This episode features Sam Liang, CEO and Founder of Otter.ai.
Sam got his Ph.D. degree in Electrical Engineering from Stanford University in 2003, specialized in large scale distributed internet systems. He was a core team member at Google Map Location Service before he became an entrepreneur. He previously founded, Alohar Mobile, world’s first mobile location context platform that was successfully acquired by Alibaba in 2013. As a serial entrepreneur, Sam found his current company Otter.ai, focusing on building ambient voice intelligence technologies with deep learning to enhance professional productivity.
Sam shared his career path and entrepreneurship experience with us during this episode.
Highlight 1: What makes Otter.ai different from other mainstream voice assistants in the market?
Highlight 2: How does otter.ai work in real time?
Watch the full episode:
Wenli Zhou: Robin.ly is a video content platform which will help researchers and engineers to develop a better understanding about leadership, entrepreneurship as well as business insight. We aim to prepare you to become the next generation of leaders and entrepreneurs. It's so good to have you here, Sam.
Sam Liang: Thank you for having me here.
Wenli Zhou: So you ended up in Stanford, getting your Ph.D. in electrical engineering,
Sam Liang: Right, I did my PhD in EE but mostly was focusing on software. My professor is a computer science professor. This guy's - actually his name is David Cheriton. He's the guy who wrote the first check of hundred thousand dollars to Larry Page and Sergey Brin to start Google back in 1997. So that hundred thousand dollars became a few billion dollars, probably one of the most successful angel investments in the history.
Wenli Zhou: It was about the same year when you went to Stanford that your alumnus Marc Randolph launched Netflix. Did you ever think of yourself doing something like what he did, like becoming an entrepreneur?
Sam Liang: Yes, at that moment I wasn't thinking that hard. But after studying at Stanford for a few years, I was immersed in that environment. You see other people around you started to do things as big as Yahoo and Google, you started to think: Okay, why can't I do it someday? Then I intentionally joined a startup just to learn what the startup life is. But I was always thinking about building my startup all the time
Wenli Zhou: You chose to study computer science back in the 80's in college. What's your observation of this 30 years’ revolution in technology?
Sam Liang: I started studying computer science in Peking University, (they had) very high security computers we can use. Every week, you can only use it for a few hours. But over this 20 or 30 years, you see so many great things happen.
Wenli Zhou: You were a platform architect at Google? Well, what's your biggest achievement working at Google for years?
Sam Liang: They called me to join Google to build a metro Wi Fi system. This was in 2006, there was no 3G yet. So there's no easy way for you to get onto the internet, while you're mobile. So at that moment, Google was thinking: Okay, what if we put a Wi Fi router on a lot of light poles in the city, and the Wi Fi routers can talk to each other and provide Wi Fi access. So that was the first project I was working on. So I actually, I still remember, I took my laptop and went to the Castro Street in Mountain View and did a lot of testing. And while working on that, I realized that, we can actually calculate the location. So that's actually how I started the Google location service project. I wrote the first line of code and designed the architecture.
And in 2007, when Steve Jobs launched the first iPhone, that was iPhone 1, it didn't actually have any GPS capability on it at that time. But you know, Steve Jobs really wanted to have a map service. He actually came to us asking for help, we actually built the service for iPhone first. And he loved it so much, and he personally demonstrated the blue dot feature on his iPhone when he was launching it in San Francisco. I still remember those days, I actually went to San Francisco Moscone Center many times, and tested the service around the area and inside it, and make sure when Steve Jobs press the button and say: Hey, this is where I am - nothing will break and nothing will crash. And it will be accurate to show its location. So it was really cool. That also helped me form my first startup idea.
Wenli Zhou: Was that the reason you left Google?
Sam Liang: Yeah, as I mentioned, I was always thinking about building a startup. So I was also always thinking of what product should I build? Well, working on the Google Map helped me understand: Okay, this is actually a lot more interesting thing to do. At that time, there was another startup, they're still alive. It's called Foursquare that allows people to check in on their iPhone or Android when they visit a restaurant, when they visit a park and check in. So I thought that was cool. But why do I have to manually check in, can we build a service that’s able to track my location all the time, and can understand where did I go? How long I stayed there? What did I do? So that's the basis of our first startup. So I quit Google in early 2010. So the first year was actually really difficult. I was having no salary at all. I had no funding at all. My wife actually was really nervous because we have two boys and we have to pay mortgage. I had to negotiate with her, convince her and say, “Give me one year, if I cannot make it work, I will go back to Google.” It was difficult, but it was a lot of fun.
Wenli Zhou: The company that you previously found called Alohar Mobile was acquired by Alibaba. What was your company’s situation when it got acquired?
Sam Liang: After a few years, we grew the team, we raised some funding - a lot of ups and downs. I think in 2013, we were talking to some VCs and we got some good offers, and we've got acquisition offers from some other really well-known Silicon Valley companies. I cannot disclose who they are. And then Alibaba, and actually also the Chinese map company, AutoNavi, contacted us, and we started talking to them. And I was convinced that they were very sincere, and they can help us bring this service to more people. So we decided to take their offer. But we were still operating independently as part of the acquisition
Wenli Zhou: That was when you decided that was a good time for greater goods, right?
Sam Liang: Yeah, because they definitely have a lot of resources, have a lot of services. So we thought, by working with them, we would be good to generate a bigger impact, that's the motivation of doing a startup and create something new.
Wenli Zhou: Many entrepreneurs’ dream is to get their company acquired. How does it feel to be on the other side?
Sam Liang: The good thing is that you get a little bit safer. It was the support of a large company behind you, you definitely have a lot more security there. In the meantime, the things you want to do and what they want to do are not always aligned. My heart is still an entrepreneur. So I want to try something even more audacious, even more ambitious. So, because we got some initial experience with startup where we've got more confidence, so I decided to move on and start a new company.
Wenli Zhou: Tell us what Otter.ai is.
Sam Liang: Yeah. Otter.ai is a new product we built from the ground up just barely three years since then. The key technology is speech recognition with AI and deep learning. So when we first started this, everybody asked us: Hey, isn't Siri already doing that? Isn’t Alexa already doing that? Why are you doing voice and speech recognition?
Wenli Zhou: Yes, what makes ours stand out?
Sam Liang: Well, if you want to do a startup, you have to do something unique. And there's no way for you to do something exactly the same as the large guys. There's just zero chance for you to succeed if you compete against them head up. We decided to do this, but we targeted a totally different market, a totally different use case. When you use Siri or Alexa, you say hot word like “Hey, Siri!” or “Hey, Alexa!” to wake up the robot. That's great, I mean, it has its use case. But if you think about how many times do you talk to Alexa every day. Some people talk a little bit more, but most people I think talk to Alexa no more than 5 or 10 times a day. So the total talking time would be no more than one minute. How do you take the consults? How do you remember the things? How do you analyze the conversation? We actually looked at the market and said, there's no such product. However, talking is one of the most common ways for people to communicate. Everybody talk several hours a day. What if I can record my whole life, what will happen? What - can I search for things I've heard in my college years, even in my high school years? What did my mom tell me when I was in high school? When you met your first girlfriend, what did they say? That will be interesting.
And also for business, I have a lot of meetings with venture capitalists, with potential customers, with job candidates. I found myself forget things all the time. So I realized that actually such a product will be really useful. But then on top of that, the AI can actually analyze all the things I say and analyze all the things I heard, and give me some insights, give me some suggestions, give me some reminders, as well. So that's why we built this.
The Otter is a product you can use on your iPhone, on your Android for free. You can also use it in a web browser by just going to otter.ai. So we launched the Otter service less than a year ago. Since then, it's growing really fast. Just a month ago, Google selected Otter as the best app of 2018. Three months ago, there was a very big conference called TechCrunch Disrupt San Francisco. The entire conference used Otter as the official voice app to transcribe all the speeches and all the panels in real time, and showed the transcript to all the people. And the transcript was also streamed live on the internet, and also made their conference really exciting as well, because they said: Well, in the last 10 or 20 years, nothing has changed for our conference technology. But now Otter brings AI into this system.
Wenli Zhou: On the technical side, from my understanding is that Otter uses AI to recognize people's voice.
Sam Liang: Right. There’re two parts, speech recognition and voice recognition. Speech recognition is used to convert song and speech to text; and the voice recognition or speaker recognition, recognizes who's talking. So that's a separate piece of technology. And actually, part of that, there's also another word called diarization. Diarization is a technology to separate each person's speech and then use just the voice recognition to recognize the identity of the person. Once you detect a few sentences by a person, and we will create a voiceprint profile for that speaker, then it’s able to recognize the rest of the speech and recognize future speech as well. So what we do is to recognize the unique voice from each person.
Wenli Zhou: What’s the accuracy right now for the voice recognition?
Sam Liang: Yeah, for the native speakers, could be 95%. Sometimes it drops a little bit with noise.
Wenli Zhou: Once, like right now, you have the power to recognize people's voice. What do you plan to do with it in the future?
Sam Liang: Yeah. Recognizing people's voice is interesting, because it helps you understand the conversation better. Because if you're only looking at the sentence, if you don't know who said it, you're missing a lot of information. Because even if it's the same sentence, if it's spoken by the Google’s CEO, who may probably mean a different meaning than spoken by Apple’s CEO. So in our conversation, people take turns to ask questions, answer questions, explain things. So if you can recognize who is talking, it helps you understand the meaning better, because you actually - if you have a lot of conversations from that person in the history, you can look at what they said in the past. So that helps you understand what he means by saying the current sentence. And over time, the system can recognize: Oh, the action items come from a product meeting, right? We already have some prototype; we haven't released it yet to recognize action items, decision points, right? When you have a weekly meeting, or a daily standup, what do we need to do? The product manager always needs to take notes, but eventually that can be done by Otter or some (other) AI system in the background.
Wenli Zhou: You can just search.
Sam Liang: You can search, you can search by keywords, you can search by a question, you can search by this speaker name. You can see all the speeches here, you can you can search for words like “self-driving car”, and you will find all the speeches that talk about self-driving car. Otherwise, the voice - if you think about in the human history, how much information is encoded in the voice, but very minimal amount of voice is actually saved. All the speeches in your life will be saved somewhere. It will be saved confidentially, and only the part you want to share can be shared.
Wenli Zhou: I know that Otter is the exclusive partner with Zoom to transcribe all the meetings. Is there any other enterprise that you're working with right now?
Sam Liang: Yeah. First of all, Zoom has been a great partner. It's the hottest video conference system in the world now. They licensed our system exclusively to provide automatic transcription. This is actually the very first media conference system that provides automatic speech recognition, automatic transcription. Surprisingly, it’s not Google, it’s not Microsoft, it’s not WebEx, it’s Zoom using Otter, so that gave us a lot of credibility. Other than Zoom, we're actually working with a lot of other partners. Most of them are still confidential right now. One of the public ones is called Bridgewater Associates. It is the largest hatch fund in the world. They manage $160 billion And it’s founder CEOs name is Ray Dalio. One of the principles he used to talk about is radical transparency. In their company, they actually record all their meetings for the last 15, 20 years. They actually heard about our company and came to us and were asking for help. Once they saw the demo, they were really excited. So they decided to use Otter for their meeting analytics as well.
Another big part of our partners are universities. Once we released the Otter, a lot of students and teachers started to use it, and they found it useful for lecture notes or faculty meetings. We're actually having a pilot project with Harvard, with Western Kentucky. We're visiting UCLA tomorrow. They are considering two types of services: one is for international students. So actually for new students who just came to the US are actually not very easy to understand the professors, which actually I had difficulty myself in my early days in the US. I had to use a Sony Walkman to record the lectures and really listened it a couple of times to understand better. But now it's like, with Otter on your iPhone, you can actually get the transcript, get recording the same time, make it much easier to understand the lectures and also for accessibility. AI is really helping so many different use cases in so many different situation.
Wenli Zhou: Otter raised Series A round in 2016 from several venture capitals. How active are you communicating with your investors?
Sam Liang: We are very active. For people like Tim Draper, Horizons Ventures - it's another major investor. They're the very early investors in DeepMind which built AlphaGo, and others like Waze, Spotify.
Wenli Zhou: Close relationship with investors, does that means that they can influence company decisions?
Sam Liang: They still - they trust us to make the best decisions. Of course, these investors have a lot of experience. They give us advice. We have a very open discussion with them. We tell them what we really want to do. Most of the time, they listen to us; but sometimes they do say: Oh, you may give it some second thoughts. They gave us a lot of introductions.
Wenli Zhou: That will benefit the company.
Sam Liang: Oh, yeah, of course. They have confidence in the technology, they have confidence in the team, they have confidence in the market as well. So because of their experience, so they can recognize or make reasonable forecasts of the future. Because whenever you build a new startup, you want to predict what the future will be. It's not what the market today, you want to predict what the market will be in the next year, in the next five years. Then you want to move early, even before the large companies like Google, Amazon, Apple realize that’s a new market. That's how new startups are born every year. And of course, a lot of them die, probably 99% eventually died. But 1% or 1.5% became very successful.
Wenli Zhou: Right now, do you feel more pressure to pave the way for other Chinese entrepreneurs in Silicon Valley than you started?
Sam Liang: Yeah, that’s why we're having this conversation. I was invited to a lot of seminars, conferences to discuss our entrepreneurial career. I got to meet a lot of engineers and students, explained to them: Why did we do this, how did we do this for people who are actually still exploring the entrepreneurial career. Either you start right away, or you can already - you could join an early stage startup to actually experience it, to see how a startup is run on a daily basis. After 3 or 4 years, when you have enough knowledge, you could do something yourself. Everybody's path is different. Obviously, it's good to experiment, to try it, rather than staying in the same place for 5, 10 years. That's too long. And you know, for people who stay in one place for 10 years, there's a term in Silicon Valley they said, that's sort of as a sealed coffin. You're so comfortable there. You don' think you can change, you don't think you want to change anymore, because everything is so easy in that place.
Wenli Zhou: Well, thank you so much for joining us at Robin.ly today.
Sam Liang: Yeah, thank you so much. For people who want to try it, just definitely download our app.
Wenli Zhou: Yes, definitely give Otter a try.
Sam Liang: And also, for people who are interested, we're also recruiting engineers. Join us, come to our website. We're looking for all kinds of engineers, AI people.
Wenli Zhou: Thank you. Thank you so much for sharing.
Sam Liang: Thank you.