This is the Neo Show, where the minds shaping tomorrow speak today. Brought to you by Neo, the browser redefining what AI can do for you.
Okay. Welcome to another episode with me, the CEO, and also we do that together with Neo Show. Neo Show we actually have been doing that on a weekly basis for quite a few weeks. The goal is actually to talk to people about AI. Many of them are founders, CEOs, and some of them are investors, professors. So that's our goal. But today I'm going to, you know, very happy to have my friend Gogan. I first knew about Gogan when he was actually a founder CEO of another startup. I let him talk about his experience, Sprig, you know, because I was at Greylock as an EIR and then people say, hey, you know, our lunch show came from Sprig. So maybe you can introduce yourself about your background first and then before we get into, you know, more media stuff.
Sure. Hey everyone. I've been a founder for about 15 years, almost a little more. I started my career by co-founding Udemy, which today is the largest enterprise and consumer learning company for adults. Now then second, I started Sprig, where Howie was at. When Howie was at Greylock, they backed my company. We raised $60 million. It was a food delivery company. It was like a ghost kitchen. We had a central kitchen in San Francisco and one in Palo Alto where we served food hot, 15 minutes ready to eat.
Is that similar to Travis's current company or different?
It's in the same general space. Yeah, I would say, but the difference is that we were an actual restaurant and we did not... Travis's company. So Travis Kalanick, CEO of Uber, ended up starting a different company called Cloud Kitchens. Cloud Kitchens today is a real estate play primarily, where they have kitchens all over the world in opportunity zones. So like low-income neighborhoods where they can take the real estate that's relatively cheap and upzone it basically, or like increase the value of it by turning it into ghost kitchens that then deliver food to you. So, if you order on DoorDash or Uber Eats or whatever in a given zip code, it's very, very common that your food is not coming from a restaurant, a retail restaurant. It's, you know, probably 80% do come, I don't know what the percentage is, but a significant percentage come from retail restaurants, but a sizable percentage are coming from ghost kitchens. These kitchens that don't have any, no person can go in and just pick up food there. And Travis runs a business like that. Sprig was fully vertically integrated. So, we were a ghost kitchen that also owned the real... like we didn't own the real estate, rented the real estate, but and had an app and we managed the menu and had, you know, we had like at our peak we had like 1,300 employees. Because we had, we hired all the delivery drivers. We had all of the cooks in the kitchen. And then we had a headquarters staff of about 60 to 100, you know, sort of knowledge workers.
And what, and then you sort of, what happened to the company? And then you put it on the LinkedIn that this is the, you know, part of the journey you learned the most. I actually was always curious why did you say I learned the most from that experience?
Yeah, I mean, you know, when you're... I was like 21, 22 years old when I started Udemy, you know, I didn't know anything. So, and we happened to have quite a bit of success. It was extremely difficult. We had a really hard time raising money. We all went into debt. It was a very hard journey. But at the same time, like when you have success, you don't really know why sometimes if you've never had true failure. So Sprig was really interesting because we had more success than Udemy in the first two and a half years of Sprig. Sprig grew extremely quickly. If you lived in San Francisco at that time, like, and you were in the age demographic, you know, sort of 25 to 40 year old, like single or married, like that demographic, like you used it. We had insane penetration in San Francisco market.
Right. Yeah, our peak was February 2016. So yeah. And then Uber Eats launched their DoorDash competitor. So Uber Eats had many iterations. Most people don't know this, but eventually they copied DoorDash. And when they copied DoorDash, they came into the market subsidizing the market. They probably spent billions of dollars subsidizing the food delivery market because they believed, just like in ride share, they assumed that the way to win food delivery was going to be keeping the ETA as low as possible. So they tried to do 30-minute deliveries or less and keeping the price as low as possible. So they subsidized the food cost as well a little bit and they took 30% from the restaurants. So this is a slightly different model from DoorDash. Essentially that killed Sprig's advantage. Sprig's advantage versus DoorDash or Postmates was that we were fast and we were healthy. We were more affordable. Now Uber Eats got rid of the fast, right? Because we were like 15 to 20 minutes. They were 30 minutes. Not enough of a difference for a consumer to care. And they got rid of the cost because they selected restaurants that were cheap. And they marketed and they offered delivery for a different type of restaurant than DoorDash, which was a more premium experience at the time and really wasn't SF. It was mostly Postmates and Postmates was super premium and expensive. So, they killed the competitive advantage of Sprig at a time when Sprig was facing major pressure on margins. We were really struggling to keep up with the rising demand meant rising costs and we were not getting the efficiencies at scale that we were hoping yet. And so we were focused on our margins while Uber Eats was coming in and taking all of our top-line revenue. And so from that point on, the week Uber Eats launched literally in SF, our business went from growing, you know, two to four percent a week, which is an insane growth rate. If you just do the math on compounding, you know, you're talking, I don't remember, but it's like 5 to 7x a year. It's very fast. Not in the AI era. Pre-AI era, it was very fast. But we're going like 2 to 4% a week. And then we went to declining 2 to 4% a week.
What is the number one lesson you would have if you would have done it all over again? And then still, do you think there is a chance for Sprig?
Totally. You know, it's a complex lesson. So what I would say is, you know, VCs always talk about how the number one thing in a market, you know, they always debate between people or market. And obviously as a VC that makes sense, you know, you either get into food delivery or you go and you invest in AI in 2010, which would have been a very bad decision mostly, or you focus on augmented reality or virtual reality or you focus on self-driving cars, like whatever you decide to do. Food delivery was a huge market, right? Obviously, between Uber Eats and DoorDash, we're talking about tens of billions of dollars in exits. Probably by terminus, like in 10 years, we're talking about we'll have at least one company over hundred billion dollars in that market, right? And so huge opportunity, but there's something that that misses, which is the approach. And the approach to the market actually matters as much as the market itself. And of course I learned through that experience that the approach to the market was actually the defining characteristic of Sprig's failure. Now the other thing that makes it complex is Sprig still could have succeeded with its approach. There's another company called Thistle that was started around the same time that did not do a lot of the mistakes that Sprig and Munchery, our other main competitor, made. Essentially, if you're not on a venture speed trajectory, you're not growing at 3 to 5x a year with like the chance of 80% gross margins or 60% gross margins, then you have to raise less capital. You have to make that capital last longer. So you don't do 18 months between rounds, you do like two and a half, three years and you have to get to profitability faster. Sprig could have gone that route and if it did, I think it would still exist today. Now there's another question of how big of a company would it be if Sprig existed today. I would argue it'd be a few hundred million. So it wouldn't be a multi-billion dollar company. Probably it'd be very hard for it to get there. Thistle is just a few hundred million company. But so for the VCs, would it have been worth it? Probably not. You know.
There is a path for hundreds of million dollar valuation. Is there any path for multi-billion dollar valuation with all the hindsight?
It's possible within the approach, but the best thing would have been to change the approach. The best thing would not have been to have a vertically integrated food delivery product, but rather to take one slice of food delivery, whether it's real estate like what Travis did with Cloud Kitchens or it's just the food, it's just the delivery part itself, which is what DoorDash, Postmates, and Uber Eats did. By taking the full stack, we made it near impossible to get to a few billion dollars in 10 years.
Interesting. So Travis's approach would have been an interesting alternative, right?
Yeah. I mean, I think he... we'll see with Cloud Kitchens. But with Uber Eats, obviously, I mean, it's a $50 billion business probably valuation. Maybe 20, 30, I don't know.
Okay. So, we spend a lot of time. We're only halfway into the intro of your experience. Now, you know, Maven, tell us more about why did you start a company? What's Maven doing? A lot of people know Maven, but, you know, probably give us insights in why do you even do it in the first place. Yeah, I'm actually very curious. Maven is a pretty new brand and it's just gaining steam, so I will take zero offense to honesty here. Do you mind if I ask the audience real quick? Two questions. One, how many people have heard of Udemy?
It's like pretty much 90% if not 95.
How many people here have heard of Maven?
All right, 10%. By the way, we're not talking about Maven Apache, the like software product, just to be clear.
So, Maven Learning is an online courses company. Okay, so let me explain Maven. So when we started Udemy, we nailed the market at the time, which was for video-based learning. You know, you go on Udemy, you go on Coursera, you go on LinkedIn Learning, you pay whatever amount of money for a course and you can watch videos that help you learn a subject. Transformative for the time, but in my opinion, it's not the true transformation that education needs. College is rising at an insane cost. It's super expensive, and adults today don't have a good way of keeping up with modern technology. So, with AI, where do you go to learn AI today? You know, are you going to go to Harvard or MIT's like extension program? Are you going to get an MBA? I mean, what does Harvard know about AI? Like, they invented AI. So, on that front, they're very, very good at the very core research part, but they don't use AI. They don't know how to implement it at your job. They don't know what the trends are and what the business opportunities are. Like, most universities suck at this. So, who's going to teach everyone here in the next 10 years, your entire job is going to be transformed by AI. Who's going to teach that? And the answer is Maven. I feel like 40% confident now that Maven will be the number one place that you will go to learn AI. So why? We have two major differentiators from Udemy and Coursera. The first differentiator is that we have real experts who know the field. So like literally the head of product at OpenAI, one of the heads of product at OpenAI teaches on Maven, right? We have someone who was probably one of the top product managers, the first product manager at Stripe teaches on Maven, right? We have figured out how to curate a top tier set of experts. Very different from Coursera or Udemy, right? Coursera is washed up professors, right? They don't know anything these days in the business world at least. And then Udemy is literally like random people on the internet. There's no curation whatsoever. Maven has nailed the curation of the expert. That's one thing. So this audience is primarily experienced professionals. You don't have time to spend 25 hours on YouTube to try to get the 10% of information that's useful to learn about AI. Instead, Maven offers you a two-week program with an expert that in like 12 hours or 15 hours of investment will upskill you dramatically in AI. That's the first thing. The second thing that Maven has nailed is it's cohort-based. So, it's live. So most of you, at least half this audience, your entire life is run by your calendar at work. And what ends up happening 24 hours before the day of your, you know, before today, for example, 24 hours in advance or 48 hours in advance, your 30-minute work blocks that you had on your calendar are gone. Someone takes them. Someone says, 'Oh, you were going to do deep work that day. Sorry, let's go into a meeting.' Right? And so, Maven essentially solves this problem of accountability and learning by putting it on your calendar. It's a live event with the instructor. You get to ask questions. There's other people in the room who are at your peer set who are learning with you. And it's a premium experience right there. You know, it's high value. So, that's what Maven is. We're just getting started. We recently in the last 2 or 3 years really built up our AI catalog and have started to become the number one place on the internet for AI and that's what I'm building today and I'm CEO of the company.
But you started a company what, 2019?
2020 when you started a company, were you thinking about AI or can you tell us why you started this company? What was the thesis back then?
Yeah, it's a great question. So, you know, when we started Maven, the thesis was that Zoom was a foundational new learning platform that essentially, like pre-Zoom and Zoom existed, we all used WebEx like 10 years ago, 20 years ago, whatever, live education was possible for a long time but the adoption wasn't there. There weren't enough users who were on Zoom calls on a regular basis to be learning live from an expert. But we know from education theory, like there's a ton of learning science on this, that live learning from an expert is the number one way you can learn. Actually, the number one way you can learn if you can afford it is a one-on-one coaching. Insanely expensive. If you want to learn AI via one-on-one coaching, you're talking about $30,000, right? It's not possible. The number two way to learn is by learning live with an expert in a classroom setting. So, I saw that that was possible actually in 2019 and was looking at it. And then when the pandemic hit in 2020, I said, 'Oh, shoot. I got to stop looking at this thing. I got to do something about it.' So started Maven cohort-based learning. The problem was, as you all remember, 2020, 2021, 2022, huge boom in tech and then a huge crash and Maven was really stuck in that crash. So we had a great first year and a half. We had a generally right thesis about cohort-based learning. The product did pretty well, made $4 million in its first year, and then all of a sudden the crash came and we were not... and we had raised $26 million. And our business wasn't perfect. It was good, but it wasn't perfect. We hadn't quite nailed the product. We hadn't quite nailed the customer segment. And so, from the Sprig days, I learned that in those moments, you get off the 18-month timeline. So instead of spending money like we were going to raise another round in 18 months, I took our burn all the way down to about 200 grand a month, cut about half the team and said, 'We're going to ride this out because I still believe that cohort-based learning is the future.' And then it took us three... we still have 10 million of our $26 million that we raised four years ago. That's unheard of in startups. You know this as a startup VC, former startup VC. And we kept the money in the bank and we just burned low amounts of money and kept building. And then of course, and we were succeeding even before AI came around, but then ChatGPT came around about 3 years ago and all of a sudden there's this huge need for upskilling in AI and who was the one company that decided to stick it out? Like all of our competitors pivoted out of learning. Who was the one company that stuck it out? It was Maven. I felt like an idiot for a few years and then finally our time came. And so now a new set of learners are coming on Maven in droves to learn AI for product management, how to build agentic applications, you know, how to use AI in marketing, right? We are starting to become the place for that. We still have a couple years left before I think we are the true winner, which is why I gave us a 40% chance of winning the market. But I think we've really done a lot and kind of got lucky that the timing worked out and just sort of were lucky that we had a long runway so we were able to take advantage of the opportunity.
So basically when you started this journey, it's more or less you saw the Zoom, you know, the sort of the rising of the Zoom actually is going to make a big difference, right? Because that's making that infrastructure ubiquitous so that you can sort of make the cohort-based learning more reality. But then a few years ago, in the last few years, you almost kind of, the way I think about it, you refunded a company with AI being the main thesis at this point. Right. So you mentioned that for people in this audience, if you want to learn AI, don't go to whether Harvard, you know, sort of the certificate program, instead Maven is a better off, right? But how do you reconcile between people still feel like a Harvard certificate program has the reputation? Maven, I don't know, do you have certificates? Like how do I... you know, how do I, you know, if they don't come to this show, right, they wouldn't know you have a better curated teacher, better, you know, the real value. So how do you deliver that message to people out there?
Yeah, it's a great question. Number one question I've gotten in my career is how do you compete with colleges which provide credentials, right? Udemy, non-credential program competed with Coursera. For what it's worth, Udemy became a bigger revenue business so we won that. We won sort of, it was kind of like one and two, both companies are great. I really admire Coursera as well. Maven, similar thing. We're competing with Harvard Extension School. We're not competing with Harvard MBA yet. We will eventually. So here's a question for you, Howie. When you are hiring for your... how big is your team that reports to you?
Not very big, but I think I know where you're getting to because I saw your LinkedIn post about Warren Buffett's statement that when he hired people, he doesn't look at which school, you know, the person would have come from. Is that where you're coming from?
That's where I'm coming from. Do you look at what school your employees come from? Do you know the school of your, like, peers, like the people who you work with the most?
I think it's a complicated question. I think there's still signaling thing. I wouldn't say I don't look at it but I think you have to look at it in a balanced way, right? We have people from Yale, you know, clearly there's signal in that. So I would say we look at it but I think it really depends on... I was talking to a startup founder, right? He was hiring this chief revenue officer. I told him that when you hire chief revenue officer, never look at the schools because most of the very successful chief revenue officers would have come from no-name schools, you know, that's just a reality in the last 20 years. But when you talk about, you know, OpenAI engineers, you know, often time they come from very good schools, so.
Totally agree. So probably like when I was, you know, in my 20s, like many of you were probably working in that time frame, like in 2005 to 2015, you know, you knew every, let's say you knew 100% of the people you worked with's schools, especially if you interviewed them. Today it's maybe 70%, 50%, something like that, right? You still know people's undergraduate education in a lot of contexts but it's going down. I think in 10 years it'll go to like 50% solidly and eventually it'll go to 30 or 20%. So that trend is happening. But here's another question. How many of the people that you know that you work with, you know about the certificate programs that they took?
Right? So Harvard and Stanford is still probably where you send your kids to undergrad if they get in.
But after they're done with undergrad, do they just stop learning? Is it over?
Well, in today's world, if you stop learning at 21 years old, when you graduate from college, you are screwed. So, what Maven is saying, hey, yeah, you can continue to go to Harvard. I don't think it's going to last forever, but like for the next 10, 20 years, like maybe even my kids will, if they could get into Harvard or Stanford, I might send them there. But after that, I'm going to teach them that they need to keep learning because the market will change. If you start as an IC and you have to become a manager, where are you going to learn leadership, right? If you are, if AI is the future, where are you going to learn AI? You didn't learn that in school. That's already in the past. So, what Maven does is it says, look, you can get the credential for your main primary degree, that's four years, etc. For everything else, you need to keep learning throughout your life. You don't care about the credential that much. What you care about is learning the skill so that you can present it at your job so that when you're in an interview, the person knows you know the skill and we do that.
Okay. So, I think we're finally getting through the introduction part. So, thanks for giving us the journey you have gone through. Now let's talk about the main topic. I wanted to discuss AI, right? All the impact of AI, the impact to your product, everything. Before everything, you know, you have been a fresh user of AI native browser in the last one day, two days. What was your initial impression?
I think there's no question that we will have an AI native browser. I mean, as I started using Neo, I thought to myself, well, this is essentially just like a very basic version, sorry if I'm screwing this up, but like it's like I currently have maybe seven to eight desktop applications open at any given time on my computer, right? Remember the time, like literally like 5 years ago when it used to be just one, right? It used to just be Chrome. Like that was it. That's all I had. And in the last 5 or 10 years, it's changed. And now I have ChatGPT. I have a desktop app. I have Slack. I have a desktop app. I don't use the web browser, which I used to, you know, 5, 10 years ago. I have Notion, right? I have Chrome. This is me. I'm a startup guy, so I probably maybe a different stack from some of you, but like that's all the things I have. Like it's annoying. Like it sucks. Like I don't want to, when I'm asking ChatGPT a query, I don't want to have to split up from that in the browser experience. In fact, I think of these things as inherently inextricably linked. I often want to go to ChatGPT to find the right website, not to give me the answer directly, which is all it does today. And then when I'm browsing, I often want a Gen AI response to what I'm doing. So I think it makes total sense to combine these two things. I don't know yet what the exact example of how that's going to work is going to be because I've only used Neo for a couple of days. It's still new to me, but the concept and the business logic makes complete sense. And it's a multi-billion dollar opportunity, right? It's like it could be, you know, a browser, you know, or AI interface is huge. So, it's very exciting.
Cool. This is a completely unscripted. Now let's get on to the AI impact to your company. Not just, you know, the product within Gen, you know, we are very excited about doing AI. What is the impact of AI when ChatGPT came, right? What was in your mind? What does that change? Or what would that change to your own product?
Yeah. So, you know, I come at AI from a different perspective than a lot of people you'll hear talk for two reasons. One is I'm non-technical, right? I never got an engineering degree. I've never really written a line of code. Maybe a little bit of HTML, but that's as you know, kind of a joke. I am an economics major who has worked as a founder for 15 plus years. So I view myself as more like a right-brain humanities thinker than I am like a left-brain like highly first principles thinker about AI. I have always viewed AI as a very strong productivity companion for work and to automate away the easiest tasks we have. But I also think it's dramatically overrated as a job replacement. It's a job enhancer. So could I have one engineer that does three engineers' job? I think that's fair. That's probably close to accurate. Honestly, we were kind of doing that before AI. We were already thinking that way. We were thinking smaller teams, smarter people, more experienced. That's what we were doing already. And AI has just made that much easier to do. On the flip side, do I think that a product manager is just going to stop having engineers on their team if they want to build large-scale products? I mean, we are... I would argue we're like a decade plus away from that. I might be wrong. I don't know if I'm being honest. The reason why I think it's going to be very hard is because at least right now it is very difficult for AI to grok and fully understand extremely large complex systems and it's very hard for it to get rid of its priors. So you have to figure out, they have to figure out how to... and then there's a reason why that happens is because the AI is trained on a very, very large model, you know, in a cloud cluster, like, you know, somewhere offshore etc. and or somewhere in the United States and that is where it uses all its compute then it compresses down and then the AI that you're using, you know, when you use ChatGPT is literally like a very compressed version of that. So because of this process, and I've only done a little bit of research on this so I'm not an expert but my understanding is there are a number of technical hurdles to get from the process that currently delivers you this magical experience in a client interface via AI to the process that is required to completely take over someone's job. So at Maven I'm still hiring engineers. I'm looking for lots of engineers. So, if anyone in the audience, not the Neo, not the employees at Gen, but other people who are listening online want to come join us, like we're hiring, you know, we're hiring marketers, but my marketers are far more efficient and productive than they used to be. And so, I think that that's really the transition that's happened within the company.
This is actually one topic I wanted to discuss with you because in your LinkedIn post, some of the LinkedIn posts you mentioned that go-to-market engineer, right? Can you explain to the audience what do you mean by GTM engineer and what's your vision?
Yeah, I mean I think broadly, right, if you look at the functions in an organization, right, whether it's finance and operations, HR, engineering, product, design, you know, just name any major function, sorry if I missed yours, but like any of these functions, each one of them is going to have a like massive change in the way that you do your work, right? And think about it like this isn't that surprising, you know, some of us remember the days of using like client-side software as the primary way that you did anything and then we switched to cloud and like what were the changes? Like if you were a PM, your analytic suite completely changed. In fact, over my life as a CEO, which has only been 15 years, it's not that long, right? Over my life as 15 years of CEO, the analytic suite has changed every company and I even missed one or two analytic suites, right? So like, and you have to learn a new mode of operating. You're not just learning a new analytic suite with a different interface, it has different power, right? Essentially in 2010 when we were doing Udemy, all of our marketers needed to learn SQL. 100% of them had to learn SQL, right? That's what I taught my marketing team. In 2025, do they even need to learn like how to use Amplitude in a complex way? Yeah, today they do. In 2030, they won't. In 2030, they'll be asking natural language. It's a new skill. It's a hard new skill. So go-to-market engineer is one of many new skills that are essentially a redoing of an existing thing. So the correlative is RevOps, right? So you have your sales team. Again, I'm not a B2B guy, so we're talking an area I'm not super expert in, but you have your sales team and then you have your go-to-market team that is trying to get leads for the sales team and optimize the sales team when they are selling to their end consumer, right? To the prospects. That go-to-market team can be completely reinvented today because you now have AI that can automate a lot of the rote tasks that that team used to do. So the phrase, the phrase GTM engineer, GTM usually not engineer, engineer usually doesn't do go-to-market, but you specifically see that a go-to-market team will be more closer to like the way they do things on a day-to-day basis will be like engineer or that go-to-market team will have engineers powering them up to do things.
