ShishirMehrotra

LubaThank you so much for coming on the show.Link

ShishirThanks for having me. Fun way to do a podcast.Link

LubaYeah, exactly. We're getting some steps in.Link

ShishirYeah.Link

LubaThere is so much to cover and talk about. I want to get into YouTube, writing, rituals, obviously Grammarly, Superhuman, but I want to get started with a fun fact I heard about you.Link

ShishirOkay.Link

LubaYou and your daughters are obsessed with Lego.Link

ShishirYeah.Link

LubaCan you tell me a bit more about that and the Lego competitions you guys go to?Link

ShishirYeah. We did Lego robotics for seven years. Started when the girls were in middle school. They're now — one's in college, one's a senior in high school. So they moved on to the next level of competitions. But yeah, Lego robotics runs these amazing competitions. The rules are basically there's 15 tasks, you can use anything you can build with Legos, and you have 2.5 minutes to solve the tasks. And we made World Championships four out of seven years.Link

LubaDamn. Okay, wait. How big is the team?Link

ShishirIt varied. Every year was different, but between 2 and 4 girls. Two and five girls. So very small.Link

LubaYeah. Wow. Were you the coach?Link

ShishirYeah, I was the coach. Yeah. As you get older, the school does it, but at that age, a parent has to do it. So yeah, it's super fun. It's a highly recommended way to, you know, to have fun with your kids, to learn a lot about robotics, about building. They learn a lot about teamwork. It's awesome. It's a very good program.Link

LubaSo how often did you have to coach them to be as exceptional as, you know, being in the World Competition four out of seven times?Link

ShishirOh, I mean, I think they would say more than they would have liked. We were a pretty Lego-obsessed house. The garage is full of Legos still. Yeah, we built everything you can imagine. You know, I think officially a couple days a week, but we pretty much worked on it every day.Link

LubaWere you the one who was like, "Okay, girls, I'm going to get you into Lego," or was it...Link

ShishirNo. My older daughter's friend did it, and she got her in, and then that kid aged out of the program, and the dad handed her off to me.Link

LubaVery cool.Link

ShishirYeah. Super fun.Link

LubaAny parallels of what you've learned...Link

ShishirOh my gosh, so many.Link

Luba...to building companies?Link

ShishirI mean, I remember one conversation I had with them was, every year we would decide how many kids we were going to have on the team. And the first couple of years, they wanted every friend of theirs on the team because it seemed like it'd be fun. And then they started realizing that adding more people actually made the team a lot less functional. And every year we would trim it back. I think one year we had 6 or 7 kids and it was just chaos, especially kids — I mean, we started at nine years old. So it's very hard to keep them focused.Link

And then over the years, we got better and better at, like, you know, why smaller teams are actually more effective. And I think that is ironically very true in the business landscape, but most people don't realize it. Actually, smaller teams are better than bigger teams.Link

LubaSo how small are we talking?Link

ShishirI mean, I think great teams, you want them to have what we sometimes think of as hive mind, which is like they're always thinking the same thing all the time. And to do that, you need them to be completely on the same page. Teams of three, four or five.Link

One of our favorite rituals here is we do these quarterly hackathons. And my rules are generally no teams less than two people. So you can't do it by yourself. You've got to work with somebody. But if you get above five...Link

LubaYeah, you can't do it.Link

ShishirYeah, that's not the point. But if you get above five, you'll start to see you spend way more time coordinating than you do working. So I think there's a magical number in there. And it doesn't mean that — obviously we have 1,500 people at Superhuman — it doesn't mean you can't have big overall teams, but you want to break things into much smaller areas so that people have a chance to have four or five people that they really work with.Link

LubaVery cool. We're going to have a chance to talk a bit more about rituals and things that you guys do at Superhuman, but I want you to talk about your upbringing. So something that I've read is that for a period of time, you as a kid didn't really know what you wanted to be, and you would, like, jump around every couple of weeks. And at some point you wanted to be a taxi driver, and your mom was not particularly thrilled with that.Link

ShishirNot thrilled by that, no.Link

LubaTell me more about that period and, like, how did you grow up?Link

ShishirWell, the taxi driver thing was later. I grew up — both my parents were computer scientists, and so early on it was anything but that. And it wasn't really well informed. It was just like any kid.Link

LubaYou grew up in the US?Link

ShishirI grew up in the US. Yeah, I was born in Buffalo, New York, and then my dad was doing his graduate school, and so we moved around a bunch as he finished grad school and then became a professor. So we were at UVA in Charlottesville, Virginia. Then we were in West Lafayette, Indiana — he got his first professorship at Purdue — and then we moved to Virginia, where he ended up at NASA. He was at NASA for a long time, 30, 35 years. He just retired. But yeah, so a very academic family. All my aunts, uncles, grandparents were academic as well.Link

But yeah, I was drawn to everything. You know, at one point I was convinced I was going to be a doctor, a lawyer, a, you know, whatever. A banker, everything you could imagine. I think I'm a very naturally interested person. And, you know, I would say my current job, while in tech, I end up being exposed to many different things. And so I spend a lot of time with education. I spend a lot of time with journalists. I spend a lot of time with media. So I think...Link

LubaPretty much fulfilling that.Link

ShishirI like being in places where you can see lots of different things.Link

But anyways, I was in college and I called my parents. I said, you know, I was thinking about taking a side job being a taxi driver. And my mom got really upset at me. This was before Uber and so on existed, because being a taxi driver was like, you went and got a certification. You had to learn the map of Boston to some huge level of detail and pay a bunch of money for a car and so on. But I just thought it was so fun to be in cars and be this observer on what's happening in the world around you. And I love Boston. And so it was this way to get to know the nooks and crannies of Boston differently. But that was what my mom wouldn't let me do.Link

LubaI love that, because I still have a bucket list of jobs that I think would be really fun to do, actually. Trying out Uber driving is on the list as well.Link

ShishirYeah. Now it's totally reasonable. You can do that for a few days. But at the time, that was like a one-year commitment, so...Link

LubaOh, I see, I see. I actually, during COVID, I signed up to be a DoorDash and an Instacart driver because I was just so curious what the product on that side of the store looked like and what the driver experience was like. And actually, we made YouTube videos about it, like, oh, this is the comparison. And then I did product manager tips for the product managers of Instacart and DoorDash about what I would change. But it was really fun.Link

ShishirI mean, it's one of their main rituals, right? They make people go through it at some frequency, which I think is, like, for every company, you want to make people use your products that way.Link

LubaOkay. So you basically kind of wanted to be a lot of different things. How did you end up deciding to do math and the minor in computer science?Link

ShishirOh, I double majored, I would say. Well, the CS and math thing was its own interesting story. My freshman year in college, I decided I would do CS. I liked a couple classes, and I remember calling my dad and I said, "Hey, I want to do CS," and I just assumed he'd be very excited for me. And he was actually really upset about it. And he said, "If you do CS as your major, then I won't pay for college." And he said, "I'm only paying for college if you major in something where the books are at least 50 years old." CS changes too fast, and it's too vocational. And so it's not worth the education. He was a physics major. And so that was his, you know — and I think he was right.Link

So in his head, I was a math major first, CS major second. I think officially I was a CS major first, math second, but it doesn't really matter.Link

LubaSo is there something you were telling your daughters, that they can't major in something where the books are...Link

ShishirI mean, this was definitely... I mean, my daughter's a math major.Link

LubaOkay.Link

ShishirSo, I mean, whether because of that advice or not, she definitely is following that path. She's at Princeton. They don't allow double majors. So she is math first, CS second. We'll see what the second one does.Link

But I think the philosophy of — there are some things you learn that are very fundamental. It's not really the math that matters. It's the learning how to think that matters. And I do think, you know, some of these deeper subjects just force you to keep peeling an onion back of, like, how do they discover the next level of this and the next level of this? You learn how concepts fit together. So I do think that's pretty helpful. I guided my kids that way.Link

LubaFor someone listening who might be sending their kids off to college in five years, would you recommend them to still go with more traditional subjects like engineering or math or physics?Link

ShishirI mean, I think there's no easy answer to that question. But obviously, first off, whatever your kid is into — I mean, it matters the most that you can't learn something unless you have deep interest in it. And there's no pushing your kid one way or the other. I don't think it's very effective. That said, sometimes they may not have been exposed to the way to get interested in it. Like, the way my kids got interested in math was math competitions, which...Link

LubaLike Olympiads, or...Link

ShishirYeah. And in particular, they're very drawn to team competitions. Math is not generally seen as a team thing to do, but if you do it that way, it becomes a team thing. I suspect my younger one's going to end up doing engineering, but she got even more into robotics. And so I think giving your kids a chance to experience lots of different things is very important.Link

You know, if there's a choice, I think learning things that are more fundamental tend to scale. I think one of the things kids don't realize, but adults realize, is kids go into the world thinking that the jobs in the world are directly mapped to every subject you could major in in college. And it's like, oh, if you major in math, then you become a mathematician. You major in chemistry, you become a chemist. But actually the jobs in the world are significantly more diverse than the number of majors. And so I think they sometimes judge majors based off of that assumption.Link

LubaYeah.Link

ShishirWhich is unfair. For Annika, for my older one, I think some portion of the math majors definitely end up being mathematicians, but many of them end up in tech, they end up in finance. So I think kids don't always understand that what you learn and what you do isn't the same thing. So you have to expose your kids to as much of that as possible.Link

LubaYeah, I can still relate to that. I'm originally from Belarus. I ended up going to Waterloo for school, and I studied chemical engineering, and I was like, I have no idea what a chemical engineer actually does, but I liked chemistry and math and whatnot. But then I ended up teaching myself how to code and switching to tech. But it is so important to expose yourself and your kids to what the opportunities are, and I don't really know what is an effective way to do that when someone is really young, before they have to make this choice.Link

ShishirI mean, a very small tactic I'd say is don't box them out of the adult conversations. I think it's amazing how many parents will tell their kids, "This is adult talk. You should go play." Assume your kids are interested and give them a chance. And some kids aren't interested. But it's amazing how many kids will go to college and then they'll ask questions about professions. And I'll ask, did any of your parents' friends do that job? And they have no idea. Like, you probably spent so much time with them, and your aunts and uncles do that job, and they have no idea. But I think there's a lot to be learned by the people you're around all the time. And kids don't think it's appropriate to ask, "What's your job? What do you do?"Link

Thankfully, my older one is incredibly inquisitive and very good with adults. And so she was constantly asking, and, you know, she'd come home and ask me how my day went, or the meetings. It's very interesting. I think she has a pretty good picture of it.Link

LubaThat's pretty awesome. So let's jump back to — you go into college, you went to MIT, you double majored in math and computer science. You then start a company out of MIT. How did you get startup-pilled back at the beginning of 2000?Link

ShishirWell, I got forced into it. A friend and I started grad school. And we picked a project, and we were working on it together, and he came and he said, you know, there's this entrepreneurship competition at MIT. Maybe we should apply. It's called the MIT Entrepreneurship Competition. At the time it was called the 50K Challenge because the prize was $50,000. Now it's $100,000. So I guess, inflation.Link

LubaYeah. 26 years of inflation.Link

ShishirAnd he said we should apply. And to be honest, I know this sounds weird because it was 1999 and we should have all been startup-pilled, but we kind of weren't. It wasn't seen as glamorous. It wasn't seen as, like, you must do it. And my parents were mostly just focused on, make sure you go finish your PhD.Link

LubaOh, you were on track to do a PhD in math, computer science?Link

ShishirComputer science.Link

LubaAt that point, your dad was okay with that?Link

ShishirAt that point it's okay. Yeah. The undergrad had a different role.Link

And we just happened to do well in the competition. We came in second in the competition and got a lot of interest. And I think it's one of those startup stories where I don't think I ever sat down and thought about, you know, should we do this or not? It just sort of took off. And investors started reaching out. And at some point you build up a team in order to do the competition, and now you've convinced a bunch of people to do this thing with you. And before you know it, you're starting a company.Link

But it's interesting. There's a lot of people with more intentional paths to startups. But both the companies I started were fairly unintentional paths.Link

LubaYeah. Do you think in retrospect that someone should always choose their path intentionally, or should you follow, you know, curiosity?Link

ShishirI have a fairly strong opinion on this for startups, and not everybody's going to like this, but I don't think people should start companies in a pressure cooker. Sometimes people will come to me and say, "I want to start a company." And I'll say, "What's the idea that you're obsessed with?" And they don't have one. And that doesn't make any sense to me.Link

In my mind, starting a company is often the most painful way to get your idea to the world. To start from scratch is the hardest way to do it. If you have a thing you really want to see different in the world, you really want to work on, in many cases there's easier ways to do it. So you should start a company almost out of — I almost view it as a curse. It's like, I just can't get the idea out of my head, so I must start this company.Link

But, you know, everybody has a different way of doing it. I've seen a lot of people who start with this frame of, "I'm starting a company, I don't know what I want to do," and they end up going in circles. There's a lot of bad incentives involved in the adverse selection of "I have to start a company," and therefore every idea is viewed through that lens instead of, like, what's the most effective way to get this thing done in the world? So I think "the startup thing will strike you as a curse rather than a blessing" is the right way to think about it.Link

LubaWhat happens if the startup idea strikes you, but then after a year or maybe longer at it, you figure out that maybe there's no market, you've got to pivot? Do you hope that you get the same level of inspiration for a different idea?Link

ShishirI mean, it's part of the reason why I say it, because every startup idea hits roadblocks. So if your level of attachment to that idea is weak, the world will knock you off the idea. There is no such thing as an idea that people haven't thought about going after. So you have to be willing to stick through the hard moment of, you know, the customer who doesn't think it's a good idea, or the investor who doesn't think it's a good idea, or the employee who doesn't think it's a good idea, or you hit some roadblock. And if you're really passionate about it, you'll go past that. But if all you're in it for is "I just wanted to be a startup founder," then very quickly you'll pivot ideas way too fast and not stick through them long enough to see them work. And most interesting things take real time. And so you have to be willing to stick through it.Link

LubaYeah. You have to will it into existence. Otherwise, as you said, the world's going to knock you over.Link

ShishirYeah.Link

LubaSo you started this company, you raised from Vinod Khosla. What did you learn from Vinod back then?Link

ShishirOh, man. Vinod. His own character. The whole experience of raising from Vinod is maybe its own fun story. We were starting this company. I was trying to raise capital from West Coast investors. At the time, the world was very divided — Boston investors or East Coast investors and West Coast investors, very different viewpoints on the world. And I had flown out to California a bunch of times — lots of interest, but nobody had really bid. So I'd given up on it. I had signed a term sheet with this investor in Boston. But that person hadn't wired money yet, which is an interesting part of the story.Link

And I get this email from Vinod Khosla. And you have to remember, this is 1999, early 2000. The world is not as connected. So I got this email from Vinod, but I don't know who this is. And the email — if anybody's gotten an email from Vinod — it's all lowercase, many misspellings. It didn't seem official, but he seemed very interested. "I saw the business plan from a friend. I'd be really interested in participating." And so I sent it to this investor who I'd signed this term sheet with. And I said, "Do you think this is real? Who's Vinod?"Link

And he calls me right away and he says, "You don't know who Vinod Khosla is?" And I said, "No, I don't." And he says, "Well, hang up and walk down the street to the bookstore." There were still bookstores. "Look at the rack of magazines." There were still magazines. "The guy whose face is on all the covers, that's Vinod Khosla."Link

LubaWow.Link

Shishir"Go read about him and then call me back." And so I go, and I learn a little bit about Vinod. And as I'm walking back into my living group, my CFO, who was my friend, he stopped me in the hall. He said, "Shishir, it's kind of weird, but $500,000 just showed up in the account."Link

LubaWow.Link

ShishirAnd what I realized was, Vinod even signaling interest completely changed the market around the company.Link

LubaThose people who weren't wiring money...Link

ShishirThey were wiring money now. But yeah, I mean, Vinod was at the very top of his game at the time. Not that he's not now. And I learned lots of things from Vinod. Vinod is definitely a very think-big entrepreneur and investor. You know, the bigger the idea, the more interested he is. He's incredibly fast acting, fast decision making. Maybe to a fault. If you had to give him feedback, sometimes he can spin things around a little bit. But it was a great introduction to — it was definitely the vibe of West Coast capital versus East Coast capital. It could not have been more different.Link

And by the way, one of his requirements was you've got to move the whole company to Silicon Valley. So that's why I moved. He sent this email on a Sunday. I met him on Tuesday.Link

LubaWow.Link

ShishirHe signed on Thursday. We got a moving truck the next Saturday, and we were all moving. And, you know, I had to give up a lease in Boston and so on. But the California switch was very fast.Link

LubaVery fast. And for someone listening who's not familiar, what was the startup?Link

ShishirOh, sure. So Centrata was a data center automation company, or we called it a data center utility company. I think the easiest way to understand the original vision is it's pretty close to what the cloud became and how you think about AWS and so on. But there was a generation of companies then that all tried to do it. And I would say it's a good case of how much timing matters, because we were all, like, 5 to 7 years too early. The core enabling technology hadn't been invented yet, which turned out to be VMware.Link

And so there's a bunch of us. And the other companies in the cohort — one that's quite famous is Marc Andreessen and Ben Horowitz started Loudcloud. They were one of our competitors, and that turned into software. But it was basically all these companies trying to build an on-demand cloud for computing, and they ended up building software for data centers instead.Link

LubaGot it. So what did you learn from Vinod Khosla that still informs your decisions maybe today?Link

ShishirI mean, I think the think-big part is very important. I think that was top of the list. I'd also say he's tenacious. I mean, he's well known for it. But his tenaciousness of going after, whether it's a candidate or a customer or an investor — he just wouldn't take no for an answer. And I got to spend a lot of time watching how he just wouldn't take a negative signal as a negative signal. Like, we were talking about entrepreneurs who gave up too fast. Vinod is the exact opposite. If he believes in something, he will not stop. And I think people will refer to me as quite persistent that way.Link

LubaYeah, definitely. Actually, this makes me want to jump to your days at YouTube. So many years later, you join YouTube when it was literally just purchased by Google, and it sounded like YouTube was losing over $1 billion a year.Link

ShishirAlmost a billion. Not quite. We never quite got to a billion.Link

LubaYeah. And it was kind of seen as Google's big mistake.Link

ShishirYeah.Link

LubaSo you joined. What was it like working on a business that had such bad perception internally?Link

ShishirWell, it was interesting because it was definitely the exception, right? You know, Google had bought a number of companies and almost all of them were turning out to be reasonable successes. And YouTube was the odd one. And like you said, not only were we losing hundreds of millions of dollars a year, we were net negative on every single view. And so it meant that the more views we had, the worse the business did. It's not a great spot. We had this billion-dollar lawsuit from Viacom. We were known for things that people didn't quite understand. You know, grainy videos of cats. And so people just didn't understand, why is this important? Where does it fit in the world? Our number one competitor at the time was a company called MySpace, which I think people nowadays probably don't even remember. But I don't think people really saw us in that frame.Link

But there was a small group of people who saw YouTube differently. And that group excitedly ran to it and said, "I bet we can turn this into a very different kind of..."Link

LubaIn that group...Link

ShishirYou know, I think I went over the same day as my boss, a guy named Salar Kamangar, who was employee number seven or something at Google, and he was a sponsor of the original acquisition and so on. There was a group of product leaders — actually the one here, Noam Lovinsky. We bought his company and he joined. He's my chief product officer here at Superhuman.Link

LubaWhat company was he running that Google bought?Link

ShishirA company called Episodic. There was a number of them. You know, many of them are spread around the industry, although a number of them work with me here at Superhuman as well.Link

But, you know, I often describe this — you can see team dynamics that are either more pirate or more navy. This is definitely pirate. So this is definitely in that same entrepreneurial spirit. You know, the world thinks it isn't going to work. And for some people that deflates you, and for some people that really powers you. And so that was our frame.Link

And we just had a viewpoint. It wasn't really that complicated a viewpoint. I remember the first time I said it publicly — I joined, and I think a month later I had to give a talk at this big ads conference. And so I gave this line. I said, "YouTube is going to do to cable what cable did to broadcast." We went from three channels to 300 channels, and now I think we're headed towards millions of channels. And honestly, that room of people almost laughed out loud. It seemed so crazy to them that this site that was mostly, you know, what felt like joke and cat videos, and didn't feel serious, was going to compete with cable television. They had all stuck us in, you know, "Oh, it's a video social network." That's how they perceived us.Link

But that was our viewpoint, that YouTube was going to do to cable what cable did to broadcast. It was a very simple idea. It drove everything about how we thought about it. If you thought about it that way, then you were very motivated to get past every roadblock and get to that next step. And, you know, it ended up working.Link

LubaSo that was incredible. How long did it take for you to start seeing the initial signs of it working?Link

ShishirYou know, there's an important lesson there as well, which is you've got to manufacture your signal. Google acquired the company, and then a group of us went over a few months later, when it didn't feel like things were quite headed the right way. And, you know, one of the first things we had to figure out is, how are we going to measure success? And it wasn't really that obvious, because the path to becoming the kind of business that YouTube eventually became was going to be a very long time. So you had to find a set of intermediate milestones that could get you there. And so we spent a lot of time inventing what those would be.Link

And we decided onget to $1 billion in revenue. We were at 30-ish million then, so it was not clear how we were going to get there. And the other one was be profitable in at least one country, and presumably in the United States. And that, by the way, is a really hard thing to measure, but we just needed something where — our best monetizing market, could we be profitable in that market? And those milestones we hit in about two years.Link

LubaWow.Link

ShishirBut I think objectively, if you look at the state of the business at that time, it was still not a very good business. But we felt like we had achieved the core signal that we could get to the next level.Link

LubaAnd what was it like for you during that time? Were you stressed out? Were you like, "This is definitely happening"? Where did you draw your inspiration and conviction to fuel yourself day to day?Link

ShishirEnormous stress. I mean, I think stress, but in a good way. You're working with a good set of people. You feel motivated. But yeah, it's existential. We had a quarterly board meeting with the CEO of Google, Eric Schmidt, the CFO, and a few of the other executives. And basically every board meeting was, should we shut this thing down? Should we sell it to someone else? So you felt enormous pressure to deliver.Link

You also feel this, especially once an idea gets to that stage, this responsibility. We were already at the point of millions of creators using it and millions of users using it. The business was not very good. And so it was quite possible that it's a very useful service but a terrible business. And gosh, it would have really sucked if we had hung up our hats and said, "That's it, we're not going to do it." And so you feel this propulsion.Link

I went on this plane ride to New York, and I sit down next to this woman, and we're just introducing ourselves. And I said, "What do you do?" And she says, "I work on YouTube." And YouTube at the time was like 150 people. I knew every person who worked on YouTube. She definitely did not work on YouTube. And I said, "Really? What do you do?" And she's like, "No, no, no, I'm a YouTube creator. YouTube sends me my check. So I just say I work on YouTube." And it really stuck with me. And she gave me her business card, and her business card said YouTube on it and had her channel. And it was just a good reminder of, like, you're doing this thing. There are real people. There was a real ecosystem being created out of it.Link

And so your stress is usually not selfish, it's not personal. It's all, you know, it's your employees, it's your customers. And in this case, it was our creators — you were giving them this chance to have a voice in the world. And for many of them, it was completely heartwarming stories of how they did it. And so yeah, enormous stress. Every mistake you made, the cost was not just, "Oh man, I'll go get another job." It was, this thing is not going to exist. So yeah, I think I thrive on stress, so it's okay. But I would describe it as quite stressful.Link

LubaTell me more about the statement of "you thrive on stress."Link

ShishirI think that for me, that kind of stress brings out creativity. I think when the stakes are high, I feel more energy. But it doesn't mean you're not — you know, you have that queasy feeling in your stomach. You have trouble sleeping. You have all those things. But you're also up at the whiteboard drawing an idea.Link

LubaYeah.Link

ShishirAnd so in my mind, there was a small period — for the most part, I think I've worked seven jobs in my career, and I've basically taken no breaks, except after YouTube, before Coda, I took a six-month break, and it was like the worst six months of my life. I just didn't know what I was doing there. It's like, "Oh, I have all this free time," but I really didn't know what to do with it. And I felt like every conversation felt superficial. It felt very low stakes.Link

And so I think stakes are a good thing. Like, hey, this matters. Put your work into it. You're going to stay up a few extra hours and finish that idea. You're going to read that proposal. You're going to hunt for that idea in the pile of ideas coming from inside and outside your company. I think that desire to win — not necessarily to win, but to survive, to see this thing work — I think is pretty healthy. But yeah, it's stressful.Link

LubaI want to potentially come back to your six-month break. But where do you draw that line between stubbornness and conviction? Because, again, you could be knocking on this wall and it's just not going to knock over. And the business might not be it. And you keep going and going, and years pass and months pass. Do you have some sort of framework for how to...Link

ShishirI mean, you definitely want to surround yourself with people who are simultaneously able to support you and are truth tellers and truth seekers. I tend to see the glass half full. Like, I tend to see the launch that didn't work, but it had this one metric that did. And you say, okay, how do we follow that thread? And everybody else is looking at it like, "What? It didn't work." So I tend to be good at spotting those things. But you also want the people around who remind you of, like, what you said it was going to be at this point. And I think you don't want to have too many of them around you, because they can knock you off the broader game. But you need a handful that can do that.Link

And I mean, the job is constantly about deciding investments. Something's not going to work, you've got to stop. So it's not meant to say that persistence at all costs is what you always do. We stopped many projects. I shut down businesses, I pivoted businesses. I mean, the business I actually started at YouTube was called Google TV. And we fully shut it down and rebooted it, and it came back as a product called Chromecast years later, completely different than where we started. And then they stole the name back as Google TV.Link

But I think knowing when to kill things is important. Sometimes the market is telling you something. We tried the same thing, it didn't end up working, we were too early. So you have to be clear about that. One of the things I often ask people is, you know, pretend to be an entrepreneur and then pretend to be an investor and just say, would you invest? I think it's a hard test, because pulling yourself out of that vantage point can be tricky. But if you look at your current business and you say, "I wouldn't invest," then you have to ask why you're doing it.Link

LubaWho were those people who were supporting you and keeping you accountable during those days? Your YouTube days?Link

ShishirWe had a number. There's definitely a lot of folks around. One that I would give a lot of credit to — we had a, I guess we now refer to him as a coach — a guy named Dean Gilbert. Actually, we had two. We had Bill Campbell, who was helping the team a lot, but the one that was closest to YouTube was a guy named Dean Gilbert, who for a while was the COO at YouTube. And then he stepped back as a coach. And he'd just been around multiple companies. He was one of the early folks at Excite@Home, which — people have probably now forgotten, but Excite was one of the best early web search portals, and @Home was the first broadband network. And he had a pretty good eye for what was working or wasn't, and he was pretty deeply embedded in the team. I'd say I drew a lot from him.Link

LubaWhat were some of the top things that you learned from him that come to mind?Link

ShishirI mean, there was a lot about, on the topic of when to go further as a business or not. I think that was interesting. I think probably the most important thing I learned from him was a way to think about personal commitment. There was a time I was roughly two years into YouTube, and I got recruited by another company, and it was an interesting offer, a company people know well. But anyways, I came to Dean and said, "What do you think I should do?" And he gave me this really interesting advice. He listened to me and he was objective. I don't think he was just like, "Stay, just because staying is better."Link

And he said, "You should make three lists." So the first list was your list of goals. What are you trying to get done? And pick your timeframe — I did 18 months. What are the top things you're trying to get done? And then the second list was, take those things and figure out what subset of those things wouldn't happen if you weren't there. So he called it — the first is your goals, the second was your marginal utility list, which was, if you left, these things wouldn't happen, or some of them would happen anyway. And the third list was the purpose list, which was basically, which of these things, if they didn't happen, would you feel really, really terrible about? Not a few months later, but years later.Link

And you're supposed to whittle the list down. So things I'm trying to accomplish, the subset that wouldn't happen without my focus, and then the subset of those that I really think would matter to me years later. And he said, if that list is empty, you should consider leaving. But if it's not, you should consider staying.Link

And so I made this list. And the first list had 15, 20 things on it, and the next list had ten things on it, and then the bottom list had five things on it. And interestingly, there was a bunch — some were personal — but there was one on that list which we hadn't yet shipped. Part of the reason I joined YouTube was I had this idea for a different way to think about monetization.Link

LubaWas it skippable...Link

ShishirThe skippable ads. And this was two years later and we hadn't shipped it. And I felt very confident that if I left, it wouldn't happen. And I felt fairly confident that if it didn't happen on YouTube, it wouldn't happen on any property. And for me, that felt like a travesty. It felt like, this seems like such a better thing. I know it seems small in retrospect, but at the time it really mattered to me. This was the idea. I felt like I would feel really badly.Link

LubaWhy did it matter to you so much?Link

ShishirI guess I view advertising and marketing as the lubricant of society. A lot of people look at advertising as annoying, and they view it as a tax. I think they miss that on the other side of every advertisement is a company trying to introduce you to a product that may matter to you. And I think that if that falls apart, the entire market becomes more inefficient and you end up with much worse outcomes.Link

But anyways, it mattered to me. It just felt, even on a visceral level — I thought about it as, if the skip buttons went away, I'd just feel personally bad that I could have done something about it and I didn't do anything about it. I know now I feel like I work on things with much higher purpose than that, but at the time it really mattered to me. And so that was on my list.Link

It's actually interesting. I left YouTube years later, and I was at this dinner party with a group of people that are not really in tech. They've never really understood what it means to build one of these products. Like, what do engineers do? And so on. It didn't really make any sense to them. And they said, "Oh, you're leaving YouTube? That's interesting. What are you going to do?" And they all had the same question for me: "Does that mean the skip button is going to go away?" And it was interesting to me how important their association with the product was. I think it mattered.Link

So anyways, Dean really coached me on how to find marginal purpose, which I think is not an easy thing to find, but when you find it, it's very motivating. Here's the thing I really, really want to have happen, and I'm quite convinced that if I weren't focused on it, it wouldn't happen.Link

LubaAnd you can probably say that skippable ads is what made ads better, because if your ad sucks, then people are going to skip it.Link

ShishirI mean, that was the theory, right? The theory was, if you make an ad skippable and you only charge the advertiser when the ad is not skipped, then you can rebalance the power. And I thought the idea was very simple because it was basically the same as Google AdWords. The AdWords assumption was, if you don't click the ad, then it's not as valuable, and you reset the balance of power. And now Google has an incentive to find the right ads, and advertisers have an incentive to be more relevant. And obviously, when Google started, the entire ad market was so much smaller than it is now, and they grew the market, because all of a sudden you as a business owner can get people to your products in ways that are very efficient. And it allowed whole new classes of products to exist.Link

And I think that YouTube has done a similar thing. Actually, quite ironically, Grammarly turned out to be one of YouTube's top ten advertisers, with this product in particular. The skippable product is called TrueView. And it's definitely one of those products where if you have a product that requires some explanation, and you can find a way to do it in an interesting way — which is not easy to do — then you can dramatically grow your audience. And from YouTube's perspective, that product went from 0 to $1 billion in about 18 months. So it was an incredibly fast-growing product. You know, now it's tens of billions of dollars on that product. It was a lot of fun to work on. I'm very glad I didn't leave.Link

LubaYeah. Have you used this marginal utility framework on anything recently?Link

ShishirI mean, I use it all the time. Anything I take on, it's got to meet all three tests. It has to be clear what I'm trying to accomplish. It's got to be clear that there's something about my approach that is unique. And then I have to care. And so the choice to sell Coda to Grammarly was definitely one of those. The choice to create Coda was definitely the same. I was quite convinced that if we didn't do it, nobody would build a product like that. And I think we're probably right that it wouldn't exist.Link

LubaYeah, we definitely will talk a bit about Coda and Superhuman. My last question on YouTube is, what did you learn from the late Susan Wojcicki?Link

ShishirOh, Susan. Lots of things. So as background, Susan's the one who hired me. So she was my first boss. Weirdly, this Google TV team reported to her, even though she handled ads, basically because she was a pretty good sponsor for these skunkworks projects.Link

Lots of things to learn from Susan, but I think probably the one that is most interesting to me is Susan had a way of thinking about ecosystems that was uplifting. If people don't know, she's now known for — she ended up taking over YouTube and doing a great job growing YouTube. But at the time, the thing she was famous for was a product called AdSense, which is the ability to put ads on any website, and the publisher, the owner of the website, makes some of the money. The advertiser pays money, Google makes some, and the site hosting the ads makes some. And the idea requires really thinking about an ecosystem with multiple parties and how to create something that isn't a zero-sum game.Link

And many, many people who had gone after markets like that were motivated by short-term gains, or they couldn't find the right balance of — in that case, it was Google, the publisher, and the advertiser — and you had to find a way to balance all their needs. And, you know, for example, it was always hard to manage ad quality on those things. And I would argue that as the ad networks have gone past Google, ad quality on those things has really gone downhill. We're now subjected to lots of very mediocre-quality display ads all over the internet. But the Google ones didn't feel like that. They felt like they were constantly aiming for a balance of what the user wants, what Google wants, what the publisher wants, what the advertiser wants.Link

And so for YouTube, she and Salar were the main sponsors for buying YouTube, and I think they saw the same promise, that this is not a zero-sum game. I remember going to her — one of the earliest major decisions we made on YouTube was about a year in. We opened up the YouTube Partner Program. Up till then, we only allowed major media companies to monetize their content on YouTube. You couldn't run ads on YouTube unless you were a major studio or label or so on. And obviously the ethos of YouTube was completely the opposite of how we wanted to think about things.Link

So we had this big debate about it, and we decided we're going to open up so any creator can apply to be part of this partner program. And nowadays, you know, it's a pretty straightforward process. If you want to start a podcast, you can go on YouTube and make money. And we had a big debate about how to do it. And then there was a big debate about how much revenue to share with the creator. And remember, at this time, YouTube's losing hundreds of millions of dollars a year. And even though that was the case, we decided to create a revenue share for advertising. It's actually a 55/45 split.Link

LubaWhy were you guys convinced that that was the right decision?Link

ShishirBecause of this — what Susan described as the ecosystem thinking. And by the way, it's still the case that I think YouTube is the only major media property that has a publishing partner revenue share model. Everything else does it with side deals and, you know, with creator funds and so on. But I think the way YouTube did it really stood out at the time, and we ended up turning it into one of the YouTube values. We called it "even playing field and fair play." And it was basically this idea that, at the time, the creators coming to YouTube were all quite small. But we believed that if we could create an economic model that you could rely on and that we would stand behind, then you would invest. But if we made it arbitrary, if we made it look like a fund that could disappear next year, or worse, if we didn't pay you, then we would see this race to the bottom of what would happen with content creators.Link

And I think Susan had seen that happen with AdSense and the web market. Ironically, even including YouTube — YouTube was a big AdSense customer. YouTube probably wouldn't have existed without AdSense. And her view was, you should share revenue even when it hurts, because the broader flywheel will move. I think she was very good at being willing to stick through an ecosystem like that.Link

LubaIt's expanding-the-pie mentality.Link

ShishirYes, yes.Link

LubaThat's amazing. So you had an amazing chapter at YouTube. You leave a very hot executive position there, where you really made the business work, and now you're building Coda in stealth for a few years. What was that like? What were you telling people, your friends, about what you were doing? How did it feel?Link

ShishirYeah. I mean, back to where we started, it was definitely the case of being cursed with an idea. As I left YouTube, there were a number of different things I could have gone and worked on. But once I got the idea for Coda in my head, I just couldn't stop thinking about it. I had a notepad by my bed with all the sketches of what it should look like. I mean, our investor pitch was basically just walking through those sketches. It's like, here's what we want to build.Link

And it was interesting because the core idea started with: the world of productivity hasn't meaningfully changed in decades. And we still had antiquated metaphors, like documents and spreadsheets and presentations are all separate things. And the second observation was that the way people use productivity tools — the way people use documents, spreadsheets, presentations — is actually like applications, but they just don't call them applications. It's how you do planning. Every CRM system starts with a spreadsheet. All of these things were actually what these tools were being used for, but not what they were built for.Link

The market size was really obvious. This is obviously a category which can support a large business. And the question was, does somebody have a product idea that'll work? And so I just got really obsessed with, I think I have a way to do it. I could picture it, and, you know, we sketched out many elements: this is what the surface is going to look like, and we're going to rethink the formula language, and here's all the different things we can do to make "a doc as powerful as an app." That's the tagline for Coda.Link

LubaAnd take me briefly to that six-month sabbatical. Did you get pulled by this idea on the sabbatical? Or were you forcing yourself to take a break and take a rest for six months, and then you came up with...Link

ShishirYeah. So I was pretty sure I was going to end up taking another job at Google. Sundar had just taken over as CEO, and he had a bunch of different things he wanted me to do. But I was just taking the time. You know, I'd been working really hard for a while. I'm going to go and explore ideas. I'm going to relax. I'm going to travel.Link

LubaYou already had kids at that point?Link

ShishirYeah, they were six and eight.Link

LubaOkay. Hang out with your kids.Link

ShishirYeah. They weren't in robotics yet. They were still small.Link

Anyway, I was assuming I was probably going to end up staying at Google. There were a few other things that people were recruiting me for, all of which looked like some version or shape of YouTube-like things. And then I had a friend, this guy Alex, who ended up being my co-founder, who was one of the best engineers I've ever worked with. And he was hanging out as an entrepreneur in residence at General Catalyst with Hemant Taneja. And he had asked for my help in figuring out what idea he should work on. So amongst all the other things I was doing, I was helping him brainstorm ideas.Link

LubaAnd you gave him the same advice, that you shouldn't be starting a company if you don't have an idea?Link

ShishirYes, yes. He ignored it, but it didn't matter.Link

Anyway, so we started brainstorming ideas and we started talking about what would eventually become Coda. And I think we both got really excited about it. Once you could picture it, it felt like a really obvious product. We had both worked at Microsoft in the past — we'd worked on the Office team — and had a pretty good sense of why those products were not evolving, and what, even if it seems obvious in retrospect, changes were going to be hard for those products to do.Link

And interestingly, I didn't want to start the company. So I told him, "You should start this. I'll invest, I'll help. And if you don't want to do it on your own, I'll help you find the right home for building it." And so I went and talked to everybody. I told Sundar about it. I talked to Satya about it. I talked to the Dropbox folks, the Box folks, every company I could find. I didn't want to start anything. I view starting companies as the hard way. It's the hardest way to get an idea...Link

LubaIt's a calling.Link

ShishirYeah, it's a calling. And so, interestingly, we would take the idea and I'd go to these different companies. I'd say, "Here's the idea," and I'd sketch out what the product looks like. "And by the way, here's an engineer who's really, really good." And everybody wanted to hire Alex. And they would kind of tolerate the idea, but they all thought the idea was kind of stupid. Not in a moronic way, but like, your path to that working is so slim. Why don't you guys just come and join us and we'll work on this other thing we're doing?Link

And as we kept talking about it, we started realizing, oh, that's because nobody's willing to put in the time to build the core of this product, and it's not going to be an easy one to build. And go back to Dean's questions — all of a sudden the marginal utility test was clear. If we don't do it, nobody's going to do it. And it was about as strong a signal as I could get of that.Link

And I remember my wife and I were driving up to Napa to visit a friend, and I said, "You know, I think I'm going to start this company with Alex." And she said, "Oh, I've known that for weeks." And I said, "What do you mean?" And she says, "Every free moment you've had, you and him are working on your stupid idea." It was not an obvious one, but it was clear that we just couldn't stop thinking about it.Link

LubaHow did you meet your wife?Link

ShishirOh, we met in college.Link

LubaOh, nice.Link

ShishirYeah, I was actually younger than my daughter is now.Link

LubaOh, wow. That's crazy. So you guys were college sweethearts. And then you traveled with her to her residencies too, right?Link

ShishirWell, she traveled with me, which is not... She wasn't happy about it, but she moved to Seattle to do her residency.Link

LubaBecause you went to work at Microsoft?Link

ShishirYeah.Link

LubaGot it.Link

ShishirAnd it turned out great. We had a great time there. Our older daughter was born there. But yeah, we had all our adventures together. We were 19 when we met.Link

LubaWow. For someone listening who's looking for the one, what's your dating advice to them?Link

ShishirOh, that's too hard a topic.Link

LubaYou don't have a framework for it?Link

ShishirYou know, I don't really think — if you're lucky enough to find someone at that age, it's great. But I feel like we have a bunch of friends who were college pairs, who found each other then, that are still happily married. I also have a bunch where it was completely the wrong choice and they pressured themselves into it. I have people that said no to every person along the way, and they find themselves a little bit older and not finding that right person. When it happens, you know. There's some amount of luck in it.Link

The other — I feel like these days... At that time, dating apps didn't exist. And so I don't even know how to think about it now. At the time, if you ended up dating a dozen people in your life, that would be a lot. And now I have friends who tell me that they meet a dozen new people a week. And I was like, if you had that level of choice, how would you ever decide anything? And I don't know if it's good, actually. But anyways, that's a different world. And this is a topic I resist giving real advice on.Link

LubaI'm surprised you don't have a framework on it, given that you seem to have a framework on everything.Link

ShishirYeah, I wish I had a better one for that one.Link

LubaLet's jump back to Coda. So you were in stealth with Coda for multiple years — for three years, I believe. And then you were building the company for eight, ten, eleven...Link

ShishirYeah, 11 years. We combined with Grammarly in year 11.Link

LubaGot it, in year 11. Tell me more about that decision to merge with Grammarly. I know that someone from your board told you...Link

ShishirYeah, exactly that.Link

Luba"You need to talk to the founders." And from what I've read, it sounded like this very swift decision, where you had a strategy around AI productivity, and they had the same strategy, and you were like, "Man, we should combine our strategies because they're the same."Link

ShishirYeah. I mean, I would definitely give Hemant a lot of credit for recognizing two groups of people who, in his mind, were headed the same way, but it was not obvious to either of them that that was the case. And so he had called and told me — I was in the middle of raising our next financing round, and it was about to finish — and he said, "Hey, before you close anything, why don't you go meet with the Grammarly founders?" And my first reaction was, "I'm not sure I get it. I love Grammarly as a product. I've loved them as a business, but it seems different than what we're working on." And he said, "No, you're probably not understanding where they're headed with this product, and you guys should spend some time together."Link

So I flew up to Vancouver and spent some time with Max and Alex, the founders of Grammarly. And I learned a lot about Grammarly that — it sounds like you've probably read and seen — but first off, it's a much bigger product and business than people think. It was already tens of millions of daily active users. It was over $700 million in revenue. It was a much more scaled product than people assume.Link

But the main thing they told me that really stuck with me is, people think that Grammarly is about grammar, but it's actually not. The core technology was not really about grammar. It was about bringing AI to the edge. And the way they described it is, if you picture the product a little bit differently: here's a product where Grammarly sees a million unique applications a day. Every desktop app, mobile app and web app that you're in, Grammarly is there, observes what you do, annotates it in a way that's unobtrusive to you and to the application, and then makes changes on your behalf. Their view is that Grammarly was the OG agent. It's the original agent.Link

And they described the main technology of doing it as the AI superhighway. "We built this highway where you could bring AI right to the edge, where people work." And in their view, there's one big problem: "We're only running one car on that highway. It's the one with your grammar teacher in it." And it turns out to be a really valuable car that tens of millions of people choose to use every day and spend hundreds of millions of dollars on. But it's a vast underutilization of this infrastructure. And as they started talking about it, we started realizing where they wanted to head with AI was actually very related to a set of ideas we were working on at Coda. And so that ended up feeling like a clear match.Link

They also mentioned that there was kind of a straightforward path for Grammarly — a popular product doing quite well. Just take it public, continue down the normal path. But they felt like this was too big an opportunity to pass up, and to do it, they wanted someone with a bit of a founder's mindset to run it. And so they wanted to buy the company, but they also wanted me to come run the combined company. I was flattered. It was definitely unexpected. But it kind of felt like starting a new company. It's the same tests: what are the goals? What's the marginal utility? What's the purpose? Am I going to care?Link

LubaAnd you didn't really feel like you had an exit.Link

ShishirOh, it definitely did not feel like an exit. I think selling Coda to Grammarly is definitely a start — I think people say that a lot, "a start, not an end." This is, by any definition, a start, not an end. And the mission ahead of us is definitely what I signed up for. But it was the same thing. I couldn't get it out of my head. It just seemed like such an obvious idea in retrospect.Link

And, you know, one fun stat is Grammarly does over 100 billion LLM queries per week. Per user per day, it's thousands. So for most people, if you use Grammarly, we are your number one source of LLM queries. You can't possibly type into ChatGPT anywhere close to what we're doing on your behalf using Grammarly, because we work at the pace of typing. So as you type, as you load every page, we're constantly figuring out all the different ways we can assist you.Link

And if you think about it that way, I often describe Grammarly, or what we're building with Superhuman, as the last mile of AI. We're taking all these tools and actually getting them to you in a place where they can actually affect what you're working on. And it's kind of like — we talked about skippable ads and so on — once you get that picture in your head of what it would feel like if my agents worked next to me, and it wasn't just my grammar teacher on my shoulder, it was my sales coach and my favorite author and my professor, and they were all with me, helping me along my path... And it feels obvious that if we don't do it, nobody will do it. Then you just kind of have to do it.Link

LubaWe're going to jump back to what is now Superhuman. But I wanted to ask you on that acquisition front: what about your startup founder pride, when you were considering joining Grammarly and becoming the CEO of Grammarly? Were you not at all like, "Oh, but I can do it myself at Coda. I've been building this for 11 years, I should continue"? How did you deal with those emotions, if any of those were present?Link

ShishirI mean, it's sort of back to — I don't have any obsession with starting companies. In my mind, when you have a great idea, you should be obsessed with getting it to the world the fastest way possible. And I think people who start with a vantage point of "I'm only interested if I own 100% of the company, or if I'm the only founder," or so on, I think they end up in the wrong spot.Link

So in my career, I've started two companies from scratch, and that's Centrata and Coda. But I would say my role with both YouTube and Grammarly, from my perspective, feels like a founder. I feel like where we were headed — it just happened to be the best platform to get that idea done. And I think that has led me to a better spot.Link

I get asked by students all the time. It's like, "Oh, I have a list of startup ideas." And I say, please don't have a list of startup ideas. Just have a list — call it "things I'd like to see different in the world" — and then be open to the fact that maybe the best place to work on it is in a startup. And when the right thing comes along, and this kind of weird-shaped product YouTube comes along, and that's maybe the right way to change how to think about media and advertising and so on — or when something like Grammarly comes along and it shifts your viewpoint on a very different kind of AI platform — you have to have the mindset to say, "I don't need a sense of ownership. I need a sense of purpose, I need a sense of accomplishment." And all those things tend to work out well. But I don't think I had any of that startup pride.Link

LubaIt sounds like the same idea as being attached to outcome versus being excited about the process.Link

ShishirYes. I see lots of entrepreneurs like that. They get very obsessed about ownership, about control. And I think that mindset will lead you astray. Do what it takes to make that idea happen. I think some of my best ideas turn into investments, they turn into companies I advise. Like, I had a whole theory on how I thought bundling should work. That turned into a great relationship with Daniel at Spotify. And in that case, I didn't have to join the company at all. And I still feel like I got a chance to see this idea turn into something. So have your list of ideas you want to see different in the world, and I think you'll feel a lot of success as they happen if you remove ownership from the equation.Link

LubaSo since you mentioned bundling — I know you're really passionate about bundling. Give me the pitch for bundling and Superhuman.Link

ShishirOh, sure. I mean, I think Superhuman is one of the easiest pitches. Maybe backing up, I think bundling is one of the most misunderstood economic topics out there. I wrote a paper maybe a decade ago now called "Four Myths of Bundling." Actually, Daniel from Spotify asked me to write it down after we talked about it a bunch of times. The paper starts with the story of a person that says all these things about bundling that turn out not to be true, and the person I was picturing was just myself. They were all things I had said that turned out not to be true.Link

For example, people think that bundles should be as narrow as possible, like you want bundles with only very related products. But actually, Amazon is a great example of a really broad bundle. If you go ask the folks at the New York Times, "What should the New York Times be bundled with?" — if you asked any New York Times subscriber, they would say, "Oh, I'd want it to be bundled with The Washington Post, because they share some ethos, they share some subscribers," and so on. But it's actually the opposite. Myth number four in the bundling paper is: the best bundle is one that minimizes superfan overlap and maximizes casual fan overlap.Link

And so the idea of how to think about bundles, how to think about subscriptions — first of all, as an economic and math topic, it's fascinating. And once you see it, you'll see it everywhere. That's why health insurance works. It's honestly why government and taxation works. They're all forms of bundling. But this core principle of bundling is about minimizing superfan overlap, maximizing casual fan overlap. You're bringing together products that themselves have very deep followings, and then using them to build bridges to people that have what you would think of as casual fanship to those products.Link

And there's a bunch of different examples of that in the world. Obviously, Spotify has done a lot of it. And what we're doing at Superhuman is definitely that pattern. We're taking products that are beloved in their categories — products like Superhuman Mail, which is one of the most beloved email applications in the world, or products like Coda, which is now Superhuman Docs, which is, I think, the most productive way to run a team in an all-in-one space. We just bought a company, GPTZero, which is the number one authenticity company. Of course, Grammarly. And our newest product is Superhuman Go, which is the platform layer of Grammarly. Each of them has a distinct promise, a distinct audience, and they have to do a good job of winning that person. But then once you do, you start to see people that say, well, if I can get them all together, then I'll prefer to get them as a bundle. And I think that's bundling playing out.Link

And we just launched the newest version of AI auto drafts. And it's a really hard problem, because I think it's one of those things where if it's kind of good but not great, it's actually annoying. It's like, don't draft mails for me that aren't good. We just hit a milestone: over 60% of mails we auto-draft are now sent, with half of them being sent with no changes. And even there, I think you want that number to be higher. But it is a remarkable improvement from what the world generally expects. And in that way, mail stops being just the place for me to work and starts being the place for me to work with, in this case, what we call the mail agent, that's working alongside me. And then as we go forward to where we're headed with Superhuman and with Go, it should become the place for all my agents to work with me.Link

To give you an easy example, today that mail auto draft understands everything about me, understands everything about our correspondence. So, for example, when it drafts an email between us, it'll follow the tone for how our conversation has gone. But as we bring the rest of the agents, it can do things like know what's in our sales system, or know what's in our support system, or read all the documents. I'm responding to someone and it says, "There's a customer, they had a support issue yesterday, so make sure you say something about that in the email." And all those things happen by being able to bring all of what we call agents — but you could think about them as virtual teammates — into that same experience. So I think we've actually made amazing progress, especially in the last year, but I think we have a lot more to come there.Link

LubaYeah. I think the superpower that comes with AI is the superpower of context. If you can have your AI have context across everything that you work with, then you can truly become superhuman. But it's that disconnect that doesn't allow us to be there just yet. So I'm really excited for...Link

ShishirYou want context, but I think the trio is ubiquity, proactivity and context. We sometimes refer to it as knowledge. You want this agent that knows everything. You want it to actually work where you work, which turns out to be really challenging. And then you want it to do it proactively. You don't want to have to ask. With most of the AI systems today, I think context has become more obvious for people: we want AI systems that have read all of our past writing, that have read all of our systems and so on. But the idea that it should take action before I ask, and that it should be present in the surface where I work, is what dramatically uplifts the utility.Link

LubaYeah. Is the surface area challenging because you have to have an open API to the apps that you're working on, for the agent to have context? Or is it something else?Link

ShishirSo the way Grammarly does it, and what we're turning into Go, is we've built a set of ways to work with basically every different application, and each one of them is a little bit different. The way we work with Google Docs is different than the way we work with desktop apps like Apple Notes or Word or so on. And so in each case, we have to do it a slightly different way. And that's really the hard engineering work that the Grammarly team did that we're turning into a platform.Link

And then in some cases there's a limit to that. And for the most common use cases, we're building the surfaces ourselves as well. So that's mail and documents, the two most common. And in those cases there's just much more you can do.Link

But can you just think about it in human terms? You know, can I go to ChatGPT, Claude or so on, and say, "Draft this email for me"? That's useful. It gets more useful when it can read all of your past information and draft you a better email, but you still have to remember to go do that. And you'll do that for your highest-stakes email. You'll do that for something where it really, really matters. But if it can do it right inline, if it can do it in collaboration with you — so as you're making edits, it's learning and changing — if you can do that all the time, then this goes from being a kind of useful thing to a life-changing thing.Link

LubaSo for instance, with YouTube, you were really convinced that the business would work. You had the convincing idea of skippable ads, and you were very persistent with making it happen. Is there anything like that at Superhuman now, where you're really trying to drive it forward and you're getting a little bit of doubt from people around you about the fact that that's the way to do things?Link

ShishirI think it's different. I don't know if I have something quite like the skippable ads experience. To be honest, that was so counterintuitive. It was so non-obvious. Putting a skip button on ads — every salesperson's view was, you put a skip button on ads, 80% of people are going to skip the ads, you're just going to see revenue drop by a factor of five. And they were right about the 80%. That is about where skip rates landed for YouTube ads. They were wrong about revenue, because they didn't understand that the advertiser sees the views that are not skipped as more valuable. So overall it doubled the RPMs, the core metric for advertising. So there was an insight that was different.Link

I think probably the equivalent here is what we call the value of assist — the value of being present right where the user is. And I think there's a viewpoint that that's specific to grammar. But I don't think it is. I think the idea of having a collaborator that works with me where I work is not just about grammar. And at this point there's not that much resistance to it. But I think early on that wasn't obvious. Like, what are the other ways that people would use that? Why would I want an agent that works alongside me like that?Link

And I always give people the human metaphor of: I write something important. I send it around to people for feedback. What if I didn't have to do the step of sending it around to people? What if they were already giving me that feedback? I have written this thing I need to send to the lawyer for review. What if the virtual lawyer gave me the first pass as I was writing it, and stopped me from doing something that maybe we were never going to agree to? I think that idea of "I'm going to build a team of virtual collaborators that work like Grammarly" — I don't know if it's as unintuitive as skippable ads. I think it's more obvious, but it definitely had some early reactions like that.Link

LubaWhat are some lessons that you are still learning as a leader?Link

ShishirI think how to make teams work well together is a topic — and interestingly, there's a group of people who view AI as a future state where we're not going to need teams and employees anymore. I view it the exact opposite way. I think our teams are getting bigger, faster. I think now the mix of virtual teammates and human teammates makes every teamwork problem harder. I think that's an area where — as you know, I spend a lot of time thinking about rituals and how teams operate.Link

LubaYou're writing a book about it.Link

ShishirI'm writing a book about it, *Rituals of Great Teams*. And I think it's still a topic where I'm constantly learning some pattern that is different than what we're doing, and hitting some roadblock. That's definitely a topic area.Link

On a personal level, I think every leader has their core skills, and you want to amplify your core skills and continue to work on the ones which are maybe less obvious. I think for me — like I said earlier, I'm the optimist in the room. How do you take people through that journey? Maybe people that aren't inspired the same way. So I'm constantly working on different ways to navigate people that see the world through different eyes.Link

LubaIt sounds like finding the right language or the right tools to inspire people that maybe don't think or view the world the same way...Link

ShishirOr hold them accountable, or drive urgency. I mean, it's very related to teamwork, but I think it's a constant area of growth for everybody.Link

LubaAnd how are you learning this? What are your strategies?Link

ShishirI mean, for both, I'd say you want to surround yourself with different people. I think a lot of people work on their companies and they lock themselves in their office, and they never interact with others. This is actually back to Dean Gilbert. One of Dean's lessons for me very early on — he says, "You spend too much time in the YouTube building. You need to get out." And that could be with customers and partners. That could be with peer companies. Interestingly, I used to have this standing breakfast, coffee, with Chris Cox over at Facebook. And Facebook was seen as a direct competitor. But Chris and I had a great time, and we would chat about things and learn from each other. So I think it's part of why I joined the Spotify board and the Walmart board. You know, they're all time commitments. But if you're open to learning from others, you can learn from lots of different kinds of people.Link

LubaWho are some people that you're learning from the most right now?Link

ShishirOh, man. Everybody. Daniel's high on my list — Daniel, the Spotify founder. His journey's been amazing. I think he's incredibly talented at all stages of a business. He can look at world scale and he can look at detail scale. Has a real learning mindset, very positive attitude.Link

I'm lucky to have a great board. Hemant — I always learn from him. He always has a different way of thinking about things. The way he did it with Grammarly, the way he makes connections, I think is really magical.Link

Reid Hoffman — I spend a lot of time with Reid. He's always got a unique view of products, but really about people. He just sees people in a different way. He made this observation that I really agree with, which is that you can often see the reflection of a person in the product they create. So he created LinkedIn because he sees every relationship as having unknown opportunity. And so he invests in them. He's connected with everybody and knows a little bit about lots and lots of different things. And he finds an excuse to interact with people in lots of different ways. And so he's just always open to learning from people. And I don't think there's many people at his level of accomplishment who will pay as much attention to an individual engineer or product person or whatever, and hear their perspective and ideas with so much openness.Link

LubaAnd to close it out: who are you still becoming?Link

ShishirWhoa, you ask tough questions. Who am I still becoming? I read a book as a kid — there's a series of books called the Alvin Fernald series, and I've learned that nobody read these, so somebody will have to look them up. But they're stories of this kid who's an inventor. And every book has the same basic plot structure. He invents something a little bit crazy and it's kind of cool, and then it has some disaster. And every one of the books — it's a really fun book, but every one of them made me feel sad at the end.Link

LubaOkay.Link

ShishirIt's like, he invented this thing, but it didn't work. And in my mind, all of my role models, idols and so on, they're all people who found ways to take ideas and make sure that they happen in the world, and do it in a way that propels the world forward in different ways. And, you know, I just want to be the happy version of Alvin Fernald.Link

LubaI love it. Shishir, thank you so much for your time.Link

ShishirAll right. Thank you.Link