window.
Bradley Metrock: [00:26:37] Just go pull up that e-mail and forward it back to three or four years later.
Leor Grebler: [00:26:42] Yeah they would laugh it off. I'm sure. That's fair enough. Not a lot of people saw saw this thing coming. But now for a product to come to market in voice, it can't just be equivalent to the Echo or the you know Homepod or Google Home. It has to be significantly different. It has to have a new type of interface that has to do something that is going to wow people. And it's not enough just to kind of come up with a product that's going to match capabilities. So with voice, there's kind of a challenge that that hardware makers are faced with. One, do I integrate these services into my product in order to make it better for the users, can I have Alexa on my lamp, or can I have you know Google Embedded Assistant or Google Assistant on my device in order to allow for streaming music through it, as I can to create a better user experience. Or do I create my own user experience. Maybe I only have a couple of local commands for just controlling the device hands-free or maybe I go with a entirely new interaction where I have a branded experience through voice. So there's these new questions that these these companies need to address when thinking about whether or not to add voice to their products.
Bradley Metrock: [00:28:07] It's a perfect segue to you talking about what UCIC does today. So what do you guys do and how do you help companies navigate these waters? Sort of explain and Walk us through that.
Leor Grebler: [00:28:23] Yes. So companies typically approach us where they're interested in voice in some way. They're working on a product or you know the next revision of our product where they want to implement voice and they're not sure how to proceed. They're faced with different services that are out there that they could implement. They're also faced with some of the questions about how they should implement or address their brand in conjunction with these other services for voice interaction. So one area where we can help them, is to go through the different options that are available. We can also guide them on how this can affect things like the the product development timeline or the cost or the type of components that they need to add to their products. So then at the next step, if it's working on different components, it's selecting the components. It's testing them, it's ensuring that they move forward with the best option for voice and their product. And then it's helping them get that product to market in the most timely and cost effective manner.
Bradley Metrock: [00:29:39] Perfect. And your web site is . And people can contact you through there?
Leor Grebler: [00:29:48] Absolutely. They can check out our website, and they can reach out to us. We've developed a few tools as well that they can use to help test out what voice interaction would be like on their product. So a few of them are going to be coming out over the next few weeks and they can also check out a skill that we created called the Ubi Portal Skill under Alexa. Yes, there's a bunch of ways that they can get in touch.
Bradley Metrock: [00:30:20] We'll have links to all of that in the show notes if you're interested in checking that out. We will make those links available. Leor, thank you very, very much for giving us some of your time and sharing with some of your perspective and experience today.
Leor Grebler: [00:30:35] It's been a lot of fun. Bradley, thank you so much.
Bradley Metrock: [00:30:37] You got it. For the fourth episode of The VoiceFirst Roundtable, thank you for listening. And until next time.