January 18, 2023

How Jasper found product-market fit: pivoting to AI-native SaaS

Sandhya Hegde
No items found.
How Jasper found product-market fit: pivoting to AI-native SaaSHow Jasper found product-market fit: pivoting to AI-native SaaS
All posts
Editor's note: 

Startup Field Guide Podcast episode 14

Our guest is the CEO and Co-Founder of Jasper AI, Dave Rogenmoser. Jasper AI is an AI-native content marketing platform used by over 80,000 marketers already. Their mission is to help anybody with a great idea share it with a few clicks. And they're also one of the first few companies that has built an incredibly compelling product in the hot new space of generative AI, raising a $125 million round early in 2022.


  • Jasper AI might seem like an overnight success story. But in reality, it took Dave and his co-founders eight years and three pivots to hit their stride and find true product-market fit as a software business.

  • Cofounder Dave Rogenmoser attributes much of Jasper’s success into deeply understanding their customers. He focused on these questions:

- What's their motivation? 

- What are their pain points? 

- What are their alternatives? 

- What are they thinking about that causes them to call you?

- Why do they cancel? 

- What are they really trying to get at here? And at a deep level?

  • Rogenmoser thinks founders often get stuck thinking that product-market fit is around the corner. But in reality, they’re not on the right track. “Finding product-market fit is usually very quick. And it may not be the fullness of it and it may not be all that you want, but usually there's signs of real life very, very quickly.”

Sandhya Hegde: Let's go back to 2014. I love that you are an 8-year-old overnight success. And so you, your co-founders, JP and Chris, decided you wanted to build a company together. That was the year that you all made the leap, so to speak. I would love to understand between then and 2020, when you've actually first started what has now become Jasper AI, what was it like? What were some of the big milestones and turning points in that six-year wilderness?

Dave Rogenmoser: When we started in 2014, we just wanted to build a business together, didn't really have any skills, didn't really know anything. We're just like, “we don't want to go get a job, so let's go figure it out.” And we tried to start this little software company that ended up failing. It was a bad idea. We didn't know anything, and it was just too capital-intensive, too labor-intensive, too long of a path to cash. And they had wives and kids and we needed to get to some cash really quickly. And so we quickly dumped that, started a marketing agency where we got clients and we would help them do Facebook ads or SEO or build landing pages. They're usually local small businesses — brick and mortar. And at the time, we didn't even know how to do the services that we were offering.

And so I would find people who could do the services and contract out to them. I was basically the salesperson and the project manager, and would sell the deal for $1,500 a month to run your Facebook ads. I would pay the contractor 50% of that. So he's going to make 750 bucks a month and then everybody's happy. That was really our first semi-successful business. And that was really hard — we didn't know the skills, we didn't know the customer, I didn't know the terminology. A lot of it was fake it ‘til you make it and saying yes to a lot of things and knowing that I'm going to figure it out as soon as I hang up that call. That was a really intense learning time.

But I really got to know marketing. That was my crash course of jump out of the airplane and build the parachute on the way down. I just learned so much about marketing because I had to, and I was doing it for all these different industries. One day I'm building a marketing campaign for an air conditioning installation company, and then that afternoon I built it to help sell airplanes for this local flight school. And then I go and I sell Yerba Mate cans through this e-commerce funnel that we were building. I just learned a ton. That's also a bad way to build an agency. We ended up being stretched way too thin because we had all these different clients that were angry at us all the time. But it was such a great way to learn broadly about a bunch of different services and industries. So we ended up deciding to shut that down after probably a year or so.

Sandhya Hegde: What would you say were your top digital marketing lessons having kind of learned how to fly that plane while building it? What were some of the lessons that you wish all marketers really had in mind?

Dave Rogenmoser: One, I was very dogmatic when I started. I would speak so confidently to clients and I would just say, "Well, you're going to need this three-page funnel and you're going to want to put the button here and you're going to want to have this." And then I've realized that it's really different for every industry and that the tactics don't just translate. But what I really grew to realize was that a deep understanding of the customer was key. And this is what was so hard about it because I was in so many industries or I just really had no idea what the customer was thinking. What's their motivation? What are their pain points? What are their alternatives? What are they thinking about that causes them to call you? Why do they cancel? What are they really trying to get at here? And at a deep level?

So somebody that's looking for airplane flying lessons, what's really the motivation? Are they rich people that just want to take the plane out for a bit? Are they trying to start a new career as an airline pilot? Are they kids? It's like there's all sorts of different angles here, and that really impacts the ad angles, it impacts the messaging, it impacts all of that. And it was just too hard to come into all these different things and learn that. And so you're left with having to rely on some of the tactics and not able to go deep.

And so I really grew in appreciation and the love I think has carried us through and is a big part of what we do today. Going so deep with the customer that you know them or can articulate things about them better than maybe they even could themselves. I found when you can get to that level and you can tell somebody something that they believe deeply, it builds an amazing level of trust with your customers, your prospects, and it allows you to just build really great marketing, but I'd say really great product, really great service, really great business. That is an art form that I think too many people skip over.

Sandhya Hegde: You focus more on the best practices and the latest tactics rather than just, you need to know what toothpaste your customer uses to brush their teeth in the morning, right? I think that makes a lot of sense. It's also the easier thing to default to feel like, yes, I have a strategy. You default to the science rather than the art, that definitely tracks. So you spent a year, you said, on the marketing agency?

Dave Rogenmoser: Yeah, we spent a year. I think we built it to about $25,000 a month. And we were just miserable. We just said, we can't scale this any further. It's just too much. We're kind of at this dead end. And so we decided to pivot hard. One day we just said, "Hey, we don't want to do this anymore." We were invited to speak at this physical therapy conference. The day before, we were sitting in the hot tub at this Airbnb and we were just like, "Guys, do you even want to do this anymore?" And it was a resounding no. And we're like, "All right, that's it. We're done with this. We're going to go to the conference, we're going to speak, we're going to tell them we're not taking on clients anymore. We're going to launch a course on how to build a marketing agency up to $25,000 a month."

And so we launched this course called 6K Success. We basically productized everything that we had done successfully. And again, we'd done a lot of stuff, I'd say unsuccessfully. But one thing we were really great at was landing new clients and getting new campaigns out the door. And then we were finally doing the marketing for ourselves. We're building these really great marketing funnels and we are running our own ads at a much bigger scale than we were before. And that was an even deeper crash course. I'm like, “OK, here's what really good marketing takes, and then here's what good community takes.” So that was the first time we kind of ran our own online community. We had all these people buying our course and teaching them how to do this stuff.

And those were skills that again, we now had in the toolbelt to carry on to a lot of what makes Jasper successful here. But it was us, I think being willing to pivot, but also keeping a thread of consistency there. It wasn't us saying, ałl right, we don't want to run a marketing agency, so let's go start a medical health company. It was like, all right, what's a better version of what we're doing here and how do we level up, take the next step up the ladder, where we can still take a lot of skills into this next business but not keep doing the same thing and banging our heads against the wall. We did that for a couple years. That was really great. Learned so much about community, learned so much about branding, learned so much about great marketing, and it's really hard to sell courses profitably.

You've got to be perfect and your funnels and your follow-up emails and your positioning and your ads, it's got to be so good to eke out a profit that by the time we kind of got to the end of that, we were just thinking, “this is a lot of work. We keep getting copied instantly. There's no kind of differentiation. Our customers just steal our course and sell it for themselves. So we need something that's harder to do.” And JP, one of my co-founders, is a developer. We didn't really know how good of a developer he was, but turns out he's a really great developer. We had this idea to go build what turned out to be Proof, which was this little pop-up that would pop up on your site, show who bought the product and when they bought it. So Dave just bought this product eight minutes ago and we built this little MVP for it, put it on our funnels for our own course, and we saw conversion rate go up.

I think the first test, it went up like 48%. We were like, "Holy cow, this is great. We could actually sell way more of our courses." But we also thought, well, maybe this is a product that we could take into a real product and go build a company around this and kind of get into software. And so we had a few friends in marketing who were also running funnels and I gave it to them. We installed it for them, just manually. There was no app, it was just all some raw code and just said, "Hey, let us know if this works or not. If you increase conversion rate, let us know. If it doesn't work, totally fine." And I think we gave it to four or five people. They came back and said, "This worked great. Can I get this on the rest of my funnels and stuff that I'm running?"

And that was awesome. So that was the day we decided, all right, everyone's saying it works great. We think there's a good business here — let's pivot. We basically shut down the other business again and pivoted into this new version where again, it was the next level up. It wasn't a total pivot into a new industry and into all new things. There were some stuff we were learning that were brand new, and there were some things that we were kind of keeping consistent there. And I think that allowed us to move really quickly. But we ran this webinar before we built the product at all.

So again, it was still this raw code. There's no app, there's no way to log in. We hosted a webinar and a couple hundred people show up to it. I'm pitching the whole thing, here's what it's going to do and here's all the magic. It was a really great pitch. And then I said, "Hey, but it's not ready yet. But if you pay annually right now, if you pay $1,000 for the year, we'll get you first access whenever we do release it." And everyone's pretty mad in the chat and they're like, "I thought this was ready now, we want it." But then we had 80 people pay $1,000 on that webinar and that was the most money we'd ever made. It was really validating that now we had people that said, "Hey, not only am I telling you I want this, here's 1,000 of my dollars to hold onto."

Sandhya Hegde: You got $80,000 in one day.

Dave Rogenmoser: $80,000. I still got a video of me scrolling through the Stripe notifications on my phone just freaking out. I'm like, this a thousand dollars, a thousand dollars, a thousand dollars. It was really validating before we had to go build this thing, people really wanted it. And so we ended up building that over the next month and getting that out to people's hands. The market was really excited about it. This is a brand-new thing, it really worked. We were the right people to take this thing to market.

And that was when we said, "Hey, this is growing fast. We don't know how to grow a software company, we need to figure out how to do that from the best in the world." And for us, I looked around and that was Y Combinator and I never kind of dreamed I'd be able to get into something like Y Combinator. I didn't have any cool pedigree or cool companies I worked for or a cool university or anything like that. It's like we just knew how to grow software fast, at least fast out of the gate. And so that was when we got into YCombinator and kind of started, I guess the next level of thinking about how to scale big software companies.

Sandhya Hegde: I have two follow-up questions. One, as co-founders, you have been on this journey already trying three different businesses. How were you thinking about the complementary skill sets — who focuses on what by the time you're scaling Proof? And second, when did you start getting the first inklings of headwinds in the Proof business? What was the kind of first sign of, “OK, not everything is going according to plan?”

Dave Rogenmoser: We didn't really know everyone's skill set when we started. It wasn't like the three of us just worked on paper and it made total sense. Mostly, we were just friends and we all liked entrepreneurship and we wanted to do something together. It turned out to be perfect and it really fell into the slot where it made sense for me to lead the company, for Chris to really be more of, once we signed customers up, he was really great at the support. He was really great at building these customer experiences that made everybody so excited to have bought. And then JP was more of the product person who would help with all the technical stuff. But then eventually he's the CTO of Proof and Jasper. So the complementary skill sets definitely fell into place fairly naturally. But there were still a lot of conversation in early days around, “we're stepping on each other's toes and saying, ‘Hey, are you doing that or do you not like doing that?’" 

With Proof, we came out of the gate really strong. We got to $75,000 of MRR really quickly. I would say it was instant validation from day one that this is really good. So there was not this long delay. I think founders get stuck and they wait too long before they've seen any validation and they just think it's coming around the corner. And my experience has been, it's usually very quick. And it may not be the fullness of it and it may not be all that you want, but usually there's signs of real life very, very quickly. And for Proof, it was certainly that. 

Proof ended up being a feature and not a real platform and we could never figure out how to take it beyond that. It was high churn, it was kind of easy to install, but meant it was easy to remove. A flood of competitors come in and clone it, giving you kind of pricing pressure.

We started to build this website personalization tool. That was what we pitched YC and we thought that was the future. You go to a website and the headlines are personalized and images are personalized, it's all very personalized for you. As soon as we got out YC, we shifted most of our focus away from Proof the popup to Proof experiences was the second product. And this is a real case study in doing it wrong. And we had this VC funding and so we felt like, well, it's OK if we don't get any feedback for a year. It's OK if this doesn't work right out of the gate. Whatever, we've got plenty of time here. Before it was like, if this doesn't work in 30 days, we're out because we don't have the ability to do that.

So we got a little sloppy, we got a little fat; the urgency was gone. We started building this tool and people liked it, but no customer ever loved it. And nobody in the demos loved it. It was just like, ‘Yeah, this is pretty cool.’ But we never had that pull from the market that I think we had from Proof that we had from some of our past businesses, that we certainly had with Jasper.

After 18 months of working on it, everyone just got kind of tired of it. It was growing slowly but not real fast. We were burning money, we were just fixing a bunch of edge-case bugs all the time. And we started having developers leave and they were happy, but they were just like, "Hey guys, I can't keep fixing bugs for the rest of my life. This isn't going anywhere." We were just like, I think the end might be near here. Because even the founders didn't want to work on it much anymore. So we ended up letting about half the team go, I think we were about 15 people at the time. 

Sandhya Hegde: This was early 2020?

Dave Rogenmoser: Late 2020. We kind of made it through COVID. Initially in COVID, made a big push, started growing again. But it just didn't sustain. And so it was October 2020, we let half the team go. We pulled back. I think this is a mistake founders make — when it's not working, they iterate too incrementally and they think, “Oh, maybe it's a little feature here. Maybe we need a position to better here.” There weren't small changes. There were bigger fundamental things that we needed to pull out of the forest. Above the forest, look around and be like, all right, if we could do anything, what would we do? And in that month we started digging more into AI and started looking more at GPT-3 and just realized, oh, these models are really good.

At the time we had been training, I spun up a little coaching business again teaching B2B SaaS companies how to run Facebook ads. There was a lesson on how to write great Facebook ad copy. And I was like, “oh, I could just use AI to do that for them. Let's build a little app that does that and we'll give it to these 10 people I was coaching and consulting with and see if they like it.” We had been so wounded by the 18 months of delay with Jasper or with Proof experiences that we were like, “If we don't really know this is working in 30 days, we're shutting it down and we're moving on to something else.”

Sandhya Hegde: I'm curious, the Proof experience is an 18-month stretch. Is it clear in hindsight or was it clear to you then that you're just not feeling the love? Was every meeting clear to you that you haven't found the aha yet?

Dave Rogenmoser: It was one of those things where I would read about product-market fit. And you hear it, you hear people say, "You'll know when you have it." And that was so frustrating to me because I was like, "That's not helpful. I don't know if we have it or not." And it's like, I'm missing the obvious here. If you're frustrated by that and if you don't know if you have it and you want the metrics, you want the answer, you probably just don't have it. Because a month into Jasper, I was like, "Oh, that's what that meant. This is absolutely what it feels like to be just pulled by the market so much faster." And with Proof, we had this joke. Anytime we'd run a marketing campaign or launch a new product, we'd kind of think, “OK, we want this marketing campaign to bring in $500,000. We want this new product to decrease churn by this amount.”

And nothing ever worked as well as we thought it would. That was always the joke — nothing ever works. But if you keep doing enough stuff, eventually it compounds and it works some. And then with Jasper, everything worked with Jasper. And we'd do a webinar, we'd have 10 times the attendees that we expected. We would launch something, we'd make five times more revenue than we thought. And it was just like, for the first time it feels like we're getting the benefit of something bigger than ourselves and we're not just grinding for every little dollar, every little thing that we did there. So that was pretty exciting.

Sandhya Hegde: I think that's the best description of what a tailwind feels like from a founder. Whenever you set a goal, the market just helps you exceed it and you just feel the love as opposed to having to grind for it every time.

Dave Rogenmoser: Yeah, it felt like we were executing perfectly with Proof. Again, we weren't really, but it felt like, man, we're doing really good work and our team is so talented and nothing we're doing is really moving the needle even when we work really hard and objectively. This is a really good product we had to build or marketing campaign we just built, it just wouldn't do anything. And so I'm just like, man, I finally got the point where I realized this is not working. It hasn't been working, if I'm honest with myself, for a long time. And therefore, it's very unlikely that it starts working anytime soon. And so we need to make a big shift.

Sandhya Hegde: How far did you get on revenue with Proof when you decided to pivot, in October 2020?

Dave Rogenmoser: We got it to about $250,000 a month of recurring revenue. We were growing about 3% a month. But the Proof experiences, the personalization product was a very, very small percentage of that. A lot of that was the pop-up business. And we actually looked at selling the company and we were talking to a broker and we're just like, "What is this thing worth?" And he's like, "Well, it's kind of a weird company because it's growing, but it's growing slowly. But it's also burning money. You need to grow way faster.” And people buy fast-growing companies that burn money or you have to get really profitable. We didn't know how to grow faster. And so we just thought, well, we need to strip down the team and we need to get rid of a bunch of bloated software and we'll just get this profitable and let that ride it out.

Sandhya Hegde: All right. I think we are ready for fall 2020, Dave.

Dave Rogenmoser: Looking back, we spent too long on Proof and obviously, that's clear in hindsight. I don't think we should have just not done it. But I think probably six months in, we could have called it and we could have said, "This is not working." We also should have been way more ruthless with a real MVP and thought we need to get this into the market with customer excitement inside of a month and outside of that. I'm not the one to say, we just shouldn't have done it and we should have known right out of the gate. But you don't need to spend 18 months on something that's not working. If you're past that, I just think the odds of it really working are pretty small in most cases. And at least with MarTech. There's probably some hard tech that just takes a long time. But at least if you're building apps and tools for companies, you could probably pivot way sooner than that.

But yeah, fall 2020, we had been in this space for a long time. This is six years of working with marketing teams, helping marketers grow their business in a variety of ways, helping people write content better, helping people do copywriting better. I knew the customer deeply. I'm a marketer by trade. We had an email list of past customers built up and people that were following us. Building a tool around AI was pretty straightforward for us. So I think we really had founder-market fit when we went into this. It was like, oh, we're the perfect people to bring this really cool thing to market. So yeah, we built... again, same thing. We said, all right, we're going to build this little tool that writes Facebook ads, Google ads, and writes landing-page copy well, and we're going to launch it within 30 days of starting on it.

We built it and got it into people's hands. I was doing demos. I started off by just doing one-on-one demos over Zoom and I still got recordings of all these and people I would show it to them and they would just start cussing and they would just say, "Holy bleep" and "This is the best thing I've ever seen." They started saying things like, this is the best thing I've ever seen since, and they would say the N64 when I was a kid, or the first time my house got dial-up internet or all these big things. I was like, holy cow, these are significant technologies you're comparing us to. And then they would just say, "When can I get it? I'm ready to pay, let me do it." And it was just instant validation again. Not just with their words but with money. They're ready to put their credit card down and there was real enthusiasm around it. And so we —

Sandhya Hegde: Who was the persona you were focused on? Was it kind of the same small business, owner, marketer? Who were the folks that you were focused on? Was it them saying, "Oh my God, this will make my business more productive, I can work with more clients"? Who are those first adopters?

Dave Rogenmoser: Most of them were serious marketers, either marketing for their own online business, probably some courses and coaching and people like that where again, that's an   industry where marketing really matters because the product is not differentiated. And so just everyone becomes a really great marketer. So it was them or it was agency owners that had a bunch of clients that they needed help with. The pain point that I felt viscerally when I ran an agency is you've got all these new clients, but it's really hard to get the copy spun up for each one of them. All of a sudden, Jasper can just bring you up to speed with a new client, day one. So it was serious marketers that I knew lived and died on how good their marketing was.

Sandhya Hegde: Makes sense. So they have already tried a lot of new tools. They're looking for solutions as opposed to, oh, I don't know, I'm going to rather play it safe.

Dave Rogenmoser: And they knew me. They had trusted me, they followed me for years. There was definitely trust there where it's like, "Hey, if Dave has something that he thinks is really good, it's at least worth looking at." But they'd never seen anything AI-based. I mean, that was all just brand new. And so this was mind-blowing to people out of the gate.

Sandhya Hegde: How did you think about the user experience, the workflow, given that it's so new? Of course, it sounds like your early adopters were bleeding-edge, we'll try it the way it is right now, perfect early evangelists. But you must have thought about how do we reimagine user experience for this extremely new, what feels like magic to the customer? How do you think about UX? I'm curious, what were the next few months like and how did you go from MVP to no, this is kind of the broader platform that can cross the chasm?

Dave Rogenmoser: Yeah, so we just guessed first. We just said, "Hey, we don't have time to do a ton of user interviews. We just want to get this into the hands of the market asap and then let them pull that out of us." And so we just took a guess. And there's a lot of things that were right, a lot of things that were wrong about that. Our MVP didn't have almost any features or functionality. I don't even think you could sign up on your own. I think we had to email your log-in and password. There's no cancellation button, there's no way to do anything. There's no user settings. It just is the core of the app. And we're like, get it in people's hands right now.

And then, yeah, we engaged. We had a community, we had our Facebook group early on and I was just hanging out in there all day, every day talking to people: "How are you using it? What do you guys need? What's bad about it? Where is it annoying?" And then we would just try to ship something new every single day in a very lightweight way. Speed was all we were trying to do. And just knowing that if we can just listen to them, build something that they say that day, this is going to work. And that iteration cadence was so powerful. And it also builds a lot of trust with your customers because they just think, “OK, they're listening and even if it's not what I want it to be right now, I've got confidence that it's going to get there.”

One thing that we saw early on is we thought this was a conversion optimization tool. And it was actually called conversion.ai was our very first name. And I thought in terms of it's going to help you convert more ad buyers, it's going to help you convert more landing page and a very direct response marketing kind of language. And then we started asking, "What do you want?" And they all said, "Well, we want to write longer stuff. We want to write long emails, we want to write blog posts, we want to write long social media posts." And they started talking about content more than they talked about conversions. And we realized, OK, conversion, performance marketing's definitely one component of this. But it felt like Jasper actually was being used and wanted to be used to write more long-form content.

And so I think one of the first big shifts in my mind was, “oh, this is not a performance marketing tool, it's a temp marketing tool and that means we've got to shift the product to handle longer stuff there.” And even right out of the gate, we launched a product and the first weekend somebody wrote a book with it. And again, they were using this little tool, it was called The Sentence Expander that took a small sentence and turned it into a longer sentence. And they wrote this whole book with this miserable tool that's not built for that at all. And I was like, it hadn't even occurred to me that somebody might want to write something longer than a few sentences. That's obvious in hindsight. But that really got pulled out of us by the market and I really did not even see that going into it at all.

Sandhya Hegde: How big was the Facebook community when you first started and you first launched your product, like early days, maybe early 2021? I assume you got a lot of word-of-mouth folks joining. But what were the early marketing tactics you used before you went out, raised money, etc.?

Dave Rogenmoser: Yeah, it was small. The very first was probably me and 20 people. So it was small out of the gate and then it kind of grew as fast as our customer base did. But yeah, we started off really small and really intimate and I knew every single person by name. I would interact with every single person. It was this intimate little thing where it was just a bunch of friends hanging out. And that was really great. Again, I didn't think of that so much as a marketing thing, as I thought of it as a product development need. And I just needed to be close to them so I would know what to build. It turns out it's a pretty good way to activate people and communities have so many benefits to it. So we invested a lot there, but it felt very organic and natural. I was just having fun.

We also did a lot of Facebook ads early on. We were good at that, it was a core competency we had. I knew that the best companies grow through word of mouth, but you've got to jumpstart the flywheel somehow. Otherwise, if I've got 20 customers and my viral coefficient is 1.1 or something, it just takes a long time. It's like, well great, well let's go buy our way to having a bigger customer base. So we did that and that worked effectively. We had an email list from past companies that we'd market to. But it really was strong word of mouth from early days. People would just rave about it and tell other friends. And so we were definitely propelling the market forward some, but we were also just carried by the market and word of mouth was great.

Sandhya Hegde: Pivoting a little bit to the tech side, how do you think about the trade-offs of working with foundation model companies like OpenAI, a lot of others as well, versus building your own, having your own infrastructure? Could you walk founders who are thinking about building AI-native software companies, could you walk them through what are the trade-offs like in terms of gross margin, in terms of control, in terms of what are even the vectors to really care about here?

Dave Rogenmoser: Early on, we didn't know how to do anything with AI. And so it wasn't even an option really. We'd have to go hire all that out and really get up to speed there. Our team was really great at community, growth, marketing, and customer depth. I think we just knew the customer so deeply and so much more deeply than our competitors because we'd been working with them for years that we knew those were our strengths and speed to market was going to be key here. And we were thinking, this is going to be a huge market, nobody really knows it yet, but we've got to get to number one really fast. And so we were just like, we're just going to ride off of open AI's GPT-3 API.

And we did train it. And so even early days, I would say it was better than most of the competition and that customers would tell us that, like, "Why is it so much better? You're all using the same thing." And so there were things that we did there. But the core of it was another company's product that we used. And I remember having a conversation with a founder that had a kind of an adjacent product, not competitive, but somewhat or at least would be competitive in the future. And they were telling me how dumb I was for using another person's API, and they were building it all in-house. And they were saying, "This is not going to work. You don't want to be reliant on these other companies. You're not a real company." All of that.

And I was just like, "Well, let's just see how this goes. Maybe you're right, maybe I'm right. Let's see how this goes." And it's obviously gone tremendously for us because that's a whole part of the company that now we don't have to focus on early days when resources are so limited. Means we can move really fast, it means that we get all the updates that they have without us having to even work on them at all. And again, there's some risk that OpenAI goes out of business in that first year, something like that. But we waited and we're like, that seems like a pretty small risk here that we're willing to accept. And so I think since then we still work with OpenAI a ton today, love working with them. They're terrific partners. We don't want to be reliant on one company. And so we've got our own AI team now and some of what they do is they're fine tuning and working deeply with OpenAI.

So what they're doing is expanding and looking out in the market at what are the other things out there that we can do to solve our customers’ needs. But that has come so much later in the journey. And I think if we would've spent a bunch of time on that early days, that would've been a huge time sink and would've really slowed us down. And so unless somebody has deep expertise in AI and really perhaps has an angle there that they think they can exploit, I think most people should probably just start with some out-of-the-box solutions, whether it's open AI or could be other models. But should start there, build some use cases around that. And then start to think about: how do you differentiate or diversify as you grow.

Sandhya Hegde: And now that you're building your own, what are the advantages you are hoping to build? I'm sure it's always a work in progress, but what does the ideal look like there?

Dave Rogenmoser: For us it's always customer-led. I think a lot of folks, especially in tech or maybe even AI, they're almost white paper–driven and it's, "Oh, this new breakthrough happened, how do I force that into the market?" They start with, "I want to use this new technology, what should I do with it?" And I think we really try to say, "What are our customers asking for that we can't currently do?" And in some cases with let's say OpenAI, again, there's some things that we can't do. We don't have access to the underlying models and we can't do a lot there. And so sometimes it means looking at other models to explore other things that our customers want. So there's just more control. Obviously, that comes with complexity. But there's more control.

There is cost savings. I don't particularly care too much about that right now. Our gross margins are fine even working with OpenAI, but there's some cost savings that will be material later on that we might say, "Hey, we need to go get more profitable. And so we're going to run our own models there." There's things like speed. We could have our models run faster for the user there, but that's not a big thing. I think for us it's really just control. There's use cases that we want to go explore that you can't do out of the box with OpenAI that we feel like it's our duty to go solve those needs for our customers. And sometimes we need to look at other things for that.

Sandhya Hegde: What would be your advice for founders getting started in 2023? If you were doing a fresh start, what would you want to tell your future self?

Dave Rogenmoser: Ride a big wave, which right now is generative AI. So I would be starting with generative AI. And that doesn't mean necessarily text, it could mean image generation. I see cool stuff with audio generation coming out. There's cool stuff with video generation coming out. So the industry right now is so early that I think a lot of people default to thinking about marketing copy or things like that. And that's one tiny little thing that happens to have worked early on. I would understand that tech, but then I would go to a customer segment or an industry that is not fully up to speed on all of this, and I would start just talking to them and I would start asking a lot of questions about where their work is inefficient; tell me about this process that you do. What do you do here? Where's the biggest time wasters in your day? What's the biggest pain point in your business?

And I'd start to ask all these questions. So let's say for example, you went to a law firm where you said, "Hey, I'm going to figure out how to do this for legal teams." I was telling somebody that the other day and they were like, they said, "Oh, and do marketing copy for legal teams?" And no, not at all. And they were like, "Well, what would you do?" It's like, "Well, I don't know." And I wouldn't go into it with a bunch of preconceived notions about what to do. I would come into it with a totally blank slate. I'm just interviewing you. Maybe there's something here, maybe there's not. But I'm guessing if you went and interviewed 10 law firms deeply, you're probably not going to come away thinking, "Oh, I need to build a marketing content tool for legal teams." It's going to be something around some process that they all think is super inefficient and terrible. And you think, oh, I'm pretty sure AI could do that thing here. Let me go build a tool for you and do that.

But I think there's a lot of arbitrage to be had in taking this technology to these older industries or different industries that haven't heard about it or use it or would be a long way off from using it. And if you know the customer really deeply and you just commit to going deep there, you're going to have so many great tools and updates that you can give them that solve a deep pain that nobody else is going to see. And I think too many people are just hanging out in the AI world and not thinking about how to take this to end users. But I think there's just a tremendous amount of money there. There's not probably a lot of money in copying Jasper and building one more marketing tool. There's not a lot of money in probably for most people competing against open AI or stable diffusion. For most people it's, hey, go somewhere where this is not and be the person that brings it there. And there's a ton of unicorn businesses by doing that.

Sandhya Hegde: Yeah. And I love your point about founder-market fit. Find something that you care a lot about and want to learn about if you're not already an expert in. This was amazing, Dave. Thank you so much for joining us. 

Dave Rogenmoser: Thanks for having me.

For more stories from iconic founders and their journey toward product-market fit, subscribe to our podcast, The Unusual Ventures Startup Field Guide on Spotify and Apple Podcasts. Then discover all of our startup building resources in our monthly newsletter — sign up here!

All posts

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.