 Hello, everyone. My name is Carlos. I am the founder and CEO of Product School and today I'm here with Vivek Raghunathan, co-founder of Neva and former vice president of Google. I'm so glad to have you here. Welcome. I'm super excited to be here, Carlos, and looking forward to the conversation. Yes, me too. I mean, it's not every day that we have someone who's been at Google for 12 years as a VP in your latest tenure there before you started your current company. So we're going to cover everything in a second, but take me back to those days where you were trying to break into tech and what was your story? Right. So my first interaction with the computer was actually back in 91. My mom was a teacher at the time and she got us a 80286 if you guys remember what that was. You know, a few weeks with that and I was hooked for the rest of my life. I knew I was going to be doing this all my life, right? So spent about 10 years in academia, four years at an undergraduate college in India, six years at the University of Illinois, and then joined Google. I put everything in a car, went to Google in 2007. It was much smaller than it is today. I think it was like maybe 10,000 people. Today it's like maybe a couple hundred thousand people. And so started off just as an individual contributor just like many of you likely are and ended up as the VP of Engineering who had brought oversight over all of ads on YouTube. My first four years, I think mostly what I learned was how you use data at scale to derive hypotheses on what can work, what cannot work. Once you have that hypothesis actually tested, drive customer value. And if you do that at a place like Google with the scale it did, you make a ton of money for Google, you make users really happy. My next four years was the funnest. I spent four years ideating a product called Google Now, which many of you like to use on your phones. It's part of what Google calls the assistant today. And we ideated this from the ground up. I worked with a really fun set of people, a partner, who's at Robin Hood Now, a large school taking who runs a big part of Google's assistant effort. Fun because there was no product. We were starting to crash and building something that he has like a few hundred million reactives. And just that act was very different. My last four years were more of an enterprise business. I spent time on YouTube ads. It's like a 15 to 20 billion dollar business. I drove all of the monetization for that. It goes with growth machine, but the thing people don't realize is how that business has pivoted from being a purely brand and TV advertiser focused business to being a lot more performance marketers business, things that focus on app installs and stuff like that. And that pivot happened when I was there. So really fun. I think breaking in big advice to anyone joining would be work on the hardest problems you can find. It takes as much time to work on them as it takes to work on the simpler ones and the impact if you will be successful is a lot, lot higher. You're very humble and make it sound like it's no big deal, but obviously it was amazing your career before you decided to start your own company. I want to talk for a second about your latest role at Google because getting to be in a VP is not just based on the amount of years you've been working for the company. So what was kind of that take me to that moment where you go from being an individual being a great individual contributor to then taking the leap and becoming a great manager? Right. So I think I'm going to say started around end 2014 early 2015. I got thrown in the deep end of the pool. We bought YouTube. Google bought YouTube back in 2006, I think. And Google had been running YouTube's ad business mostly on top of the double click stack, which many of you know is the display ad stack that powers most of Google's businesses. And I got thrown in the deep end of the pool as told to build a first class built for YouTube ad system from the ground up. That could truly use the power of what users did at YouTube to make the ads look more relevant. Of course, I like to remember my reaction when I was thrown in the deep end. I said, hey, I'd like to do this with a 6% team and I'd like to be the one actually in the weeds of the project on a day-to-day basis. And my boss at the time is my co-founder now. He ran all of ads at Google. He also ran all of infrastructure at Google. And I think he just smiled to himself. I didn't think he said it. He said, I support you in whatever way you want to be successful. Three months later, it was a 50% team. Two years later, it was a 200% team. Three years later, it was a 400% team. And really it was, we were doing very disruptive things on top of what was already a very big business. If there's one kind of learning from there is businesses that get big or products that get big, the biggest risk you have is yourself. You suffer from the curse of incriminate tourism, right? You don't do big enough things. And we were comfortable doing something fairly disruptive to what was already a big business. I think that the effects of that are being seen in how YouTube is doing in the market today. It's 40%, 50% growth. And I see a pattern here. You talked about building from zero to one within Google, which obviously comes with very high expectations of what it takes for that new product to then be deployed to market and be successful compared to maybe older startups that are constantly experimenting, launching a feature. And as long as there is some sort of traction, they just go live. So what are some of those parameters that you need to follow at Google in order to define that experiment success and send it to market? I think going back to my time at Google, it's not actually very different from many of you who would be in a startup. I think the big difference is you have distribution. So if you get some kind of fit, you don't have to wait for distribution. You can throw distribution as the problem. That is a little bit of a double-edged sword. You can also throw distribution at products that don't have fit and still be somewhat successful at becoming like Google. If anything, I think in the roles I was at Google and the roles were very different, right? The ad business on YouTube is much more of an enterprise-focused business. The Google Now business is a lot more of a consumer-focused business. For enterprise-focused business, I would say ultimately, your sales teams will sell what they can sell. So if you're building products that they are not able to take to market, if you go to market function is not able to commercialize your efforts, you're kind of dead in the water. So the most important thing that I spent lots of my time when I was at YouTube was just deeply understanding what the customer needs were, deeply working with my cross-functional partners in the go-to-market teams and understanding what we could bring to market that would actually help them grow the business. And trading off this what I'll call fierce urgency of now, you always want to sell what you had yesterday because you can sell it versus like inventory inventing the future and having the right mix between those two things. Yeah, and I can imagine that patience is important because some of the things that you are trying to do are for the long term. However, we've seen a lot of, Google has been famous for shutting down products that were kind of successful. At least they had some millions of users, but maybe it wasn't a good fit for the long-term strategy. Yeah, I think, and to me, this is why I said the ad side of the business and the consumer side of the business are very different in my mind. I think the ad side of the business that editing or that pruning down or that natural fine-tuning of what happens is happening because your successful products are getting more customers, they're getting more revenue, they're getting more resources thrown at them and that loop or that flywheel can even, the consumer side to your point is much harder, because you can get to 100 million. Yeah, I can imagine that shutting down something that is working is harder, I mean, that's less obvious than something that is not working. Right, and you see this in the number of messaging products that Google has had over the years where you keep it reading on something and like may not always work, right, it's never guaranteed to work. I think Google now in many ways, I think was lucky to be bundled as part of Android in the early days, so it was able to get lots of users and that flywheel was able to kick in. I know also Google has been famous for just basically introducing product management when it wasn't really that cool. You guys created one of the first, if not the first, APM program and obviously a lot of famous product managers there. So when you were there as a VP, what was your relationship? What was that secret sauce between working between engineers, designers and product managers? Yeah, I'll say a few things. I think some of these will also talk to things like, myths and misconceptions people have sometimes, right, but I'll also give you like my perspective on how I went about doing that. I think single most important thing, we thought of ourselves as a joint team. We did not think of ourselves as functional organizations that had goals of their own that were working with each other in some kind of, you know, there's no such thing as the product that writes the PRD and the engineer writes the design doc and the engineer design doc implements the, any time I had heard the conversation in my teams about, oh, you know, the strategy is not the right one. I just shut that conversation down on day one. I told them we're going to think of ourselves as one combined team or, you know, get everyone in the room. And the same thing helped for my product peers and my design peers. They very much thought of execution of the strategy as part of their mandate and as something they did together. And if they hit any roadblocks, we worked on it together. That is like, you know, biggest misconception. One is this notion of like very tight boundaries between the various functions. I think things are a lot more fluid. If anything, I think the APM program makes it even more fluid, right? The second thing, I think that I saw very much was sometimes, you know, as engineers and PMs and designers, we have this should be very data driven, which is kind of the, or should we talk to users a lot and the market a lot, or should we go with our gut, right? And each function brings some aspect of that to the table. And in my mind, great product building is adaptive, right? If you have very little data, no clear picture of who your user is, you don't have a clear picture of their need, you need to start with your gut and build a shitty MVP and talk to users and get their feedback, right? If you have a clear visualization of the product need, if you walk into a meeting with your potential user and they tell you, here's a demo of this thing you're going to build for me that I already built, like you have your user, right? You just go and execute, that's all you have to do, right? And then if you're Google and you can acquire users in drones, then you can rely on metrics and you can do a test the crap out of everything, right? You can, you feel the right shade of like lilac that you should have like the link as just to share the experimentation side, almost they being adaptive to what strengths you have in the teams you have is kind of super important. And not all teams at Google were built like that. There were big teams with lots of design support and there are tiny teams with one product manager and 50 engineers and like, yeah, your style has to change depending on which team you're in. So now, at the beginning of 2019, you and your co-founder who's also an executive at Google decide to start a company from scratch. Why? It's a, I'll give you my story and you know, if you have Shridhar on here, you can ask him his story. My story is fairly simple, right? Like I told you, I did three things at Google. I worked on search, I worked on Google Now and then at Nats for YouTube. And when I built Google Now, I, and when I worked on ads in YouTube, I had this like knowing feeling, like a little nagging feeling in the back of my head that I was missing something, right? We would have conversations about should we trade off privacy or should we trade off or should we be more personal, right? Like if I could know that your package was on the way, should I tell you on your phone that your package is on the way and you should go pick it up at your doorstep or should I knock that involves me having to parse your email? Should I do that, right? Where is the line between privacy and personalization? And when I was at YouTube, we spent lots of conversations about, you know, how much, you know, advertisers love reach and frequency. And we would have these conversations around, you know, how many times is seeing the same ad like, okay, like users don't like it. That's what your customer wants. What is the balance between what your users want and what your customers want? And so, you know, with this in mind picture me early 2019 Shridhar had left Google. He was at Greylock, which is the venture firm. And he's a venture partner there. And he's figuring out what to do next. And he's pulling on this thread. And the thread he was pulling on was you can't reimagine search the product without reimagining the business model that underpin search, right? And the key insight he had was let's build a new search engine. We'll reimagine search, but it will start with the premise that the customer is the user. There's no other customer of the product. It's a subscription product. It'll be ad free on day one. It'll be private on day one. But from the freedom we get from that underlying business model, we can rethink the product itself, right? I thought it was the coolest insight I've heard in my life. When I heard the insight, I was like, yeah, you know, I need to do this, right? And so it was, you know, a good friend of mine once told me you cannot, you should leave to start something if you cannot not start it. You know, that like height we took up the Stanford ish was that was that insight for me. So it was pretty clear, like, you know, told my wife, told my kids, told my boss, like arranged for things that I could leave for, you know, in a month and a half. And we go far into the company in early February. Hey, good as to you, my, my admiration goes to you and your co-founder. I think it's fascinating that, you know, I can totally see the PR headlight, like two Google executives leave the company to reinvent online search. I think it's incredible. But at the same time, it's very risky for many reasons. You mentioned your wife and your kids. So I can imagine, obviously, there's some unfair advantages based on your experience working at Google and also in general, at building products. But still, building now from literally from scratch, not having those hundreds of engineers at your disposal. How did you have to shift your own, your own work style, like to really make sure that you are being efficient with this new environment? I think it's a great question, Carlos. I get asked this question a lot. And I'll kind of break the answer up into two or three pieces. So you'll get a feel for the various pieces, right? What is one of the inputs, right? What would you do differently now versus, you know, 20 years ago, or if you're in a big company, right? In a big company, you can take the support system in terms of the infrastructure for granted. You can't really do that outside. But what's changed is, you know, you can build on the shoulders of giants, right? Cloud services are incredibly cheap, like data processing frameworks are everywhere, Spark kind of works. You can rely on amazing data analytics platforms, whether it's Snowflake or Redshift or Mold or Amplitude or any of these things. They can get your stack running and piece it together by renting out the pieces that you care for much faster than you would if you were in a bigger company, right? So I'd say that's like one big thing. Second big input is need to be very careful to understand. And, you know, we all do this when we build product. Where are you differentiating and where are you, you know, doing table stakes work, right? One way of saying this is, you know, you're trying to acquire loyal subscribers. Where are you going to get loyalty and delight and reward from your customers? And where are you investing in something? Because if you didn't do it, they would leave like it's a churn preventer versus like, you know, it's what sales would call an objection handler, right? And being very crisp and clear about you can't differentiate everywhere, that like products that try to differentiate everywhere, differentiate nowhere. So what are those one or two things that could be very different? And then everywhere else, you're just trying to make sure the experience is good enough so you can get off the ground and like rely on the bet that you're one or two things are. So actually that's the second big thing, like is figuring out what those differentiated bets are for Niva, that's a very deeply personal experience. You can put your personal content in and we'll search over it, but the experience gets a lot better. You can choose which providers you like, which providers you hate. The second big thing Niva does is we are very publisher centric. We try to blend, blur the lines between searching and browsing. We bring in answers from great content on the web into the search experience, right? So I'd say that second big thing is figuring out what those differentiators are that are both great value to users, solve a real user problem, but don't kind of are not easy to replicate by a bigger incumbent, if you will. Third big thing, how do you hide out and build out for a problem as big as this? I like to structure my team as pods. They're roughly autonomous five group units that can run with missions by themselves and they're loosely coupled. They know what their goals are. They're aligned with everybody else's goals, but they're roughly autonomous to do whatever they want. And that lets us really run all these goals in a fairly focused fashion without needing everybody to talk to each other to make sure everything is just perfect. I'm excited to check out Niva. So what is the status of the product today? Oh, perfect. We came out of stealth mid-June of 2019. We are in alpha right now. You can go to the website on www.niva.com and sign up and try the product. We're going to be coming out of our alpha pretty soon and doing a much bigger push in the next six to eight weeks. So stay posted, but if you can't wait for that, like www.niva.com, sign up. You should start becoming a Niva customer using us as the end of this call. So that is an interesting rollout process, like this in the startup land where we are used to company shipping, let's say within months. And you mentioned that you've been in alpha for two years now coming out of beta. So what was that logic behind taking longer just to build before you test with market? Right. So I think a couple things, right? First, search as you know, it is a much broader surface problem. The thing I find most fascinating about search is just the diversity of human needs, like 10 to 20% of all searches that every search agency, even the biggest ones, are completely unique. They've never been seen before. It speaks to the diversity of human needs, but it also speaks to the bar you have for making a search experience great, right? People trust their lives with a search product. Like people type in health conditions that a loved one might have to see what the prognosis is. People type in summer camps for their kids to see what their kids could do. People will prep for an interview or for a job, like using their search engines. The bar for a search engine being great, super high. At the same time, we've been iterating with our users for much longer than this. The second kind of thinking we've had a lot more, it is really important for us and our users to be able to use Niva for all their search needs. Not kind of one time a day, not kind of like, but you know, as their exclusive search experience for long periods of time. And so lots of our alpha has been just iterating with the number of users we have. It's in the high thousands and work with them on making sure that we're meeting all their needs to the place where we feel fairly comfortable. And now we can, so I would say some of the labels there, alpha and beta, are almost monikers, right? Our alpha is as close to a fully flashed out product as your one. I mean, online search is something we'll do every day, multiple times a day. And I agree with you, there's a big opportunity to reinvent and be more efficient around it. What would you say are some of those big level opportunities for the future of online search? Right, it's a great question. I think one thing we all learn as people who care about product and users is you need to be in love with the problem and the mission statement, not necessarily your tactics, right? I've been super passionate about search and discovery. I think it's a utility that we don't fully grow the value of it in our lives, right? Let's take a utility like internet connectivity. I was in undergrad in 1998, we had a shared 64 kilobits per second line. It's let that sink in 64 kilobits per second. I have a gigabit per second line right now while I'm talking to Carlos, right? Things got 1600x better in a 20-year time frame. I think search and discovery is like that, we just search with our most intimate needs, right? To me, the vision for search is it's something that has three aspects to it, is deeply personal, right? I'd like it to be my search experience, something I'm in control of. I should be able to tell it what sources I trust. I should be able to tell it I care for these forums more. I care for this official content more. I like indie sites. I don't like big box sites. I don't like aggregators. If you're a programmer, you should tell it I care for Python 3 and not Python 2 or I care for TensorFlow and not PyTorch. If you're a designer, you should be able to tell it I care for Figma, not for Sketch. All those things should just get better, right? So that's one. I'd like it to have memory and context. These are overloaded words, but no one wakes up and says, I have 40 searches I want to do in my life. You don't do that, right? You have tasks you want to do. Maybe you have to do this interview or you have a planet trip or you use various tools to solve these tasks, right? So just help me solve these tasks and use all my context. Let me give you some examples. At the easy end, I may say typical size of an employee pool. Do I mean an employee stock option pool or do I mean a swimming pool that I'm going to have in my office for my employees, right? If you knew about me, it likely means the former. It likely means the email from my lawyers about what should the size of the stock option pool be, but why is it that search engines can't do that? That's at the easy end, right? This is not very hard. At the middle end, search for me so I don't have to search, right? Make sure I don't forget things. Make sure I don't... I'm super sloppy. Make sure I don't forget the email I forgot to reply to. Make sure I... When I come into a meeting, I'm as prepared as my sales friends would be to have these briefing books on all the people in the meeting and here's what you need to know and here's what they care for. That should be just something I don't kind of access to, right? Take advantage of all the context you have to... That's kind of still like somewhat easy. There's something impossibly hard, right? I'm trying to sign up my older kid for a summer camp for the summer. She's nine years old. She likes sports. She likes arts and crafts. Me and my wife are tiger parents, so we want to make sure her summer also has some like intellectual activities, right? We want to travel in the last two weeks. I just gave you a tax description. I told you what we want, right? The search engine should be able to use all this context and iteratively problem solve this with me. Instead, here's what we're doing. We have a Google Doc to write down all of the information about these camps. We have a spreadsheet to check out her schedule. We have lots of searches. We have lots of browsing. We have emails to friends to figure out what they're doing. Just solve this for me. Make this one amazing experience. This is impossibly hard. I don't think anybody can do this right now, but that's what excites me, right? And finally, I want an experience we can trust. More so than ever before, I don't trust the content I read online. I want an experience where you get the best content from sources you trust, where the friction to consuming information from a source you trust drops to zero, where great content from publishers is amplified. Neva will actually share at least 20% of all our revenue with publishers who we partner with. We just announced a partnership with Kora and Medium. We have more publishers coming up. I don't want to worry about whether this review I'm reading is false or this site was recommending me a product is getting paid to push it. I don't want any other stuff. I just want content I can trust from sources I love. Those three things put me in control of experience, memory and context and experience I can trust. I think if we solve these three, if someone solves these three, they'll make search 1600x better than it is. That's the opportunity in my mind. My last question to you is there are a lot of product folks on the audience who are also implementing search as part of their products. It is not maybe a public search but it's a search as part of their own e-commerce site or any other type of product. How do you think that a search engine like Google, Neva and others can be integrated to do some of those micro searches? Right. The first thing I tell you is pick me, talk to me because it's a problem I care for. I want to understand your needs. The second thing I'll say is try and build your product as a SaaS app. Try and use things like OAuth and SAML and have an API on the get go. I think where we're going to end up in the future is we're going to have the ability to cover together what I'll call search tech with your own repo and be able to use things like semantic search and lots of advances in NLP to power great search experiences. To do that, build your system from the ground up so you can integrate into all of this tech pretty easily. To do that, I'd say most important things, don't try to cover together something completely by yourself. Try to use open source if you can get going with it and you can switch over to something better over time. Try and build it in a way where you have OAuth and SAML and APIs from the get go. I think to do those two things, you should be fine. That's awesome. I think we're moving into a world where now as a product manager, it's not about just reinventing the wheel and trying to build every single component. It's more about understanding how to fit third-party components into your core product and just add value in a differentiated way instead of just trying to build yet another search engine. I could not agree more, Carlos. I think knowing which parts to build and which parts to buy are probably the most important choice you have to make right now. It's been a pleasure to chat and learn from you. Is there anything else you would like to add? No. It's been an amazing experience being over at your show. Thank you so much for having me over. If you want to try the product, just go to www.neva.com and give us feedback on what's working and more importantly, what's not because as product people, the most important thing you want to do is make our products better. Thank you so much for your time. Bye, Vivek. Bye, Carlos.