A PhD at Stanford for Internet distributed systems and working on their professor David, who is known for being the first invest, Investor in Google. And also we have another speaker a moderator tonight is Xu Wei shell, managing partner of Tina Avengers, so he and Sam will have her. We expected a very great chat in the second part of tonight's event. So without further ado, I'll let Sam take the stages. And also before that just a very friendly reminder that we have a red icon on top left corner, which is the live chat script by otter.ai so you can just click in and see how the transcripts goes with this very useful tool, and relaxed in a meeting. So set the stage is yours.
Right, thank you, Sofia. Thank you, Shipway and tumor manager for organizing this event. I don't know I forgot it for Sophia mentioned, we are honored that Ching Yuen is one our investors and have provided a lot of help in partnerships recruiting and customers. So, just want to thank Chinmaya mentors for your support. Um, so yeah, I would do a quick presentation, Then. Then I'll have a fireside chat with Seth Sheehy, let me share my screen. Yeah.
Wait a second.
Can everybody see the slides.
Yes, we can see it.
Well, okay, yeah. So you're also broadcasting on clubhouse, I guess.
Yeah, yes, I am broadcasting that clubhouse.
Okay, so those people won't be able to see the slide but. But hopefully, if you share the transcript URL, everybody can read it as well.
Yeah, yeah, that's what I'm going to do.
So yeah, just a brief history about AI was founded in 2016. And we're based in Los Altos here. I'm actually in our office and also the US right now. Although everybody else is working from home. I like to come to office. The team is actually super talented, the picture I'm showing here is onstage at San Francisco TechCrunch Disrupt San Francisco in 2018. That's when TechCrunch Disrupt selected otter, as the official voice app to transcribe live all the speeches in TechCrunch Disrupt, as everybody knows, the conference, it's a very high profile in this conference in 2018 Actually, Oprah CEO was speaking there, Dropbox, CEO was all the speeches were captured and live transcribed by otter. We were actually one of the hottest startups there. And, actually, Simon, Simon same in this Zoom meeting Simon is here on stage.
So, you know when you build a startup you need to think, you know hey what problem are you trying to solve, you know the problem we're looking at, or. The problem was meetings. Everybody in enterprises No, they're just so many meetings. You need to have every day. After COVID It happened actually, in the beginning, people thought okay I can work from home. I don't need to commute, which can just help me relax more, but it turned out that, you know, though, people don't commute, they actually schedule out even more zoom meetings. So people, you know, really exhausted by zoom fatigue, you know, talking about in a meeting problems, you. It's hard to have everybody you know have common understanding. Remember what had been discussed what have been decided. What are the action items, what are the flags, if you discuss with you know for me. We talk a lot with VCs customers we discuss a lot of numbers. Now, how do you remember all of those. So, typically, take notes, you know, take notes during a meeting, it's actually not that easy. In the past you have a paper notebook you write, use a pen to write. But now you may, you know type on your laptop but there are distracting. So that's why we created otter. To solve this problem, especially for the teams. You know when you have a product team meeting, you may have 1020 people in the meeting, you want to make sure everybody get the notes, everybody is synchronized and also there are always people who cannot join the meeting live, how do they get the information. So as Sophia mentioned in for this event, you can actually get live transcript by clicking on that link, And the upper left corner, corner. This is all done by order. It's nicely integrated with Zoom. Yeah, but in fact actually zoom license alters technologies and provide that to other zoom users as well, although it's to get the full functionality you need to use otter directly. So this is a huge market opportunity. There's all kinds of estimation on the number of meetings. You know there's one source set, you know, every day, it just in the US are 25 million meetings are happening every day. People talk a lot. There's tons of information being changed. And also, there are a lot of meetings that are may not be necessary, or you may need to have everybody attend meetings, attend all the meetings, so if you take up those meetings, a lot, you know, if you know a lot of people attend unnecessary meetings actually can cost people billions, 10s of billions of dollars, and in terms of time. If you're a middle layer or higher, upper level management, management, you spend actually what 35% of your client meetings. So, we have the honor to help people collaborate and communicate better. This actually have applications, or use cases in many customer segments including enterprises, education, media.
See, you know, Otter, working in the background for this type of zoom meetings, You use this for all your own meetings. Actually when system cut your calendar with otter otter, it can actually automatically start for all your zoom meetings and can make sure all the meeting notes are shared with everybody. You invited. So the real time transcription uses author's own speech recognition engine, you know this is a question. We have been asked all the time. Hey, did you guys use Google API, Do you guys use Microsoft or Amazon. Actually, if you are a good scientist or engineer, you can test it against the Google, Microsoft, side by side, you will see that our provides much better accuracy and higher performance. This is actually very surprising to a lot of people, you know, when we talk to VCs talk to customers, it's hard for them to believe, you know, a start up with, you know, just a few dozen people how can they build this. So there's a lot of, you know, secret behind it but, of course, we do work super hard. And we build our own AI engine behind this, you know we crowd millions of hours of audio data would build that. The Data Selection pipeline, select the right data to train the deep learning system. So definitely a lot of hard work, you know before we launched the product, you know, it'd been. It took us more than two years to just build the foundation before we launched the product, but it definitely, you know, Other companies like Google nuance, Microsoft, or have been working on this for like 1020 years. We, we did build something much more advanced and can catch up really fast.
So
you know there's a lot of questions in why do we do it now. There's many reasons. Number one is actually there is a big market demand, obviously it is described, and now it's more clear to people a few years ago when we started this, you know, and other people were skeptical, they say Hey, isn't speech recognition, a solved problem. No. didn't work google assistant or Siri, you know, don't they already do this but, you know, the difference is that are those system actually cannot process meeting conversations very well, you know, they can. Well use Siri, you can say hey Siri. What's the weather tomorrow, or set alarm at 3pm. It can process a short question or a short command but for long form conversations in meetings, they actually don't work very well. So this is where we actually build our own proprietary technology and peut otter. On top of it. So, see this is the future of work in terms of remote collaboration blend of synchronous communication with asynchronous communication. So, lots of interesting things to do. I'll just quickly go through the market it's recognized otters. Matthew, you know, we, it's a small start up it's actually pretty surprising to see that otter has been covered everywhere, a New York Times, Wall Street Journal. A year ago when we launched the otter live notes for zoom. I was invited to demonstrate this feature on CNBC national TV. Some people were asking hey what's the stock symbol for otter.ai. The no this is such a stir. A smart, smart startup in Silicon Valley. Otter, and have been, you know, selected by Apple as Apple day also won the one of the five best daily helper apps award by Google. Oh, Sophia mentioned David Sheraton, you know, we have lots of support behind us, this is Professor David Sheraton, who wrote the very first check to Larry Page and Sergey to start Google in 1997 I think. Yeah, so he's a investor in order and a strong supporter to us. And also, Eric. Zoom CEO, you know, he came to us in 2017 in Tech has their engineers to test our system and found no otter has much higher accuracy then Google that Microsoft. So they decide to work with us. So so far we have transcribe more than 100 million meetings with more than 3 billion minutes, so it's increasing every day, rapidly. On the right hand side it's. Some people might recognize him. It's a well known VC Tim Draper, says a big fan of Otter. See, he's wearing a t shirt and he's using otter in all his own meetings, you know, as a one on VC. There's here it's got founders pitching him all the time, of course, it's he's on a lot of boards. So he's using ordering, or his meetings. Yeah just actually this is our team's holiday party before COVID happened in Palo Alto. It's part of the, the team's very diverse
and very fun thing. So yeah, actually we're, you know just received a $50 million funding or looking for more people to join us, engineers, back end front end, AI, scientists, engineers, so if you're interested in AI interested in the new collaboration product, you should check out our career page. So that said,
Thank you Sam. That was a great presentation. Every time I learned new things from just listening to your presentation. So, I'm dying to get into this next section because we have so many interesting questions we want to ask you. So, let me just jump into it. So, so, so some of you may not know, but this is Sam's second successful startup. So my first question is about, you know what is the difference in your experience building otter from the first time when you have our first startup.
Yeah, I'm just sharing the screen in case anybody hasn't seen the live transcript. You can see if this is happening in real time. And in case you hear anything interesting you want to highlight things, there's a button in the lower left corner. So this helps you know remember important things and you can go back on a time. So, you can do this yourself, as well. So a second time. A lot of things are different, of course, we learned a lot from the first time. The in terms of understanding the market, understanding the business side, understanding what kind of customers, we should target. And of course, in terms of team building in terms of allocating resources. There's so many of them, you know, only the one Stephanie, taught me a lot of lessons, of course and also help us build a good social network, you know, I actually met my co founder. In our first startup. And, you know, after the first startup was acquired, you know, we were thinking about something new. And we came up with this idea and say hey, you know, let's do something even more crazy than the first startup, you know, think about all the voice information in the world. How can we capture it, how can we organize it, how can we make it available to everybody. So it's a, it's a very audacious plan, actually, you know, when we started, we didn't have any speech recognition system, so we actually say, Let's build it ourselves, because I would test everything, and nothing worked well, you know, we tested Google with technically Microsoft. It just the accuracy was so bad. And then we say hey, that's actually opportunity for us, because, you know, nobody else is able to do this well. So yeah, delude ourselves.
So this is a good segue to the next question, as you mentioned that you own your own speech engine, and NLP tech stack. Use started because out of necessity, because there's no other good engine available, but nowadays the, the other speech engine are also getting better. And we know there are companies taking advantage of that building their own applications on top of, you know, third party speech engine, especially from big tech companies. What do you think is the advantage of owning your speech engine NLP models, especially you know, in this sector.
Yeah, it's actually really critical. You know this is a common question VCs, ask us all the time, you know, there's several things. One is how do you compete with the big tech. The second is oh, you know what if, you know, this new startup just use Google API. They don't need to build their speech recognition engine. For the second case, Secondly, they don't understand that there's a number of big disadvantages of using a third party API number one is actually cost you have to pay them, which is actually very expensive, that will limit it, your use case and will make your product, much more expensive for your customers. Number two is actually, you need to send all the data to Google. That's actually there's some concern for, for a lot of enterprises, they don't want to send data, or Microsoft. The third thing is actually if you don't control the stack, there are very limited number of things you can do. For example, customer coverage is important feature otters, provide. it allows you to add special names, special terminologies acronyms or medical terms into the author dictionary. Other API's doesn't allow you to do that. It's actually really critical, you know, people's names really important to recognize correctly, and new words are amended all the time, actually I look at the transcript on YouTube. I think just a while ago and look at it, they still don't recognize COVID-19. But if you look at otter otter can recognize it, so he learns it constantly learning new words, it. If you watch the live transcript, you can see it's actually also constantly correcting words when IDI hears more words.
Yeah that's really awesome. And the other thing I noticed that. Now, there are startups actually using otter as the third speech engine to calibrate or even build their services. Do you have any plans to, you know, open otter as a sort of API driven services to other companies, or is this something that you're not really interested in.
Yeah, we've been asked that question all the time as well. You know people did recognize we have higher accuracy, high performance in terms of low latency. At this moment, That's not our priority to open an API. Maybe it's some later time. We are a focus is the author product and the end user experience. We want to create a new collaboration platform on top of honor. So that, you know, the number one use case is for meetings to, you know, for people taking meeting notes share meeting information, get the meeting information as easy as possible, whether you're in the meeting, or not, you can get, actually, you know with otter you can join three zoom in at the same time, and you may listen to only one and just watch the live transcript from the other two. So allow people to a get actually even more efficient.
Yeah, I'm not sure about multicast at three meetings but we are a loyal user of Otter, and we just had our LP meeting because cross off multiple funds we have more than 100, investors, and only half of them are able to join live, so we were able to use auto generated transcripts to send to them afterwards, and they are very impressed by, you know, this kind of quality, and this, our ability to generate the transcript very quickly. So definitely very happy about, you know, Otter, even as a normal user, but as an investor, we also know that yeah, you know otter has seen rapid growth in the last four months, any insights on the you know, this tremendous growth during the COVID pandemic, and any any learnings you You're, you're able to share.
Um, yeah, actually even before COVID We were already growing pretty well. Then, after COVID happen, you know, as I mentioned earlier, the people just are having too many meetings. And with resume fatigue and also because they're busy everybody is working from home, said to harder to communicate. Um, and, in a people actually often, you know, working from home often distracted by other things as well, especially that some people have kids, some time, there's something happening in somebody's ringing your doorbell. You need to leave your desk in when you come back and people, what did people talk, Lee were absent out other action to help you handle all those cases. And, of course, we said, you know for international companies, people are located in different timezone so the otter, allow you to get information at a different time really efficiently actually if you've missed a meeting. And you use otter. If it's a one hour meeting, you actually don't need to spend one full hour to get all the information. The search capability and help you find the information you're interested in really fast. Now also if somebody else really already highlighted certain things for you, they say Hey, Sam, you just need to look at this paragraph, that's all you need to know. You don't have to read or listen to the other parts. So that's actually help people save time. So, the, the value is getting stronger and stronger and more visible to people. That's why more and more people are adopting it, of course, means is we got a lot of organic growth, because the people tell each other about the order. Once they use it themselves, they see how this is so useful. Now I'm going to use it myself, for my next meeting and they share it with their meeting attendees, so they just where we grow. Yeah,
that's really powerful. I think there's a implied facts in your logic here is that we read much faster than we listen. So so I can scan through the transcription, with all without highlighting with highlighting even faster but even without it, I can, you know, scan it faster. Now with the power of so much data that you have processed. What are the other things that you think in the in the you know, future vision of Otter yet besides just being the best transcription engine and the application of course.
Yeah, there's definitely tons of new things to do, you know, first of all, the We Still, you know, although we have really good accuracy but it's never good enough. You know people are definitely very critical or you know people always expect better. The, especially, you know, for about 100 back on noise better handle. You know, the oftentimes you have someone dallying from a phone line, their voice quality is lower, or their home, Wi Fi network isn't very good, so the voice is distorted. How do we handle that. Of course Mr command is there Oh, including myself. It's a immigrant, you know, we have accents, you know there are people coming from Asia come from Europe, or other places, everybody, you have different accent, how do you handle that really well. And then there's the contextual intelligence, you know, if you cannot hear any words correctly can you fill in the blank, the bass sound the other words you hear, you know, it's actually in the future, can the AI can actually interpret or understand words, even better than human being, because it has all the historical information. And then the next is you know, once you have the transcript. What can you do with it. How much can you understand it. Can you understand the questions can you understand the numbers, how do they relate to each other. Of course, everybody asked us, Hey can you generate automatically generate the meeting summary. To be honest, you know, you may see on the internet some, someone will claim they can generate it is actually still a and solve the problem. It definitely needs a lot more work to be able to generate, just a few bullet points in this, these are all the important points from this meeting. Yeah, we that those are all active areas we're working on. So this is, this
is really exciting. You know, speaking of processing the data a lot of people know, there's another company called a gong, they're valued at several billion dollars in my mind, you know the the tech not the core technology is probably, you know, not as good as otter but they they find a good it's a vertical application. They're they're able to, you know, attack the sales use case. And, you know, mashing the template and matching some keywords. What's your view of, you know, finding this kind of vertical niche and then be able to to increase the value versus what you're doing.
Yeah, God is a great company, they, Stephanie, as you mentioned, it's a great use case for salespeople. But we decided to take a very different approach, we like to target meetings in general, you know, for, for this conversation, this is now a sales call. And, you know the number one reason we take this approach is that we see actually the, the potential market size is bigger, you know, you know, they, they went deeper into the sales vertical but we Carver a lot more, more types of meetings could be a project meeting it could be an interview. It could be a webinar, you know, it could be a virtual events. So otter covers all of this really well. Another use case is actually education online education. You know with the COVID, a lot of classes are moved on to zoom or Google Meet or Microsoft Teams otter actually help them take lecture notes really easily so the. We see our user base is way bigger. So yeah, it's very different strategy.
Yeah, I love it, I, my main point is that otter I believe still undervalued because, you know, I think you guys should be becoming a multi billion dollar company soon. But I love what you're doing. And, you know, obviously the integration with Zoom has been around for a couple years now. This year you also added the integration with Google, Meet. Even though I'm not a, you know, big user of Google Meet but I know a lot of people using it. And so how does it feel dance. Dancing with elephants go these companies are obviously much bigger one is, you know, close to $100 billion. The other is more than a trillion dollar, and then you have Microsoft on the side, playing toying around transcription engine of their own, even though it's a garage, you know, project, I don't pay too much attention to that quality. What do you feel about you know, this partnership with the giants, and are you worried that one day they may come and try to compete or acquire you, you know, what's your view.
Well this is, this is the
once is a capitalism. You have to compete. You just have to assume that people want to crush it. That's just a reality, you know we we accept that, you know, we just need you need to get comfortable with being uncomfortable. Microsoft, you know, Microsoft actually made a lot of great progress we actually, they see, we see their their accuracy has been improving. Pretty fast as well. So, there will always be competitors, you just have to move faster than all of them you have to be smarter, and also on both the core technology and the product offering. On the product side, how do you make that easier. How do you make the collaboration better, how do you let people interact with your product every day, multiple times. So, you need to have bolts, you know, in, that's our strategy, you know, have both the core AI technology, and the, the best user experience, you know, another thing I actually want to add is that we have a good mobile experience as well. So, on the go, you can get your or your meat, making notes on your iPhone and your Android. Actually, right now in Zune. They actually don't have that.
Yeah, I use the mobile app, a lot, it's really convenient for me to just one tap, started the transcription and doesn't matter whose meeting it is. And then if I have to, you know, step away I can always come back. Are you using the website to do the otter and then use the mobile phone to view it when I step away for a few minutes.
So,
this leads to my next question. I always feel that there's something similar between otter and zoom in that, that you really care about the product experience, but you actually are a technologist at heart just like Eric yuan or Eric is a technologist, but he is laser focus he's maniacally focused zoom on the Zoom user experience, and he made a lot of important trade offs and made a lot of important decisions along the way some helped him at the time some hurt him at the time, but ultimately resulted in a great user experience and focus, he's not. He refused to be distracted. So, in some way. I feel you're like that. So, can you share with us, like how do you, you know, go from a technologist to someone who can really think deeply about the product and make those trade offs and make those decisions.
First of all actually know we build this service, for two reasons one is that we see that we have the need ourself. You know I have so many meetings every day, I have bad memory I can remember things. And when I communicate and it's I, you know, for example, I talked to this VC. Oh, what did he say, you know I need to share with my co founder, what, you know, sometimes I don't remember all the things, you know, so the sharing part is, is really critical. So, the recording product, actually. I myself, although, you know, I, my thesis, Stanford was on distributed system, actually before that I was actually really into computer graphics, animation and human computer interaction, I was in actually Stanford graphics in HCI lab for quite some time. Even before that actually I was at Silicon Graphics was coding X Window. You know, it's actually I was having a lot of fun. You know, building animation we transition. It's helped me create you eyes. Interesting user interfaces. It's so in sandwich I'm actually fascinated by, you know any good user experience and user interface myself. And again, you know, we eat our own dog food in our own county, We use the product every day. So we get to feel the pain, if anything, doesn't work. So we discover any bugs, we see in you know flow that doesn't work. Um, I guess this is one advantage if, you know, I guess, you mentioned, Eric, Eric, he uses zoom himself every day so I'm sure he you know he discovered a lot of problems about using it and he will work with their product team their engineering team. So when you have a, you're the founder or CEO who just who is so obsessed with the product, I think that's a good, good combination. It's now like, oh, you build something that you never use yourself, then it's hard to really feel the pain and feel the need to improve it, and to innovate, you know, we, we come up with new ideas all the time and we also have brainstorm sessions in our own team you know everybody actually is really passionate about a product, and then they think about new ways to improve it, new ways to provide new value to people.
Yeah, so, to follow up little bit is that, you know, coming from, you know, startups, as well as large company, myself. What is the behind scene product feature discussion looks like. So, you know, maybe your team members saw some cool features, because I know I'm guilty of sometimes sending some features to us that hey look, they're doing this and they're doing XYZ like you could do that too. And how do you control this process so that you have a methodical way to decide what feature to build what features, not to them.
That's a good question. It's hard question actually, I wouldn't say we know exactly it's a lot of time it's, it's our instinct. And, you know, some sometimes people say hey we just would look at the current user data and see how they do it. It's, it's sometimes the right way to do it, but sometimes you, we need to know how to interpret the data, because it's an honor. It's a new product. The, the new product people don't necessarily know how to use it yet. You know in what situation do they use it.
It's,
it's hard to have a general rule, you know, A lot of times you say hey, nobody uses AI so it's not useful, but it's not necessarily a truth. Sometimes you just have to follow your gut feeling.
Yeah, I think you have a pretty good gut feeling my observation, at least is that I feel like every time I ask you a question, you're really thinking, and you're thinking about. From what I can see is that, is there a better use of my resource than doing this, versus doing that, and then you usually come up with, as I observed, like based on what is the best use of your resource because as a startup, even though you just raise $15 million. There's still a limit, limited resource that you can deploy, and a limited number of engineers, you can hire. So, at least I feel like, you know whenever I talk to you, you don't just give me the brush off you don't give me the answer that I want to hear but you really think about it and making those trade offs that
we don't claim we know all the answers of course something, Sometimes it's trial and error, we try something it may not always work, then we say, Hey, okay, that doesn't work then we have to do it in a different way, of course in the AP test, sometimes, sometimes you just, it's really hard to say. Yeah, yeah well you met something new, you can't. Yeah, you can't expect that everybody will understand it right away.
And that's really the honest answer I think I really appreciate that. Now, like more and more people are getting vaccinated, and there's a general expectation or hope that the business will open again we can get back to office says now added maybe quarter capacity. Now in a few months at half capacity or by end of the year we have a beat for strength back. So, some people associate otter, zoom, and a few other things as the sorta like a pandemic, you know company. Obviously I don't fully agree with that. What are the, you know use case that you see that will stick with people, even after they go back to office.
I think this is the new norm, even though some people will return to Office, some people will still be working remotely. Maybe not 100% of their time. So a lot of meeting will still be hybrid, some people may be local, in the conference room, oftentimes there are people remote as well. And, you know, the different timezone problem, it's a very international company that's very common separately and this is where actually our strategy is, is interesting because we actually have before COVID You know, we actually put already put in a lot of effort in mobile. It means that this is actually our initial very ambitious goal we say hey, no matter where your conversation happened author can help you capture that and share it. So that's why we, we put in a lot of our effort building the mobile apps on iPhone and Android, and desktop, you know there right now you can run otter in web browser as well. So, in Australia, too, to plant this virtual meeting with real meetings, actually.
Yeah so
person meetings.
Yeah, that's a really good point, because you know when we first met. There is no pandemic right we are in person, and I immediately see the value of Otter, even for in person meetings or phone calls, because you know, I normally take good notes by notice for some of the research cause even with my good notes, I'm only capturing, you know, a little over a half of the knowledge, information, maybe 60% Maybe if I try really hard to 70%. When I go back from my transcript, I pick up additional 30 40% that's really valuable not only to me but especially to other people because I took notes on the things that I heard, and pay attention to, but I missed the things that I'm that are my blind spot. So, definitely, I feel like this is true no matter if it's an in person meeting or or remote.
Yeah, for our own company, as I mentioned and we eat our own dog food actually all my meetings in the last five years are in order. You know, I can search any time, any place you know instantly. Within seconds, I can find. Okay, what did that customer ask or any problem they have, you know, this is actually really important when we listen to customers, sometimes you have to listen to them several times to fully understand. And also, when you get good feedback, you want to share that with your team as well, to let them hear from the customers firsthand. This is where actually you know their sharing and collaboration is so important. It's so it's not a single user. Note only, although some people use it that way but we see it's more powerful when the entire team use it together.
Yeah, I see a lot of people asking a question in the chat already one of the question I also am interested is about languages, and, you know, related to that translation, you know, obviously, English is the biggest language in the world, it's the, you know, common language for most of the international business, but there are other languages in the world, what what is your view and what is your timetable that that you know that auto will attack other languages, in the future. Yeah, we
cannot give an exact timeline. This is also a common question, we're being asked all the time. Other languages and translation, this, this is our no future roadmap. This is why also, you know, we're looking for more people to join us, because there's so many interesting problem to work. You know, additional languages and translation. Yeah, these are the future projects. Yeah, no more interesting, you know just can you, you know sometimes people use two languages at the same time, in one may think, so can you automatically detect that automatic. You know transcribe it or right away. And of course mean for international meetings to try and translation will always be useful.
Yeah.
So you mentioned you just raised a big round. Obviously, congratulations on the successful fundraise and also enables you to hire more people, as you mentioned you are hiring. You know, actually, you know, a lot of big tech companies still not fully open up the floodgates to hire yet. People are still watching so it's really good news that for you to hire, and what type of people you're you're looking for, you know, obviously this is an exciting time to join orders that you know high growth trajectory, there's a lot of new things potentially people can work on but uh you know can we maybe elaborate like the different type of people you may be hiring.
Yeah we're hiring across the board on the engineering side we're looking for machine learning people deep learning, and both speech recognition, natural language processing. Actually we're looking for vision people as well. If you have experience there, there'll be great backend engineers. Front end engineers DevOps, because we are processing, huge amount of data. Now the voice recording for one hour meeting could be 100 megabytes, just for one meeting, so we need to make sure that database is fast, it's reliable. And, you know, search, you know, how can you search, so many meeting transcript really fast. How do you index it. You know, read it in under rank or things the right way, based on relevancy. And then, yeah, we're on the growth side the sales, marketing, we're also hiring so across the board.
Cool, cool. I think that's a good segue maybe to switch to open q&a Sophia. Do you think we're ready to do to switch to q&a
Yeah, yeah, thank you so much for this great conversation, I think there's a lot, and great insights. So I think there were already some questions in the chat. And Simon has been very actively answering that they're through that s AP
product, and Simon did a lot. Good in the past user experience. So, enhanced
Yeah, That's very, I can easily tell from the active response in the chat, so maybe we can start from there and people, if you have if you have questions you can just unmute yourself and we can make it very interactive and so that's C, see I mean, sure. Do you do have any questions to ask Sam.
Yeah, I'm a medical oncologist, so my biggest interest is how to kind of translate my conversation with patient, and, and use that to build my note, because that's the most time consuming and labor consuming part for a physician, I think, AI especially pulled out I order we'll have a extremely huge help for our physicians really I think this is a good field for you guys to explore more
of you now said thanks for the question. Actually, you know, before Sam. So because I already saw Simon's answer on this one. I agree with him that you know otter is a general purpose platform, obviously you can use that to build your transcript. But we haven't have another investment in our portfolio called Deep scribe. They specialize in, you know, note taking for patient doctor conversations. So their goal is very different from otter, so they're not really aiming for transcription accuracy, they're aiming for the bullet points that you can put into electronic medical records that diagnosis symptoms prescriptions and, you know, name of the drug and these kind of things so it's a more specialized play so if you're interested you can look it up deep scribe dot, dot, a common belief, yeah TIPS
TIPS crap.
Yeah, so it's like deep, very deep, and the scribe Scribe is like, you know, transparent scribe. I think it's dot AI better right.
Let's see, okay, yeah,
yeah. Sam, what do you think, you know, I jumped in, for for this one so I don't know if you have a different thoughts on that. Oh,
the.
Actually, right now we're not doing anything really. We're not doing anything directly with healthcare. Although, we actually have seen doctors use in order themselves, although we didn't really target them directly, but people, doctors or healthcare provider discover alter themselves actually on the flip side, the lot of people when they see their own doctors they find funny useful to use order we actually have heard a lot of stories from our users when they visit doctors, they cannot remember all the instructions or the terminologies. Doctors told them they added, they just turn on otter and take notes with otter, and they go home then they read this and we actually had a lot of users share those stories with us.
Yeah,
yeah, okay. Yeah, that'd be very helpful for the phase patients as well, because a lot of time, how the patient may not be able to remember everything and they would like to share with their family members, and not all of them will be there with them, so that'd be a great help. But that will may, may have some liability issues. So, maybe some some limitation or disclosure I have to be used for those transcripts. So, Yeah, another issue to remember
security that's a very important. You want to take seriously, you know, this is why, you know, we actually were ready. Have you all have all the data encrypted. And of course you know computing more security measures to make sure that data is all confidential and secure.
Great, yeah. Thanks so much,
that our security engineers were also hard.
I think this also answers, one question from the chat, regarding privacy that who asks about how do you handle the risk that meeting transcription service becomes a commodity. Do you have anything to add to that.
A security wise is actually, of course in the meeting, and voice is very sensitive information, but essentially it's actually not so much different than your email, your document your Slack messages, essentially is the same. So there, they're they're a common practice. You know, in terms of encrypt information. So, the general model, it's actually pretty well understood.
Yeah I think whose second question is about, what if this transcription service becomes commodity. My interpretation is that, you know, what if big tech, just give away that kind of transcription capabilities.
Well it's a touch on that earlier, is both the technology and the product and the technology itself, it's actually to be honest, I heard this five years ago and say hey, it's all commodity now. And, but, you know, I say mentioned earlier, it's a still an unsolved problem, how do you understand that what kind of acts and how do you understand all kinds of words. What if it's not clear. And of course on top of it, it's all about the all kinds of analytics and understanding of the content is still an unsolved problem. And in a commodity or not it's actually a very expensive operation. This is why when you use a Google API they charge you a lot of money, actually.
I see. So you're basically. I think what you're implying is that by owning your own tech stack not only you can achieve greater results but you can also optimize your cost structure, much better than another.
Yeah, both. Yeah, and also it is by owning this it's actually allows us to provide much higher quality. It's actually. Another important part is learning, dynamic and learning in a longer term. When you talk to author often Audra can learn the way you speak. It knows, it can almost predict what you're going to say, even before you say it. It knows how you pronounce words. So it's getting more personalized.
Yeah, there's a new question. Yeah, I think we've kind of covered it, but this specific question is about otter does our have a planet enter the Chinese market, I think, you know, it's sort of correlated to this language question.
A short term we don't want to do that. Yeah, there,
there are other languages that are more of a reason based on the focus of otter that the English market is big enough, or is it something else.
Yeah, English market by self is really big enough, in a way, all penetration is still quite small, you know there's. See 100 times bigger opportunities is still we still untouched. Just for English.
Yep. Yeah. So, I think you and Simon both said the same thing. Yeah,
that's another another market you know we, you know, we got a strategic investment from NTT Docomo. It's, you know what, we already have quite some users in Japan, using orders English word, and of course when they're interested in Japanese. Transcription as well. So that's actually potentially very interesting market for us. And we have really have a good partner there in the great largest telecom company in in Japan and Japanese someone
raised their hand, I think, Chris, you want to unmute yourself and just asked directly.
Yeah, thanks. I do have a question, so I'm a lawyer, user of voter, I use all your entities to four hours a day for all my meetings, as a user, like if I identify a spelling error or terminology that is not recognized by algorithms while alters transcribing and wondering whether otter is planning to support manual tax correction by the user on the fly, and also will alter consider keeping human in the loop in the future. Thanks.
Oh, you mean correction in real time.
Yes, yes. Oh,
I see, yeah right now, the, you can correct words afterwards. It's, yeah we this is a common feature a lot of people ask. We're still trying to figure out the right prioritization, this is definitely something we want to do but we cannot promise when we'll do it. The.
Thanks. Yeah.
So I assume that if we started correct these words and adding custom vocabulary, it would affect future recordings or future transcriptions, right,
yeah. Yeah, that's something we're building to automatically learn from the corrections. These are the things in a weak if because we control the full stack so we can do all of this, but if you just use a third party API, you won't be able to support this kind of learning curve of variety is actually a really important part.
We have another question from the chat. Sure, Leo is asking, in terms of data privacy, what are the advantages and disadvantages compared to big tech
advantage disadvantages, big tech, of course, they have more people than us at this moment, they can.
As I mentioned, I mean, this is a, you know, security definitely takes a lot of engineering work. It's, it's relatively a well understood, it just, you just need to put in a lot of work. There's, there's not.
Blue is asking specific about data privacy. That changes your, your answer and.
