Raul is a Techstars alumnus (NYC’18) with 15+ years of expertise in software development, product management and business management, prior to TypingDNA he co-founded other startups, helped launch several other innovative software products and coded core software components used by millions.


Ciprian Borodescu: I’m here with Raul Popa, CEO, Co-founder and Data Scientist at TypingDNA – a company that recognizes people by the way they type on computers and mobile devices. I’m super excited, and it’s an honor to have you on this podcast. Thank you so much for being here.

Raul Popa: Thank you very much for the invite. I’m happy to be here with you and your audience as well.

Ciprian Borodescu: So, you’re now in New York. What are you up to these days?

Raul Popa: Mostly working from home, like pretty much everyone else. At TypingDNA, we recently raised a new round and we’re hiring more people and things are going a little bit in that direction. So, a lot of onboarding new people, working on a growth plan, those kinds of things.

Ciprian Borodescu: So yeah, at the end of 2019, beginning of this year, TypingDNA raised 7 million in Series A, a round that was led by Google’s venture group Radian Ventures. And I just wanted to ask you, what was the impact of closing this round for you and for the team?

Raul Popa: I think it was a good opportunity from the very beginning, when we met with Radian Ventures to actually start thinking about raising a new round. Initially, we didn’t plan to do that. We planned to sort of do a round a little bit later. But at some point, after discussing with Radian – or while we were discussing with Radian, we realized that it would be a good opportunity to do that then, and then, you know, I think it was a good decision because after we raised that round, two months in we had this lockdown, and I think it would have been much harder to raise a new round, despite the fact that we actually have really good growth. We had record April, record March. I’m really grateful for how the company is going and how things are going on our side. So, things are changing for the good. But as I told you earlier, we have more resources, more time into onboarding people. I think working remotely is hard if you work with people that used to work on site. But working remotely with people that you never even met, you just hired them, and you have to onboard them remotely – that’s even harder.

Ciprian Borodescu: Yeah, that’s challenging.

Raul Popa: So if you, for example, worked previously with lots of freelancers then probably it’s easier for you. But for us, it wasn’t like that. We had three offices – two in Romania, one in the US and we always had people in an office talking to people in another office. So, you know, the remote idea was there, but the people were not working completely remote from their own home most of the time, and we actually had a chance to meet with everybody. Now we don’t, so we’re kind of in a little bit different situation. But things are working out. Other companies did this very well, and we’re trying to learn from them.

Ciprian Borodescu: That’s quite a challenge. And I was planning on actually asking this question a little bit later on the podcast, but I’m going to ask you now because it ties really well to what you said. How do you go about scaling a team, from the founding team, to the first employees and then to over 25 or 50 people? And I’m mostly interested from a personal point of view because I’m also an entrepreneur, a founder, and a CEO. What are some of the stages and the challenges you’re going through as the company grows?

Raul Popa: This is a very good question. I still don’t have all the answers, to be honest. I’ve been involved in other startups and co-founded other companies and also being involved as an employee in other companies – early employee. So, I kind of understand the dynamics. I’ve seen the dynamics at work in different stages. I think there are stages and stages. At first, it’s one or two or three founders that build something unique. And, initially, your first goal is to convince co-founders to join if you’re the main founder, if you’re the first person with the idea that wants to build something. And it was my case – I guess it’s your case, as well. And then, first, you have to convince some people to go along with you without being paid, because at first, you don’t pay them. So basically, they have to give up their job or whatever they’re doing and, for a while, work on the promise of the idea that maybe we will get investment. So the second phase is you want to get investment. To get investment, you have to build a business plan and think about how this can turn into a profitable business or a business that you can exit from at some point – those kinds of things. Then, once you have the first round investment, the next goal is to build something because you initially have a prototype, so to build something, to build Salesforce, or to try to sell yourself – those kinds of things. So, it really depends on what the goals are.

Raul Popa: But at some point, you reach a level where you’re 20 people, let’s say. Until then, every person wore a number of hats – so not just one hat. So rarely, rarely, at that point, you have people that are extremely specialized, even developers who work in multiple technologies. So you kind of want to have people who are able to do multiple things at once. But then, from 20 people to 100-150, you have a specialization, if you want – you know, growth – in which people get specialized in a certain thing. And they also start grouping around projects. So grouping around departments like marketing, development, and sales doesn’t work anymore so you want to group them around projects, around clients, around types of verticals, around these kinds of use cases, and so forth. So there’s a new layer of management that comes in, like Event Management, Product Management, this kind of thing – a managed project, more or less. And the thing is, this is a very, very interesting stage. It’s not the stage where you transform your startup into a corporation. It’s a stage where you still connect with everyone and you kind of know everyone. It’s called a ‘tribe size’. Basically, in a tribe, everyone knows everyone or at least met with everyone at least once a year. This is the definition. Once you reach 200 to 250 people, you’re over that size, and you cannot do that anymore – and from there on, it becomes a corporation. And I haven’t been there, I don’t know exactly what happens after you’re 150, that’s completely new for me. But being 50-100 people and so forth, I’ve been before and I have good knowledge on what to do right now.

Raul Popa: And it’s actually very, very challenging and to do this remotely is even more challenging. So, now we are in a challenge to grow to about 50 people this year. That’s what we want – about doubling the size that we are at this point. And most people are actually either functional – the really important people in some places where we didn’t have them – or they’re very senior, or as senior as you can get at this stage. So, a lot of these new employees would have never joined a company at the previous stage because they had a very good salary and position and probably a much better-established company. And this is how it typically goes. So, the people that you hire initially are more risk-tolerant, they want to work in a fast-growing environment, in a fast-paced environment, in a startup. Then, as the company grows, you start hiring people that are more specialized, do very well one particular thing or a few particular things. They also come at a different cost and with a different set of knowledge and skills. Very interesting.

Ciprian Borodescu: So, in 2018, you guys graduated from Techstars, New York – and before asking how your experience has been going through the accelerator, I just want to share a question that we got during Techstars Montreal AI last year: who is the CPO – the Chief Product Officer? This ties to the question that I have for you. When do you think it’s a good time to hire a Product Owner or a Product Manager?

Raul Popa: I think it’s very important to understand the role of the product first. A product person is probably the second most important person after the CEO. Actually, the product person is the CEO of the product.

Ciprian Borodescu: The mini CEO.

Raul Popa: Yeah. And actually, I’ve been a Product Manager for seven-plus years. Before starting TypingDNA, I worked a lot as a Product Manager, Head of Product. Of course, I also did data science, machine learning – those kinds of things. And before that, I was an engineer for 7-10 years and managed engineers. But I transitioned to Product Manager, and I do understand the role of product. Actually, I worked with multiple products, and I was the sort of Chief of Product, before. I think in a startup, when you’re, like, less than five people, you don’t really need a Chief Product. Basically, one or two or three founders should be able to understand the client very well. Otherwise, how do you create a company if you don’t understand the client? So at first, you have to sort of understand the client, and the best way to understand it is to actually be one. And in most of the startups that succeed, one of the founders really understands the client because they were in their shoes. And it’s a better role than having a product person that is hired to do product. As the company grows, at some point, you really need people to focus on particular types of clients. Let’s say, if you go after three verticals, you want to have a person who focuses only on one vertical and wants to understand how that type of client works with your technology, and what kind of things they want to see in a dashboard, those kinds of things. And, you know, that’s when you need a product manager.

Raul Popa: About the product owner thing – which is a different thing – basically, if you’re working in a Scrum management framework, then you will need somebody to manage the requirements, and to do acceptance later on, and so forth. And that would be the product owner – actually, to own their responsibilities around building or improving a product. And sometimes, if you have a smaller team, so let’s say you have five people, you don’t have seven people in a group that will develop a project like you need in a Scrum project. So a person does pretty much everything. You know, you have a Scrum master who’s also a product owner, who’s also maybe the chief architect, and also does some development, maybe here and there and testing and all those things and talks with the clients, and so forth. And so, from my experience, the best product managers that we had – and since we’re from Romania, you don’t have a lot of schools that prepare you to be a Product Manager. Actually, I don’t know if there’s any school in Romania that prepares people for product management.

Ciprian Borodescu: No, absolutely. Just working with different startups and having the chance to do that.

Raul Popa: Yes, and what I saw before is that the best product managers come from mainly two directions. It’s either engineers who have a very good idea about how clients behave or what clients need – these are very important. And so their engineers understand very well the insights of what they’re building, but they also have a taste for what the customers need. Or it comes from the other side, from the customer support, customer success, sales’ perspective where you understand very, very well the customer, but you don’t understand the insights and then you have to work with somebody more technical to actually build a product to actually see that can work. For companies like TypingDNA – which is deep core technology, it’s a deep tech, we actually built a completely innovative technology – there’s no way a product manager can come from sales. Yeah, you can say, “Hey, you can use it in this kind of way.” I can use a button on which I press and then it rains, but it doesn’t mean that that’s possible to build. So you want a person that understands how the product can be built as well. So you want that person to understand these two things: what the customer needs and if that can be built. They’re very important. I think at the beginning, only the founders are able to do this. Later on, I think you can you can get specialized people that probably, as you said, they devote themselves in previous companies starting, I think, usually from these roles. I’ve seen product managers – one person in particular – that started from a marketing and communication role, but also it was a marketing and communication product. So in a way, she understood very well the customer. So yeah.

Ciprian Borodescu: That’s really well said, and I totally agree with that. Cool. What are some of the do’s and don’ts of building an AI product based on your experience at TypingDNA?

Raul Popa: I think it’s overrated the fact that you’re building an AI product. A lot of companies are bragging around the fact that they’re using AI, machine learning to solve a certain problem whereas, they can do well without actually using AI and just use a rule base or real people doing something. They’re not always solving a problem that needs AI, or where AI would be really useful. And so, I think out of, like, 100 startups that are bragging they’re AI startups, for sure less than 50%, actually, for them it’s crucial to use AI. For us, we are an AI company. Without AI, we would not be able to do what we do. It’s really impossible. But even then, we don’t use the most sophisticated AI technologies out there. We do use some very complex ones, but not the most sophisticated ones. And it’s also, I think, overrated, the fact that once you build an AI product, a lot of investors, for example, talk a lot about things like defensibility. So, how defensible is your technology? How hard is it to replicate?

Ciprian Borodescu: The classical question.

Raul Popa: Yeah. And I’ve seen founders trying to come up with arguments that they made the technology from an engineering point of view that is so superior, that nobody else can do that. And I think that’s bullshit.

Ciprian Borodescu: I think in today’s world, it’s really hard to have that position. Yeah.

Raul Popa: I think it’s ignorant and arrogant to say anything like that. What somebody can do, another person can do for sure. Maybe it takes time, maybe it takes more money, maybe it takes different brains, but for sure somebody else can do it. I think the best defensibility lies in other spaces than the actual technology. But what you do have sometimes is a leg up, you have like a window of opportunity, you have the technology ready now, and others don’t have the technology now, so you can reach clients that need the technology now; even those that need the technology in six months, or a year or so. For others, it will take a lot of time to get there. I’ve seen a lot of companies bragging about the fact that they have tons of data. As a start, you don’t really have a lot of data. So, it’s really hard first to brag about and it’s really hard to be considered a defensibility tactic – usually bigger companies have a lot of data. And what we see a lot is that a lot of new technologies come up that you can get to really good results with less amount of data. So, that is what makes the startups fight against big corporations like Apple, Google, Amazon, and so forth. There was this thesis that, you know, without a lot of data, you cannot build a very good algorithm. And I think this is turning upside down and it’s not true. It’s true in some cases – it’s mostly not true lately. I’ve seen a lot of startups challenging this assumption and I think because of that, not even big companies can argue that they have superiority in terms of defensibility. The best defensibility that you have is to move fast. That’s really what it is about. So if you’re able to move fast, that is really, really what you want to see. You want to see that your numbers are growing, you want to see that every week, every month or so you’re doing something new, you’re doing something in the right direction, and others are not doing that. Or even if others are doing that, you’re motivated to do that for your customers and your target audience. That is something that lights people up. Like, they want to be involved in these kinds of startups, investors want to put their money into these kinds of startups, you know, clients want to work with these companies that are doing all sorts of things, and moving the needle forward, and so forth. That’s, in my opinion, what matters. So, the biggest no for AI companies: don’t brag that much that you have an AI product. Try to use that to your advantage and focus on moving faster, not just saying “Hey, we have this technology. Nobody else is able to do it, so we are safe.” Nobody’s safe. We’re in a moment where technology innovation is at its fastest speed. We never moved as fast as we move at this point, especially in AI – and AI is the center of the speed in innovation. So, really, you cannot claim that you’re safe if you’re doing an AI startup. You have to continuously evolve. That’s my opinion.

Ciprian Borodescu: Absolutely, I am totally behind that. And I think we’re now in an age of actually applying AI and a lot of the research has been done in the past 50 years or something like that and we’re just applying that, and we need to find new ways of applying machine learning or statistics or AI, whatever you want to call it. Okay, so are you guys still working on improving the technology powering TypingDNA right now, or today, the focus is mostly on the applied part and the use cases and the product, like you mentioned?

Raul Popa: We’re doing both. Our vision is very simple. Our vision is that people are typing more than they used to, and definitely more than they talk at this point. And so, we’re communicating through typing more than we used to. That’s very important. More than we talk. And if you look at, like, 100 years ago, less than 5% of people were able to read and write – just to read, over 90% of people were illiterate. And 1000 years ago, even more. So basically, the entire human race was used to communicate orally, and that was it. Our statistics show that around 2014, we were at 50-50 and the trend changed in a matter of few years. In 2017 already 75 to 80%; in 2018 80% of people preferred texting on a mobile phone compared to talking on a mobile phone with another person. So, things like WhatsApp, Slack, email, Messenger, prevail at this point. People want to use them. And there’s also this asynchronous nature of communication, where I type my message when I have time, and you read it and reply when you have time. This allows for more conversations, basically, that can be compressed in the same amount of time. If we do this podcast right now, if you need that person to be ready when you’re ready to talk, then it takes a lot of back and forth to put that together. If it’s more people, even more, it’s very, very complicated. And I’m not happy, or this is not necessarily a good thing as just, this is happening, and we cannot say it’s not, and we cannot shy away from this. So, people are communicating through typing more and more and this becomes the standard method of communication. And why this is important? It’s important because not a lot of people agreed with us at the beginning. If you remember, three, four, or five years ago, a lot of hype was about voice assistants like Alexa, like Siri. People thought, especially older people – and investors are usually older people, and they thought that, hey, the world is the world of assistance where you talk with an assistant in your car, under the shower, and so forth. Younger people don’t drive, younger people have a lot of time and have a mobile phone in their hand. And you know, I have a daughter, and we’re in the same room, sometimes she will send me a WhatsApp instead of talking with me. So things are changing so much that young people don’t really like to talk even in a closed environment. So I think this was overlooked a lot.

Raul Popa: And at TypingDNA, this is where I’m going with this idea. At TypingDNA, we believe that the technology that we’re building will bring some of the biometrics that we’re lacking in the typing communication world. In a voice-based communication or face-to-face communication, we see the face, we hear the voice – so, those are biometrics that we use to recognize whether it’s the same person who says they are, that we’re talking with. In typing-based communication, we lack that. So, let’s say you communicate with somebody on Slack, and they ask you for credentials from something that you’re doing and it’s a colleague of yours. What do you do? Do you ask them for a password or something? How do you make sure that it’s not somebody who stole their phone, at the subway, who is somebody who wants to get into your account, who has a very targeted attack, and they want to get in, and this is the way they’re doing it? How do you know that’s not a false claim, a bad actor? It’s really hard. And with big companies, the problem increases exponentially. So, you want to be able to verify somebody’s identity everywhere they type continuously, not just when they log in. So, there is where our vision sort of lies. And because of that, the products that we have right now are not necessarily enough for all these things that we want to tackle. So, some of the things that we want to tackle is continuous communication through typing on mobile devices, on desktops, and all sorts of devices and operating systems and applications, and so forth. And then different APIs that will tackle different things. So we’re only at the beginning. We are the best, I would say, in what we’re building, but we’re only at the beginning of this new typing-based communication era in which we lack any form of biometrics, and TypingDNA does that very well and wants to do that very well in the future – to be the layer that we need for additional security and protection. So it’s maybe a little bit longer answer, but I think that’s core to our vision.

Ciprian Borodescu: Absolutely. It’s really important to understand the vision. And just out of curiosity, did you guys filed for a patent for the technology? Or do you intend to? Is that an important thing at TypingDNA?

Raul Popa: It’s important to understand what patents do. Patents protect the people who are afraid of patents. If you understand this very well, then some things are worth patenting, others are not. We are in the process of patenting a couple of things that we’re doing. Other things we think that is better to keep them as a trade secret. So, for example, Google search algorithm, the famous Google search algorithm, was never patented. Why is that? I mean, shouldn’t that be protected? And I think that has to be protected and that’s the way to protect it, by not patenting. And, you know, a lot of technologies are like that. So, if you’re building a security technology – something that would secure people – you don’t want to open it up for inspection to everyone who wants to understand how it’s done. So it’s not only about protecting against competitors, it’s only protecting against attackers, and people who want to reverse engineer or anything like that. But if that technology is going to be working on people’s devices and is subject to potential reverse engineering, then you should patent it. Why not? I mean, if that’s already there, and people can look into it, you should patent. And then there’s design patterns, where you have a flow, your idea is to use your technology in a specific way, you can patent that flow, so other companies cannot patent the flow themselves and restrict your ability to do that. Because we are afraid of patents. So we want to make sure that our patent is clear.

Raul Popa: To make sure that your patent is clear, you can do it in a different way, as well. You can also have – and probably you know this but not sure if your audience does – one thing that we learned is that actually, when you open-source something, when you do that, you have something that is called a prior art. So, once you put something out there, nobody else can patent that because it’s already out there because already somebody else did it. So if you only want to make sure that nobody else restricts your way to do that, the best thing to do is just open source that. Even then, if you didn’t open source, but you still created the technology that you used somewhere in production, you still have something called first commercial use or prior commercial use – so, you used it commercially before. Even if it wasn’t open, you can demonstrate that, so nobody can force you with a patent to not do that anymore. So you can do that even if somebody patents that in the future. So, there are a few protections in place, so you don’t really have to patent if you don’t want to. That’s one of the things that we learned. But also, we’ve seen that other companies, for example, at the next stage, when they have 10, or two or 20 different patents on things that they’ve built, the company values a little bit more. And this is very true for deep tech companies that are built for being acquired at lower one, two-digit valuation. We have bigger expectations at this point, so we’re not really looking at the idea of just selling out for a small valuation. So, you know, the value that patents will bring to the whole company, I think is marginal in this case.

Ciprian Borodescu: It makes sense. You can do so much more with the technology that you guys are developing. What was basically the door opener when selling this to companies? And then, the follow-up question is, did you go with a vertical or did you spray and pray and try to identify the vertical, to discover the vertical that works best?

Raul Popa: There are longer answers and shorter answers. I will go with medium answers – or short. I think, first, there are lots of use cases that we found, some of which, you know, clients came to us and told us about; for example, we have, BPO companies that have the support from India and Pakistan – you know, third-world countries sometimes – to companies like Microsoft or Salesforce, and so forth. So, they use the intercom and talk to the clients to help them, right? And what happens is that these guys have something called like a thin client that runs a very shaped version of Windows. And, at the end, there is one person that is certified and they know the technology – let’s say Salesforce technology – and then they also speak good English and so, they can do the support; they have been trained, and everything is well. But they have five more brothers who also want to do support because they pay by the hour – and they’re not certified, they don’t speak good English, they don’t know the product. But in most cases, there are very, very simple things that you have to do. And that drops the quality of the support.

Ciprian Borodescu: Interesting use case.

Raul Popa: That’s where the attacker actually collaborates with the actual owner of the account. And what we want to do here is to look… Like, these people chat with six people at a time, right? They have six windows open on their computer, and they’re typing responses to different questions all the time. That’s all they do, they type. And what these guys want to do is they want to use TypingDNA to look at what these people are typing and make sure is the same person all the time. If it looks suspicious, then that particular chat goes to somebody who supervises and looks at the actual content being typed. If the text being typed is legit, it’s the correct way to address that issue to the customer, then that user you can leave. That user is unflagged, even if it’s probably his brother or somebody else who’s not… This is not really the problem to actually catch the person who’s doing wrong, but actually to catch the person who’s doing wrong and influences the quality of the outcome – in this case, the quality of the support. This is one very marginal case that actually the clients came and told us about this case. And there are millions of people doing this kind of support jobs from remote.

Raul Popa: And there’s so many others, like verifying employees when they log in from home, when they log in from other mobile devices. Imagine that once you unlock your mobile phone, everywhere you go – you are in a park with the phone unlocked or at the store – and you’re browsing your Facebook page or whatever, you’re scrolling through the Instagram – once it is unlocked, if somebody steals it from you, they have a window into everything you have. So initially, we thought that, hey, if somebody steals your phone, they will just sell it on the black market, or they will make some calls and they will use your phone credit. But you can deal with a very targeted attack in which you’re probably a CEO of a company or you have a position working in a bank and approving credits or something like lending or stuff like that. And then, somebody steals your phone, does something bad as you – they impersonate you and talk to friends or people at work or ask a friend saying, “Hey, I’ve been in an accident. Please send me this amount of money.” All this is done through typing. That’s the thing. And once somebody steals your phone, your phone is subject to anything like that. And we believe that people should install our technology – enterprises should install this on their employees’ devices, mobile phones, computers; now with the work from home and work from anywhere, people take their devices anywhere. And you know, even banks. Banks were not used to using Slack. They were used to meetings. Like, in a real room, meet with everyone, talk, sign documents, those kinds of things. In real life, physically. Now, everything’s done over Slack and emails and that being said, we still have a very small number of verticals from where we see a lot of clients coming to us in order to make sure we address their needs very well. We have specialized people in sales and customer success. And the products are becoming more specialized for these use cases. But it’s the other way around. So, we believe TypingDNA should be a platform for everyone, not just for one particular vertical. In most cases, it’s not like that. Most companies are actually built to only support a particular vertical.

Ciprian Borodescu: Got it. Now, for the final special section on the podcast, lightning questions and answers – a series of fun, short questions that you have to answer really fast. Are you ready?

Raul Popa: Yeah, sure.

Ciprian Borodescu: Okay, let’s go. So, Game of Thrones or Lord of the Rings?

Raul Popa: Lord of Rings, I think. I mean, Game of Thrones, without the last season. They have to rewrite that a little bit.

Ciprian Borodescu: Joe Exotic or Saul Goodman.

Raul Popa: I’m not familiar with the first one. I believe it’s one show on Netflix or something like that, right?

Ciprian Borodescu: Yeah. Tiger King. Yeah.

Raul Popa: Yeah. So I don’t know. I’m not very familiar with that one.

Ciprian Borodescu: Okay. What’s your favorite movie, man?

Raul Popa: Wow. I think there are different movies for different categories. I mean, definitely a movie like Forrest Gump or Schindler’s List. Those are movies I like. Also, The Pursuit of Happiness, I think is very, very cool.

Ciprian Borodescu: Yeah. Alright.

Raul Popa: I don’t have one particular favorite.

Ciprian Borodescu: Cats or dogs?

Raul Popa: My wife and my daughter like dogs, so I’m kind of into that as well. But if I would be single and super honest, I kind of like cats a little bit.

Ciprian Borodescu: Awesome.

Raul Popa: I would have one of each.

Ciprian Borodescu: That’s a good answer – a balanced answer, I would say. The last book you read?

Raul Popa: It’s a book called Range by David Epstein – it’s about the fact that generalists are doing much better than people who specialize in only one thing. It’s really good. It went out in 2018, I think it’s quite fresh. So, it’s worth reading.

Ciprian Borodescu: Romania or the US?

Raul Popa: This is a tricky question. Maybe none of those. But I like the US for now.

Ciprian Borodescu: Okay. And the last question, the bonus question – I know you’re passionate about entrepreneurship and leadership and I need to ask you this. Who’s your favorite leader? Who do you look up to, in moments of doubt?

Raul Popa: That’s, again, tricky. All the questions are tricky. I think I look up to God. I mean, I don’t know. But if you’re talking about real people, there’s this leader who’s very famous in Silicon Valley – died about four years ago – called Bill Campbell. I don’t know if you know him. He was a very important coach for all main leaders in Silicon Valley. You know, people like Steve Jobs, or the Google founders, or Jeff Bezos, Bill Gates. He coached all these guys, and he’s really a leader to look up to, if you want one. I’m actually reading The Trillion Dollar Book, written about him by Eric Schmidt, of Google. A really cool person. Unfortunately, he’s dead, but I would definitely like to be coached by somebody like that.

Ciprian Borodescu: Alright. Awesome. Thank you so much, Raul. It was a pleasure to have you and thank you so much for sharing your wisdom with us and with me. Thank you so much, man!

Raul Popa: Yeah, thank you very much.