SPEAKER_01: Hi everyone. Welcome to YC founder of Fivesites. My name is Anu Hariharan and I'm a managing director at YC where I work with our growth stage companies. I'm here today with Jaleh Rezahe, founder and CEO of Mutiny. Mutiny uses AI and data to convert website visitors into customers and was part of YC's summer 2018 batch. Welcome Jaleh. Thank you for having
SPEAKER_00: me Anu. How are you? Good. And I'm so excited to have you here because I've known your journey
SPEAKER_01: from the gusto days. But today we're going to talk about AI but we'll also touch a little bit on your early journey as well. Mutiny recently announced that they acquired Intellipse, an AI marketing platform. So many people listening today are founders and CEOs. So we're going to discuss two things. One, acquisitions as a growth strategy as well as how AI will impact the next era of growth. But first, as I said, let's start with some background. Mutiny was founded four years ago and the product has really evolved massively into becoming a platform for go-to-market teams. Mutiny uses AI to identify ideal customers for your company, determine sections of websites that will increase conversion and produce copy that converts visitors into customers. So Jaleh, I know a little bit of this background but how did you really come up with this idea? Can you share a short version of your founding story? Absolutely. So this idea for Mutiny was very much inspired by my own experience
SPEAKER_00: leading marketing at Gusto. And when I first took over the marketing team, we were initially struggling to scale our revenue and customer acquisition. So we were spending a bunch of money on marketing but it wasn't translating into revenue at the rate that we wanted. And CAC payback was well over 20 months, which as you know Anu is very bad for an SMB business. And so we did a bunch of discovery looking at the data, talking to other companies to understand and ultimately what I arrived at was the simplified view of growth, which is that there's really two levers to growth. There's demand and there's conversion. Demand is things like advertising, content, PR, all the stuff that we associate with marketing and conversion is this practice of creating a system to turn all those hand raisers into customers. And that's the piece that's massively underdeveloped for most companies, Gusto at the time included. And so I look to the best companies at this, which tended to be B2C companies like Airbnb, Uber, they had all solved this problem by essentially hiring hundreds of growth engineers and data scientists to go and optimize their web experience. And so we decided to do the same thing. We built out a dedicated growth engineering team and we invested in things like form optimization, website personalization, onboarding, all sorts of things to convert visitors into customers. And over my time at Gusto, we saw how impactful this was. Overall, our revenue grew by 100X during that period. CAC payback went down to less than 10 to 12 months. And so we were able to have this really rapid growth, but it was really difficult to build all of the stuff from scratch and hire all of those engineers. So my co-founder who was also from Gusto, we started to think about how can we turn what we were doing into a no-code platform that every company could use so we could essentially serve as the world's growth team and help all companies solve this conversion problem and be able to essentially grow a lot more quickly and efficiently. And so that's, and our day one was actually day one of YC. So you've been with us pretty much since the beginning of that journey. Yeah, I remember talking about this idea with you way back
SPEAKER_01: even before when you were applying to YC, but can you actually double click a little more? What exactly was difficult to build? And I do remember the 20 month payback was like, you know, as you said, we never had good startups to be 20 months. But for the audience here who are mostly founders, because they may not have, some of them may not have seen what the struggle is going to be. So can you elaborate what was hard, what were you trying to solve and how long it took? Yeah, totally. So basically what you are doing
SPEAKER_00: as a growth team that's trying to solve conversion is you're trying to build a system with inputs and outputs, right? So you have, the first step is connecting data. So understanding who is coming to your website, which part of your funnel do you have problems and therefore should prioritize. And so there's this whole data investment piece for you to be able to just aim your team and your engineering resources in the right direction to affect revenue. And once you know, like, let's say, you know, restaurants are not converting on the website, because restaurants have a different value proposition. And our website is optimized for startups, which is, you know, it was a very real thing that we face at gusto. Or you realize, hey, our signup form relative to benchmarks is underperforming. And we think that we should have a higher conversion rate and that we have opportunities to improve this signup form. You know, there's, there's hundreds of things that you know, that that that can be important. And you essentially look at data to to estimate what the revenue impact can be if you were to make improvements in that step, then what you do is brainstorm and ideate with the team. Okay, well, if we know restaurants are underperforming, what do we do differently for restaurants? Should we personalize our logos and and images? Is it really more of a call to action issue because they don't want to buy by going through the online experience, they want to talk to somebody and see a demo, right, so you start brainstorming based on what you understand about the different customers for what the solutions might be. And then you size all of these ideas the exact same way that you would build and prioritize if you were building a new feature, except that, you know, instead of things taking years to build, you know, these tend to be things that take, you know, four weeks to like nine to 10 weeks to build. And so so then we would we would prioritize based on the amount of time that it would take our confidence in in the impact that we were estimating for it. And then we would then say, Okay, out of 100 things that are listed in the spreadsheet, these are the one or two things that we're going to prioritize and build. And then you have engineers go and build those things in as lean of a way as they possibly can. And you have a measurement system that then shows you whether the the change that you made the thing that you build actually improve conversion or not. And then you learn and you optimize from there. So you have this system in this loop. And you can imagine that if instead of taking you months to, you know, come up with ideas and build them, if we could collapse that to a matter of, you know, minutes, or hours, it would have a massive impact on the team's ability to be able to launch new experiences and optimize their conversion. And so that's the that's basically what mutiny does. So instead of the company having to build all of this infrastructure, we, during our onboarding, you know, we have all of this pre integrated data sources, we have our own data sources, we have our own conversion tracking, etc. So during, you know, about an hour of onboarding, we plug into all of that infrastructure, we have analytics already set up and everything. And then once the once they're onboarded, then they can just, you know, go into mutiny, and then our AI starts recommending, what are the things that they should be doing to drive conversion, and it's all no code. So then they can just launch those things in a matter of minutes, and, and see the impact. So it basically collapses that very long process. That also took engineers, which obviously, they don't grow on trees. It's, you know, it's the hardest to build. It collapses that to a matter of minutes for someone who's not even technical.
SPEAKER_01: That's great. And so, roughly for the audience here, when should a startup think about using mutiny? Is it from day one? Or what's when is it right for them to sign up for mutiny?
SPEAKER_00: So on the, so we have essentially mutiny is built as a platform that we then activate different use cases on top of it, right. So think of it as if you're a growth team, you can basically build whatever you want. But it's, you know, it's very time consuming, but you can go from signup form to referral to onboarding to, you know, website headline personalization. And you essentially because you have to custom build everything, you can kind of bounce around the entire customer journey. But since we are building a platform for all these companies to use, we essentially open up canvases one at a time, right. So we started with website on page personalization. And then we added components and embeddables. So things that you can just, you know, surveys and, and little things that you can just insert into your website. Then we are adding A B testing. So you don't have to have a separate A B testing tool, you can just use mutiny for all of this. And we have and a one to one A B M landing pages. So these are all separate canvases that we that we release. And so it really depends on what you want to use the product for. And so for our inbound product where we help optimize the website, usually I recommend that the website has pre has product market fit. And at that point when they have, you know, 20,000 or so visitors, then that's a great time to to start doing a lot more optimization on top of it. But as we add in A B testing, essentially the size of that website requirement goes down. So I think much smaller startups can start using mutiny probably starting in a few months from now. And then our outbound product, which creates one to one pages for target accounts. I mean, that's something that we at mutiny we're using when we had five employees and were pre product market fit. So that one doesn't really have a website visitor limit. As long as you know who your target customers are that you want to go after you can you can start using that. But generally, you know, mutiny customers are small to mid market. So you know, folks like notion, Brex, Carta. And we're now starting to work with larger customers like Snowflake and Qualtrics.
SPEAKER_01: That's great. That's awesome. Well, one thing here for founders that say in the audience, I've noticed one of the you know, common I would not even say misconception or mistake. But you know, one of the things founders tend to overlook because you're so busy trying to get the product market fit, and you have all these customers joining you, that you put up a website and you look at it a year later, it probably doesn't hold through anymore. You'll have to make changes because you have different customers coming to you. And so that's something definitely to keep an eye on. And mutiny is really great in that sense. Now let's you know, you mentioned AI a couple of times, jelly. So I want to double click on this. Especially I think like, you know, I think I saw a meme this week, that is now focused on generative AI. And it's such a broad definition. And I've always seen these things coming, though, let's demystify it. So let's, you know, mutiny help, you know, what you said is really helps improve conversion, you're focused on both inbound outbound, and you're, you also have a, you know, AB testing product, but it transforms conversion from a niche AB testing tool to a platform that every go to market team can use. But what does AI even mean in this context? And how have you leveraged that over the last four years? Yeah. So I mean, every company has this inefficiency problem that we talked about,
SPEAKER_00: right? So on average, when you look at the marketing spend of a company, 19 out of every $20 is wasted, because the experience isn't optimized for that audience from that part of the funnel. And so the more that you can make that efficient, then the faster your company grows, and the better all of your metrics and all that good stuff. And so the first generation answer to this problem was essentially AB testing. But the problem with AB testing is that it's very tactical, and you are trying to optimize to just the lowest common denominator. And so it ends up being one person on a web team running experiments on websites should the button be green or red. It's not this big strategic enabler for the company. But when you give the entire go to market team x ray vision into every visitor, how they're converting, and then also pair that with the ability for them to change their website for all of those different segments, then you basically go from this niche task to every marketing team, every growth team is now making the website a core part of their their strategy. So the ABM team wants to create one to one pages for all of their target accounts, the paid team wants to match the ad copy to the landing page, you know, the lifecycle and growth team wants to upgrade freemium users to paid and so it becomes this now, you know, this huge part of the strategy to drive more revenue. Now, where AI comes into this is making all of this easy to do for all these different users. And so, you know, as soon as you connect data, you end up with 1000 questions, right? Which segments are underperforming. So for example, we connect and identify over 70% of the visitors, and there is hundreds of attributes around the visitor such as their size, their industry, persona, use case, right. And so once you once you shed light on that, there's all these questions around what which segments are underperforming, which segments should I prioritize? Who should I start with first? What parts of my web page should I change? You know, I don't have unlimited resources, like where should I focus in order to get the biggest bang for my buck? What copy is going to resonate for these different audiences? Maybe I have eight variations of what I want to do differently for startups versus enterprises, but which one is the right one. And so we essentially use AI to answer all of these questions. So from the who to the where to the what, we use a variety of different technologies to basically automate and make it easier for each of these things. And so we're very, you know, one thing that probably will emerge from this is we are not very focused on technology, we're focused on the end user, like what is the value that we can add for them? How can we make their lives easier? And then we pick the right technology, whether it's AI or otherwise, to essentially enable the user to be able to do that task a lot easier.
SPEAKER_01: I want to double emphasize what you said, like, it's so important to focus on the problem and not the technology. Because too often, I think people want to start an AI company, and then it's like, what exact problem is it solving? So you, you mentioned something very interesting. I do see this with a lot of vice startups, data is, you know, growing from day one in a startup, and the amount of data you can collect is, you know, incredible. But to some extent, the questions you ask, inform the answers. And sometimes, it's also like, are you asking the right questions, you touched on customer segments, which customer segments to focus on? Can you touch a little bit on how you approach it at mutiny? Like, you probably see a ton of data? When do you know what are the right questions to ask? And when do you say, these are not questions we need to optimize for now? Because I think that could be pretty damaging to startups, if very early on, they're focused on all the questions?
SPEAKER_00: Yeah, totally. So I mean, we, we approach it from a marketing and user lens, right? So then this goes back to like, how do you put together the right founding team? So Nikhil and I joke that we don't know how to solve any of the same problems. And, you know, I bring I used to be, you know, the user, right. And so I bring a lot of that to the table. And then he is, you know, an engineer's engineer, like, he loves the data, AI, like that whole side, right. And so we essentially bring these two things together. And so the way that we think about it, and a lot of this we validate in in user questions as well. So for example, in the early days, one of the most valuable things that we did was we followed the customers home, if you will, so we were essentially an extension of their growth team. And so we would follow them and we would say, Okay, let's let us come to your team meetings. And we want to be end to end a part of this problem that you're trying to solve with me at me. And the the thing that we saw kept coming up. And I remember like amplitude in particular was one of our early customers is they would come into mutiny and execute on personalization. But then it would take them a while to in between experiences. And so as we double clicked into that, and we're like, well, what are you exactly doing to do this? And they're like, well, we're trying to figure out how to prioritize between different audience segments. And so then, you know, we dove in with them. And we said, Great, like, can we can we take a look at this with you? And we saw that, you know, they were essentially trying to figure out which segments are larger than others, which segments have anomalous conversion behavior, what has been changed in behavior, maybe a segment recently got added to their website, or recently had a change in conversion, right? And so these were the types of questions that they were trying to answer in order to ultimately prioritize, okay, like, where do I actually start? And which segments do I prioritize? And so when we saw that, you know, that gives us a very clear answer as to Okay, these are the types of questions that we should be automatically answering for our customers. And so we, we did a few different things in that particular example, one was we built in the flow, the data, we started visualizing the data in a way that they could immediately see some of those things for themselves. So we gave them better hooks to to be able to answer those questions. But then that also informed, you know, our ability to automatically identify answers to those questions for them, and have those types of recommendations go into our recommendation engine.
SPEAKER_01: So you mentioned, that's a really helpful example of a real customer amplitude, right? So you mentioned data and the use of data. And I know that, you know, you also touched on this earlier that your recommendations or the use of AI to answer some of these questions depends on the data. Can you talk about since you started mutiny? What are some of the advances in AI over the last four years that you've sort of leveraged in mutiny itself? Yeah, absolutely. So, I mean, I think the, the data set is, is really, really important.
SPEAKER_00: And so, you know, I think we started with, like, what is the problem that we're trying to solve with AI, and then make sure that we solve that problem at almost like every stage of the company with or without a very large data set or a super sophisticated algorithm from day one. And so for us, the problem was, we want to solve the art and science of conversion, right? Which really comes down to like, we want to automate and we want to guide. And so the initial value of mutiny was the no code piece, right? We want it to, without engineers, marketers could launch, you know, changes to their website and learn from their own data. And that in and of itself, without any AI was just really valuable to be able to give them that ability. And then the AI piece essentially just makes it easier and easier on top of that, right? So the V1 of our AI, like in the very early days where we had no data, it was, I was the AI, right? So I was like, okay, great. Like we will help you come up with, we will, we will help guide you. We will help you come up with ideas for tests because we were just trying to solve that cold start problem and get people going so that they could, they could start, you know, using the product so that we could start generating a data set. And then as those early users started to, you know, launch experiences, our data started to grow. And so we knew we wanted to ultimately use AI across customers in an anonymized way to, to, to bring in things like benchmarks, recommendations, right copy, et cetera. And so we, our data layer was standardized. So our definition of healthcare across Notion and Amplitude and Snowflake is the same. And so with that standardized data, we had that standardized data, but there wasn't enough volume to, to be able to automatically, to build models and automatically say like, okay, these are the parts of the website that have to change. Here's, you know, what you can learn from other customers. And so what we did instead in those early days is we launched what we called community playbooks, where a user, as soon as they would have a winning experience, we would in the app, ask them, Hey, can you share this playbook? And they would say, you know, yes, pretty much we had a hundred percent acceptance rate. They would, they would say yes, and they would share it. And we would extract that playbook, which was a screenshot of what they had done on their website before and after with mutiny, as well as metadata around who was the audience and what was the conversion rate. And we would then recommend that playbook to other customers that had similar audiences on their website. So this was like, you know, V1 and it essentially was going back to solving that user problem of how do we inspire them? How do we guide them? How do we, you know, how do we make this faster for them? And so just seeing what other companies were doing you know, maybe we didn't have enough data to automate every single step, but we could at least put this in front of them. And then it allowed them to be able to come up with more experiences. And so that again, helped us get more people using mutiny faster in the early days, which, you know, kept building our data set. And then at some point, you know, that data set started to grow. So we were able to do much more interesting things. So being able to identify anomalistic behavior and clusters and audiences on the customer's website, so that we could, we could say, hey, these are the audiences that you should really focus on. And then if we had a matching playbook, then we would add that to it. Then we started to have enough data to be able to really bring in benchmark information across customers, and be able to predict performance for them, right? So to be able to say, hey, you know, for this audience, we estimate this is the revenue impact that you can have if you make these types of changes. And then as our data set has grown, we now are able to build models where we can tell the customer where on their website they should focus on making changes, whether it's particular pages or even parts of the page. So we can, we essentially can identify, hey, like this section of the page, if you personalize this, you are 32% more likely to be able to increase your conversion rate. And so we have these estimates that are coming from the aggregate data, and then our model is able to identify these different parts of the page. So we've been able to, you know, go from just the audience to like, actually now get very concrete on the page, what they could be doing to make those changes. We also starting, starting about a year ago, started writing personalized coffee. And so this was something that, you know, we leveraged GPT-3 for this. And I'm happy to talk more about that, if that particular technology is of interest, but essentially we're able to, you know, we have a data set that tells us who saw what, you know, what type of content and did that lead to an increase or decrease in conversion. And so we take that data, which is our own proprietary data, and we feed that to GPT-3 to then produce personalized coffee for different parts of the page for the customer. So it's been very much a progression of, you know, starting out with the first, you know, product being that no code piece, you know, added enough value for us to be able to get in the game and start partnering with our customers. And then with AI, we've basically just, you know, chipped away at this ease of use problem around the, what we call the art of conversion, which is like all of the strategy and where you should be focused on. We've essentially just chipped away and made little bits of that easier and easier and easier over time. And so it wasn't sort of this like one time, you know, thing. And it's going to, we're going to continue to, to, I think, operate in that way as more, you know, AI technologies become available that we can take advantage of. We'll basically just incorporate that into solving that customer problem ultimately.
SPEAKER_01: What I love about what you shared, Jalil, is I remember in summer 2018, your one line for YC was personalized website for B2B marketers. And, you know, your evolution, it started with focusing on customer problem. You were the version one AI trying to help customers answer the problems. You collected more data step three, then you used it for benchmarking. Then fourth, you gave better recommendations. And now with GP3, you're starting to realize that vision that you laid out, you know, almost four years ago. So it's really interesting to see that journey. And I think for the founders in the audience, the real takeaway here is a lot of startups usually say, oh, I don't have data and I don't know how to leverage, but I think Jalil's example really highlights. Start with a customer problem and do things that don't scale. If initially you're the human trying to figure out what's the recommendation, how to really get to the answer, you start with that and data grows. And then over time, as data grows, you can use models and inputs and improve your recommendation engines, you know, to help service your broader customer base better. So Jalil, did you have moments? I mean, it sounds like, you know, it's only been four years, but there are a lot of changes in just mutiny as well as in AI in general. Are there times you felt you could have done more with the advances being made in AI? Are there potential, like different solutions you would have considered?
SPEAKER_00: Yeah, I mean, I think GPT-3 is probably, it's amazing how much is happening in AI. And there has been a lot of positive wins that we didn't expect when we started building mutiny. And so GPT-3 is probably a really good example of one of those, you know, we had on our roadmap to write copy, but we just didn't think that we would get there for a while because we needed a much larger data set in order to be able to like truly generate tax. And so, you know, and for those of you that don't know, GPT-3 is basically a large language model that's made by OpenAI. And it's trained on natural language data from basically the entire internet. And in layman terms, like it means that it writes really well. So you can communicate to it via text, and then it can produce text for you. But as the deficiency is that it doesn't know what works for whom. And so we saw this really great opportunity to leverage that technology, but then marry it with the preparatory data set that we had. And so, you know, a lot of companies, they basically just build an interface on top of GPT-3. And that, you know, that can be relatively commoditized over time, because that's something that everybody has access to. And so what we do that I think is pretty unique is we essentially give GPT-3 data in a structured way about the different audiences that we wanted to write data that we wanted to write text for. So we would say, hey, like, this is what's worked for healthcare companies across our customers, and we feed it that data. And as a result, it's able to then and then we ask it to, you know, to write text. And so as a result, it's able to generate highly personalized content. So I'd say there's these moments along the way have been really great as a startup, you know, you have wins, headwinds and tailwinds. And I think on the AI side, we've had a lot of we've had a lot of like positive momentum that where we have seen something and we've been able to just leverage that. You know, there's other things around just like natural language processing, there's just a lot of back end AI models that you can you can start to use and move a lot faster. So I think all of that has been have created like, you know, higher momentum and have been positive moments for us.
SPEAKER_01: What are some things that you think are still underdeveloped? You kind of said or another way to ask that question is, you said GPT-3 was an inflection point, right? You kind of mentioned it as something revolutionary that happened because of which, you know, you're in this next frontier of, you know, using AI for personalized recommendations, but maybe you can look at it as things underdeveloped or what is the next inflection point you think?
SPEAKER_00: I mean, we're pretty excited about Dolly, because in marketing visuals is also pretty big. And so I think there's a lot of opportunity there on just general visuals, images, video, a lot of our customers. It is interesting, though, to think about how does that translate for for marketing and for brands that have really strict brand guidelines on, you know, what things have to look like. But in a lot of our like early experimentation with with Dolly, there's, there's a lot of promise on being able to feed it data in a way that you're able to generate visuals that can potentially work and be on brand for a particular website or for a particular usage. So I'm really excited about like what we can do on the visual side in the very near future. That's great. Can you maybe for the founder audience explain what Dolly is or how you've used
SPEAKER_01: to that mutiny so that they have a better idea of the use case? So we haven't so we haven't productized anything around Dolly yet. But that is something that
SPEAKER_00: our engineers are working on. But essentially, Dolly is very similar to it's another open AI model. So it's very similar to GPT-3. And it can essentially generate images from input text. So the underlying model is, I'd say it's similar to GPT-3, but it was trained by matching text to images instead of text to text like in GPT-3. And so you're able to basically speak to it and give a text, describe what you want to see from it. And then it is then able to generate images that matches your description. So you can say, give me a raccoon that is smoking a cigar on the in Paris. I don't know why that popped into my head, but our team does a lot of weird stuff with raccoons because that's obviously our mark. So they like to have fun with that and it will generate that for you. So you can give it like really good input and then it'll create images for you.
SPEAKER_01: Do you think we ever hit a world where all images are created by Dolly?
SPEAKER_00: It's hard for me to predict the future. If I was good at that, I would be a multi-billionaire today.
SPEAKER_01: True that. But it's amazing to see the evolution of technology and how you use these tools today and you're like, wow, geez, I don't understand how I have been doing this work for the last 10 years. Yeah, that's really amazing. And I think what's been really cool is, so I do, we have an advisor
SPEAKER_00: who is from Square that we worked with and he is in comms, he's not a very technical person. And he was telling me that he plays with, like he has this game that he plays with his kids with Dolly. I don't know specifically which interface he uses to it, but essentially him and his kids come up with cool stuff that they can tell to Dolly and then it generates images for them. So it's amazing to see how much AI has become accessible to everybody. I think it's at a major breakthrough where before it was sort of behind the scenes, this backend thing, and now it's in our lives and people are actually interacting with it. So that's been really, really incredible to see.
SPEAKER_01: You mentioned a very important point earlier. It's exciting to see GPT-3 in Delhi, but not all ideas are that core. You said for example, meaning some are just built on top of GPT-3 without much depth, whereas in your case, you actually give the proprietary data sets and it's very targeted to a specific use case. So a founder that's dabbling in AI and wants to start a company, what would be some caution or cautionary advice, if you will, you would give them on how they should think about using these technologies?
SPEAKER_01: I think you probably hear this a lot from VCs, which I think is probably pretty good advice.
SPEAKER_00: You have to think about what is the ultimate long-term vision of the product and what is the, I know the term moat is really overused, but essentially, what is the long-term defensibility of the business? So that is something that I think is valuable to think about in the early days. And you should have a good answer. If that's not at all part of the strategy, I definitely recommend thinking through why if you're successful, you're not going to immediately get commoditized. Because the rate at which technology moves right now, it's just so fast. It's crazy to see how quickly once you discover something great and you put it out there and everyone talks about it, how quickly everybody else starts to copy you. And so if you don't have things that are hard to copy or that will eventually become hard to copy, I think it'll just make it harder for you to stay ahead of the game. And so that is probably my recommendation around that.
SPEAKER_01: And so that's very good advice. But one of the things we also tell in VCs is try to launch fast. You've definitely heard us say this in the 12 weeks. And so how should one, what advice would you have for founders in terms of leveraging open AI, leveraging GPT-3, and yet focusing on the moat question which you laid out, which is very important. Yeah, I think your vision and long term view, it's very separate from your day to day execution.
SPEAKER_00: So for us at Mutiny, we had this broad vision of we want to build this platform that essentially replaces the marketing system of record, many disparate tools, as well as a whole bunch of custom growth engineering and data science. And so really big vision, really big market. When I shared some of this with Paul Graham at YC, the way he described it was he's like, oh, you're the other half of Google. You know, Google is like helping everybody, bringing customers to the doorstep of your website. But then there's nothing that helps convert them. And so we, you know, that was the vision and how we thought about where we want to get to. We wrote a vision doc that essentially lays out, you know, in about eight pages, what the opportunity is and what we're trying to create over the course of a decade with Mutiny. But then when it comes to execution, you have to have a very clear, OK, what is step one? Not even act one, act two, act three, but what is step one of act one? What is the first thing that you're going to launch? So when we did all of that and then put that paper away and then we started, we incorporated, started YC and in two and a half weeks released our first product. And in I think we had our first paying customer in less than four weeks. Right. And we had a very simple UI that we love. We have screenshots of and we love sharing that every once in a while to embarrass my co-founder. But, you know, it's it wasn't necessarily the most beautiful thing, but it got the job done. And it allowed us to start to work with customers, understand the workflow and then keep going from there. And so I think I sort of see the long term vision as it brings a lot of clarity around where you're trying to go. It helps you bring other people along the journey with you, whether it's investors or employees, you know, folks that are going to help you build everything. But then your day to day execution is very fast and it's very focused. And, you know, you're not thinking about, you know, the 10 year thing when you're building the first version. And then and then, you know, you kind of build and as you understand what the customer workflow is and you feel really confident, then you start you can start to like refactor. And, you know, this is something that we're thinking a lot about right now as we are moving into more of that platform phase is, you know, how do we as we expand and build new use cases? How do we still give those teams the autonomy to build new products in the way that we built them when we were really young and let them be able to leverage the platform, but essentially not be overly tied to to it so that they can figure out exactly what's the use case, get to product market fit around that. And then and then once we have that, then we can kind of continually refactor the the primitives so that they're more easily accessible for future use cases. So I think this problem doesn't go away. You know, this how do you how do you enable your long term vision, but still move fast? It's just something that as a founder, you have to deal with at every stage of your journey. And it just looks different at different stages. But it's the same problem.
SPEAKER_01: Yeah, and actually, this is going to help us transition to the second topic, but it's almost like the build versus buy question. And, you know, you have to constantly ask that. And that's when the long term vision really comes in handy. And you have to answer yourself. Answer for yourself. What is your moat? And what's you know how and that moat in the lens of how how much improvement is it from a customer standpoint for the problem that they're trying to solve? Right. And so the build versus buy. I think Jeff Lawson from Twilio has a great framework for this, which is, you know, if you ask developers, they'll say build everything. But you mentioned this at the start of this Twitter space that engineers don't grow on trees. So you can't put resources to build everything. So you really ask yourself how core it is to what you're doing. The problem you're solving for the customer versus when do you use off the shelf solutions? So let me switch gears a little bit. In the last 15 minutes we have, you decided to grow your team by acquiring Intelips. And now mute me as one of the larger engineering teams with production experience in modern marketing. Why did you decide to pursue this acquisition? Like, what was the thought process? Yeah. So, I mean, right now we are.
SPEAKER_00: In a recession and certainly expect more of that in the next one to two years. And so I think we're just in a space where companies need to convert every dollar. And we have built a product that makes marketing dollars more efficient. So we've literally built a product that everyone in the market now cares about. So as far as I'm concerned, like this is our chance to really lean in.
SPEAKER_00: And I think as startups, all of us are building things that are going to change the way an entire industry operates. We're all here to create a new future. And so to do that, we have to look for inflection points where something happens in the market, where the old way is not going to be as good. And now a much larger portion of the market is open to your new and better way. Right. And so we see it as this is really that for mute me specifically. And so we want to lean in and we want to take bets and we want to move faster. When I was at VMware, I was there during the last recession. And it was kind of amazing the way I saw that company double down on these types of market inflection points. So we sold virtualization and that helped companies cut down on server costs and energy bills. And so when we were going through the OA recession, we really went nuts on like, hey, like help. We'll help you consolidate your servers. And it allowed us to grow really quickly, even as a public company. You know, my first year at VMware, we doubled revenue. So it was you know, I think it's really important to take advantage of these these tailwinds in the market. And so for us, you know, we want to make a bigger bet and and really lean into creating our category. And A.I. is a really big part of making me easy to use and make it more accessible to more people. So we wanted to basically accelerate the roadmap. And so that was the reason behind, you know, why it makes sense for us to to just move faster than, you know, hiring folks like one at a time.
SPEAKER_01: And how did you know you wanted to work with this team? Because I'm sure you had you know, you could have explored many teams. So how did you come up with that decision? Yeah.
SPEAKER_00: So I mean, we were interested in the Intellips team and the skills that their team had developed while building the company. You know, their CTO, their senior engineers had really good experience with marketing A.I. and in particular things like GPT-3 and prompt engineer and things that are just newer. And so we saw a great opportunity to to really expand our our capacity for being able to build this sort of stuff more quickly. And I think in acquisitions, like you always have tradeoffs, right? The acquisition is faster than hiring or building or training. But then you risk not being able to integrate what you have acquired into the rest of the company. In our case, we wanted this really talented team, but we didn't want to create a microculture within me. Right. So those were the the tradeoffs that we were thinking through as we were evaluating this. And so really, for us, what it came down to is, do we believe we can we can bring in this really incredible team that we want to work with, but do it in a way that that allows us to enhance our culture as opposed to divide us. Right. And so it was all about do we have the same culture? Do we have shared values to be able to fully absorb this team? And for me, like, you know, I'm a big believer in the personality and the values of the founder does really spread in the organization and it becomes their culture. And so I knew the founder really trusted him and knew that he had not only built a really smart team, but that they were going to be good humans. So that was sort of what opened the door and made us interested. And then as we started to talk more, it became really obvious that we were cut from the same cloth. Right. So we were both very transparent about our intentions. We were very open about our needs, their needs, you know, the numbers, what investors wanted, what different people on the team wanted. And so we were able to really put together a deal that worked for both sides. And we did end up spending a lot of time with every individual that joined to make sure that they were going to be a perfect fit for the culture. And then once we acquired the team, we we purposely did not put them all on an island. So they were based on their abilities. They were distributed within within the company. And so now it's you know, it's been a huge culture enhancer for us because, you know, we have folks that are very much aligned on values, on the vision of the company, the product that we're trying to build. And, you know, there is now this group of folks that do have the shared history. And I think that makes us stronger. But we're ultimately one mutiny and one company all building towards the exact same long term vision.
SPEAKER_01: And how long did you take to build this trust and assess the culture? Because you actually what you articulated sounds like you actually spent significant amount of time with the team and you knew even before the acquisition the odds of it working out. So for the founders here, how did you think about how much time you spent to get to know them really well?
SPEAKER_00: So we spent I think from beginning to end, might have been six weeks for the whole thing. But we we spent the same amount of time with every individual as if they were as if we were hiring them in onto the team. So probably six hours per individual and a lot more time with the key players.
SPEAKER_01: Got it. Yeah. Super helpful and very important. I think that's something, you know, people can't usually calibrate. It's like some if you need to spend time as a three to six months or is it and it depends on the size of the acquisition. But I think the one takeaway here, which you mentioned, I think, which is profoundly helpful for people and I would highly recommend it is think as though you were hiring them. Right. Because that's the bar if they need. If you have an incredibly high thought for hiring and for advancing your culture and your values, then that's the lens you need to bring when you're acquiring companies. Do you expect to acquire more companies? Jelie, how should founders and CEOs determine whether it's the right strategy for their company?
SPEAKER_00: Yeah, I think acquisitions can definitely be tricky. And so you want to be really clear about what your goals are and why an acquisition is the best way to achieve it. When I was at VMware, we made a lot of acquisitions and I would say most of them failed because we were overreaching and hoping that, you know, some vision that we put on a slide deck would miraculously appear. I think like at some point, I remember we acquired a company that was kind of similar to, you know, like an old version of Slack and we were an infrastructure company. And so, you know, it just didn't fit with the rest. And the successful acquisitions that we made were the ones where they were really aligned with what we were trying to do as a company. And they just fit with the rest of the business. Like it's as if like we would have made them, you know, we would have done that ourselves. And they were just able to accelerate the timing for us and the culture and the individuals and the way they approach their work was very similar to the culture of the company. And so, you know, and then when I was at Gusto, I was there until about, I think, 500 employees and we were not very acquisitive and, you know, probably there was an opportunity to do more of that. And so at Mutiny, we are approaching it as just being really clear about what our goals are and then deciding whether an acquisition makes sense. So, for example, in our case, we do still have an acquisition use case and we're very interested in bringing on smaller teams where the founder wants to stay and help build a product area inside of Mutiny. So, you know, as I mentioned, our vision is to build a conversion platform that serves multiple different use cases, website personalization being just one of them. And so as we, you know, as we expand, we want ex-founders who are very entrepreneurial, who know how to move really quickly and be very user focused. We want them to join our product team and to build for those customers. And so it's a great opportunity for a founder who has all of those skills, but maybe is tired of doing all of the other things like fundraising and, you know, and all that stuff and wants to join still a nimble and fast startup, but have a little bit more, you know, a little bit more focus on just building the product and then growing that business area within Mutiny. So that's the particular acquisition that, you know, that we're open to and we're very interested in if the culture fit and the alignment around the product is there.
SPEAKER_01: It's funny you said that I haven't met very many founders who like fundraising, so that's a good point.
SPEAKER_01: That is true. It comes in cycles. One last question since we have two minutes and I thought it would be a good question to wrap up given the topic. What are your thoughts on how AI will impact the next 10 years?
SPEAKER_00: I mean, I think, you know, we talked a little bit about this of AI used to be really back end. Right. And so from the perspective of founders, I think it's really exciting because before, if you wanted to leverage AI, you base it had to be the end all be all like your whole thing had to be like building this machine learning team to, you know, to build these models, etc. It was really hard to build something bigger and to go all the way to the user. Whereas now I think there's enough productization of some of these back end AI technologies and models that as a founder, you can basically build the product that you want in the area that you're really passionate about and tap into AI as a tool in your tool belt to really accelerate the thing that you want to build and the value that you want to give to customers. And so I think that's really exciting from the founder perspective and from the user and growth perspective. What's really exciting is that I think AI allows us to automate a lot of crap that nobody wants to do. And so when I think about the world of growth, you know, there's all of this optimization that somebody has to do. Like the reason people aren't driving conversion, the reason there is all of this waste is that somebody has to sit there and create thousands of permutations and optimize those things. It's just at some point it becomes impossible and really, really tedious. And a good example of this, if I dare use an analogy from Facebook, is when Facebook ads first came out, they weren't really effective and only gamers could use them to have good unit economics and actually grow their business. But then when lookalike audiences that was all powered by AI came out, then everyone was suddenly able to use Facebook and be successful with it. Right. And it just kind of grew from there. And so I think that in particular in growth, AI is going to allow a whole group of people that are not technical, but understand what they're trying to do to be able to now have access to things that previously only very technical folks could do. And I think that's really exciting. And I think there's a lot of opportunity for us as founders to think about what parts of the world and what problems exist where there's still a lot of manual and technical steps and how can AI help automate and streamline that for end users.
SPEAKER_01: Well said. Well, thank you so much, Jaleh, for taking the time to share your own journey with Mutiny, but also the recent advances in AI. I'm sure the founders really appreciated it. Thank you again.
SPEAKER_00: Absolutely. Thanks for having me. Anything for YC.
