Michael Lingelbach (00:00:00):
You really want to find a product or a market. Where there's people that are using you casually, but then there's a professional upsell version of that.

Pablo Srugo (00:00:09):
How many users do you get in that first month, for example?

Michael Lingelbach (00:00:13):
A million or something.

Pablo Srugo (00:00:15):
Wow.

Michael Lingelbach (00:00:16):
We grew to a million in revenue relatively quickly.

Pablo Srugo (00:00:18):
How fast?

Michael Lingelbach (00:00:20):
Months, five months, six months. We pretty quickly hit like an eight-figure run rate. Been really exciting too is also seeing the enterprise adoption. I think we sign an enterprise contract every couple days now. Pretty big ones now too, and that's without outbound sales. And I think I've said this a lot that free users can be really false signal. That just because someone uses it for free doesn't mean it's something that solves a core pain point.

Previous Guests (00:00:43):
That's product market fit. Product market fit. Product market fit. I called it the product market fit question. Product market fit. Product market fit. Product market fit. Product market fit. I mean, the name of the show is product market fit.

Pablo Srugo (00:00:55):
Do you think the product market fit show, has product market fit? Because if you do, then there's something you just have to do. You have to take out your phone. You have to leave the show five stars. It lets us reach more founders, and it lets us get better guests. Thank you. Michael, welcome to the show, man.

Michael Lingelbach (00:01:10):
Thanks so much for having me. It's good to be here.

Pablo Srugo (00:01:11):
Dude, so AI is. I mean, obviously changing everything. You have one of the hottest kind of AI apps right now. Tell us maybe a little bit. I mean, I've played around with it. But maybe give us just a short idea of what Hedra is and what it does? Just to set some context.

Michael Lingelbach (00:01:24):
Yeah, so Hedra's a generative media platform. So we're both a product company and a research company. We train models that are natively omnimodal. That means they can operate jointly over image, text, and audio. And the goal is we're really trying to empower content creators from either prosumer people on TikTok. Who you might have seen some of our viral content from. Some are more enterprise marketing agency clients. To really be able to bring together these tools and create content natively with AI. So we take this from two approaches. One is how we build models with net new capabilities, and then you're going to see some really exciting stuff from us there soon. But also how we fundamentally rethink the UX and UI around these products. I think as you've seen with like Cloud Code versus IDEs recently. I think oftentimes as the capabilities of these models increases. We actually need to rethink how we're actually building the core product experience. And that's something I spent a lot of time thinking about at Hedra.

Pablo Srugo (00:02:21):
Are you only using own models? Or are you also able to use other models?

Michael Lingelbach (00:02:25):
Yeah, so we allow you to use multiple different models. We have our own proprietary models. Which are more popular, typically because they offer differentiated functionality and are very cost-effective due to our architecture. But we also let people use their same content library with basically all full leading model. So that people don't feel like they have to hop around to multiple subscriptions, but they can stay within a Hedra studio.

Pablo Srugo (00:02:50):
Is that like? Walk me through the thinking on making your own models. I had this conversation with another kind of a series B, purely AI app. And he was very like his strong opinion was, the vast majority of people that build their own models end up. I think he referenced like Harvey or something like that. But the vast majority of apps that build their own models end up just discarding them. Because, the big foundational models, they might not be as good as your model today for your app, but they will get there and then they'll just destroy you. How do you think about that? And the time you're investing there?

Michael Lingelbach (00:03:20):
That's a good question. I think for language models, we've kind of figured out the general interface to language models, right? It's like text in, text out, not to be simplistic. And that just tends to work for most of the LLM use cases. Especially when you think about where the LLMs are getting the most adoption. Which is search and code generation. I think for visual media generation is quite different. Because a lot of times this is a much more creative pursuit. Which is more about, there are many ways of communicating intentionality, between you and a model. And I think that opens up a lot more parameter space of like, not only how these models are designed, but also how you can communicate intentionality to them. So I think LLMs are just much more farther along, and they have a very well-defined path. And it's really just about scaling them at this point. And working on reasoning, improving general intelligence. I think for media generation, it's still too early for that, and we're seeing things like fundamental architecture changes happening even now. So that's why I'd say we're not in the stage where you can just build on top of a general purpose media generation model just yet.

Pablo Srugo (00:04:27):
Got you and then just maybe for context, like if you. For those haven't seen Hedra, pretty simple UI. I mean, you go in, you just type some stuff, you add in like an audio or text. You add in maybe an image or a start frame or whatever, and it'll create a video for you kind of based on all that.

Michael Lingelbach (00:04:44):
Yeah, that's how it works right now, and you can also do. We have an audio studio product. Where you can make voice cloning using different partner voice models. We have a whole image generation suite now. So really it's about that core workflow of how do you bring together different content types and assemble them into something that you can actually publish as a short form video.

Pablo Srugo (00:05:04):
What about, so here's the idea. I just tried to do this right now, right before I interview you, and it wouldn't let me. I'm curious if this is something you would end up doing or not. We do these long form podcasts where we record the video. But it's remote and people don't love watching remote. So I was like, I would love to put in this audio and these videos. And then have you make it so it looks like we're in the same room together chatting. Is that something that could be done?

Michael Lingelbach (00:05:30):
Yeah, I think that's a great idea and something that we're thinking about. You know, historically, we've worked a lot on. How do we have good dialogue as part of a general purpose video model. And now a lot of what we're thinking about is how do we bring net new capabilities that are useful for this workflow that we've kind of grown to own. Really around, you know, how do we make really good podcasts? How can we make really good marketing material? I think a lot of that is making video models be able to more flexibly generate and edit at the same time. So I think in the future version of Hedra, you could chuck some video in and some character references in. And say, hey, edit this podcast and get something out that felt really professional with very little work.

Pablo Srugo (00:06:12):
So let's now, we've kind of gone on a bit of a tangent just to set context. But, yeah, tell me then about. Let's go through a story, as much detail as possible. I mean, you started this in 2023. So it's been two years? 

Michael Lingelbach (00:06:23):
Yeah, about two years now. 

Pablo Srugo (00:06:24):
What's your background? How did you end up in the place where you started Hedra?

Michael Lingelbach (00:06:30):
Yeah, so I was a PhD student with Fei-Fei Li and Jiajun Wu at Stanford. And I was in the last year of my PhD. And you know, ChatGPT a had come out like, half a year or so. And before, and there was just all this excitement around building applications and new technology in the space.

Pablo Srugo (00:06:50):
What was your PhD in before that in politics?

Michael Lingelbach (00:06:53):
I was in the computational neuroscience program, but most of my work was in the computer vision embodied AI realm. I had just wrapped up a project. I had submitted my Greenlight meeting to go do my defense. And then I told my advisors, hey, I want to take some time off immediately. Because I feel if I wait to do my defense in five months, I'm going to miss this big opportunity, and they're really receptive to that. And so originally I was going to go back. I was going to go set up the company, build some stuff and then go back. And then I just ended up raising money. And it just kind of led to the really getting locked in. And I've been meaning to go back and finish my defense now for like two years. But, you know, it's just hard when you start a company. You just get so fixated on everything and building. And hiring that like thinking about going back, and doing like a month for finishing my PhD is actually, you know, seems like a lot of time.

Pablo Srugo (00:07:50):
Maybe if you don't mind going deeper on what you were doing and how you saw. And what you saw, because I think that's obviously key. I mean, everybody knows what happens afterwards and kind of the success you've had so far. I mean, you raised a $30 million Series A a couple months ago from A16. But, what exactly were you doing? And what did you see in that ChatGPT moment that made you so decided to jump in?

Michael Lingelbach (00:08:12):
Yeah, I think for me. It was realizing that a lot of the clever approaches were that we were doing academia were ultimately nothing in the face of scale. And that, if I wanted to build something impactful. I felt, I needed to do it in an industry setting at that time. I think academia has a lot of extremely valuable contributions still to be had. But a lot of the core productization of the work is really just about finding that core consumer or enterprise need, and building a model that can really serve that. And then building that workflow around that model that will solve that use case. And I felt, you know, for me, I didn't have the concrete idea or anything when I dropped out. I just felt there was something there, and I took some time to think about it, and talk to people. And I kind of came to this conclusion that. There's this at the intersection of a lot of my interests and where I felt there was a gap in the market. Because I come from a theater background. I grew up doing a lot of acting, performance and arts. But, I worked in computer vision and, I was in this community of people that were also very startup-minded, I'll say. Karen for Cartesia was in my cohort, Demi from Pika. I was really interested in this idea of there were these companies that were doing avatars, and then there's companies that were doing generative media. But, there wasn't this company, I was thinking about the intersection of them. How do we make these general purpose models that are really configurable around performance. And I just felt that was a core missing heat in the market. I saw the rise of these faceless TikTok channels. I saw how, you know, companies were doing brand messaging, testing, and rerecording the same video like 40 times. And I just saw something in there about, could we make something that would automate all of this from both a platform and model perspective? And that was kind of that core moment for me that clicked. And it's two years ago, so it took a really long time to get there.

Pablo Srugo (00:10:19):
Two years isn't that long, by the way. I mean its an amount of time.

Michael Lingelbach (00:10:22):
It feels like a long time, it feels like a lifetime ago. I feel like I'm a totally different person now. 

Pablo Srugo (00:10:28):
I mean, a lot's happened in those two years. That's legit.

Michael Lingelbach (00:10:31): 
Yeah, I mean it's weird because in some ways a lot of stuff has changed but in some ways a lot of stuff hasn't. Like, I don't use ChatGPT anymore, I'm more of like a Gemini Anthropic person. Depending on what I'm doing, but I opened up chat GPT the other day and I was like, oh, this looks like I remembered it back in 2022. But that's still one of the fastest growing products in the space, you know? But on the other hand, now we have all these really advanced application layer companies and the core models are just generally much more capable. But it just still feels like a long time. Just when you're going through, we've been through two office moves. This is our New York office, which is kind of crazy. So now we have a second office. It's just like, from when I was in my apartment, and talking to three other people to this. It just feels very different.

Pablo Srugo (00:11:18):
And so, in the computer models you were building that you're saying are not really productize-able. What was an example of some of the things that you were working on, in terms of computer vision?

Michael Lingelbach (00:11:27):
So during my PhD, I did a lot of projects with neural rendering. Towards the end where I was really interested in explicit 3D representations. And I think one thing that I realized pretty early on. Is that a lot of times these really explicit things where engineering hacks oftentimes will get subsumed by the model. So, I think like 3D stuff is extremely interesting. Especially for like the realm of gaming, but for generative media. I think these big models are just getting really good. And so that's an example of something that's still has a lot of interest in academia. But, I think for certain use cases, these large scale diffusion models are just solving a lot of the problems. 

Pablo Srugo (00:12:09):
So they were like, you were like hyper specializing on specific use cases. But you think that the generic ones are just going to solve that over time. That's kind of your idea there?

Michael Lingelbach (00:12:17):
I think it's more that there's certain technology that will get subsumed by models. So I think a lot about a genetic orchestration is a good example. If you look at a lot of products that emerge, especially the AI coding space. I think we're seeing this massive adoption of Claude code right now. And Claude code is just kind of this unified Claude plus tool use, and it turns out that's like enough when the model gets smart enough. And so that's more what I'm saying is that these, there are still specialized models. You know, Claude is probably not as strong at search and being a companion as maybe Gemini or ChatGPT. But for coding, building a unified model and building a deep productization of that can be enough to just solve a huge class of problems. 

Pablo Srugo (00:13:00):
So by the way, you mentioned this. I'm just curious, why'd you move from ChatGPT to Gemini? Any specific reason?

Michael Lingelbach (00:13:06):
You know, it's one of those things where it's just Gemini 2.5 Pro is really good, and Google has really good search indexing, and it's built into Google Workspace. So it's in email and drive, and everywhere now. And I think it's also very fast. So for general purpose, I just have it in our Google Workspace subscriptions. Because we pay for this through the company. So I just use that for all my day-to-day questions. And then I'm a huge Cloud for Coding user. I think Cloud Code is amazing. So that's my primary work set. But now I use Neovim and Claude Code together in the terminal, and that's just 95% of my AI use case now.

Pablo Srugo (00:13:48):
So going back to the storyline, you decide to drop out. Walk me through specifically, what you're doing day to day. What are you doing in those five months period?

Michael Lingelbach (00:13:59):
Like, the early pre-Hedra or post actually founding the company?

Pablo Srugo (00:14:04):
Pre-Hedra, when you dropped out but you're figuring out what are you going to do?

Michael Lingelbach (00:14:08):
Yeah, I mean. I was kind of talking to people that I wanted to recruit. I was like, it's really hard to convince people to join you when you have no money. I think people.

Pablo Srugo (00:14:17):
That's right.

Michael Lingelbach (00:14:18):
Underestimate the difficulty. Because you're like, oh, you drop out of Stanford, and everything falls in place. It's not exactly like that. But then I started being, I didn't know what a VC really was. It's kind of embarrassing in retrospect. But I'm like, oh, what's venture versus growth equity? Who do I talk to? And so I just spent some time building prototypes and trying to convince people to go on the journey with me. I did have a co-founder at the time. So we were trying to put together some things.

Pablo Srugo (00:14:50):
And you're building a prototype of what? What is this thing that you're trying to solve for at that point, right? Versus what is today?

Michael Lingelbach (00:14:56):
Yeah, it was originally trying to build essentially a compositional avatar plus scene rendering engine. It was more or less pure end-to-end model-based, more compositional rendering. 

Pablo Srugo (00:15:08):
So, you put some text, you get an avatar, and they do something?

Michael Lingelbach (00:15:11):
More like you have to take a video, and that video gets turned into something that's a representation of that video. It was more like assembly than pure generation. We made a really cool prototype. Well I made a cool prototype.

Pablo Srugo (00:15:24):
Was this pre-Sora? Pre the Sora video they put out and all this stuff?

Michael Lingelbach (00:15:28):
Yeah, this was pre-Sora. 

Pablo Srugo (00:15:30):
Okay.

Michael Lingelbach (00:15:31):
This was 2023 at the time.

Pablo Srugo (00:15:33):
Right, yeah. Right after ChatGPT came out.

Michael Lingelbach (00:15:36):
And yeah, we ended up raising money, getting the seed round done. Because there were some investors that were really interested in space. I really connected with Index at the time. Then it was kind of more about the.

Pablo Srugo (00:15:47):
How much was the seed round?

Michael Lingelbach (00:15:51):
We raised seven for our seed round originally.

Pablo Srugo (00:15:53):
That's a big one. I mean, obviously in the world of AI it's happening a lot, but what's the story there? How did that actually come about?

Michael Lingelbach (00:15:59):
Yeah, I had been thinking on these ideas. I had kind of a concrete vision at that point. Pretty similar to our vision now, maybe much less mature. I didn't really understand customer profiles. How focused you had to be if you were trying to do enterprise versus doing something that's just purely prosumer. But, I think the core idea of building content creation workflows around these models, giving people more long form video controllability, having people at the center of a lot of the video was there. And then I had, kind of, people had figured out that I was doing something. Because I just have a lot of friends that would, get approached by investors. And then they'd be like, Oh, you should talk to my friend, Michael, who's building something in the space. And then we got scouted by a couple funds and then I had gotten connected to a couple funds through mutual friends from Stanford or through my broader network. And then that kind of led me to meeting with a bunch of people and getting a lot of investor interest. And then I really connected with Cat Wu, who was at Index. Who's now coincidentally one of the leads of Claude Code. I think you may have seen they got poached by Cursor and then they just went back to Anthropic recently. And she really sold me on Index and working together with her. And that kind of was, what led us to kick off with them. And then Abstract and a16z participated in the seed round as well.

Pablo Srugo (00:17:27):
When you raised that seed round, was it just you? Or what was the team like?

Michael Lingelbach (00:17:31):
Yeah, it was me and my co-founder at the time, and then a couple other engineers who joined early. Terrence is our employee number one, and he's still at the company. Our oldest employee.

Pablo Srugo (00:17:45):
You raised the seed round. What's the thing that you're focused on doing? Focused on building first?

Michael Lingelbach (00:17:51):
Yeah, so I had wanted to get something out to market. So we were trying to build on this product idea that I originally had. This technological approach, and so we spent a few months working on that. And at the time, I was also trying to recruit. I didn't really get how to recruit for a startup. I know it sounds kind of weird, but like you ping all your friends in your network and that's how you get some people. Oftentimes the best people, but I didn't understand what working with a recruiter was like. I didn't know what an ATS was. I didn't understand an interview loop. It was very vibes based early on and quite messy.

Pablo Srugo (00:18:25):
Is that good though? That's potentially a good thing, no?

Michael Lingelbach (00:18:28):
I think it's good can be good and bad. Because I think oftentimes when you're building a startup. You want to have really small time to market. But also you kind of want to lay the foundations that when you get to market. You're not building on, shaky terrain. So a lot of our early system was super unstable, very crashy, and I think had we brought in really experienced staff level engineers to anchor the team. I think probably it would have been easier to scale, and that wasn't something I really appreciated. Especially because, I was a researcher by trade. I'd never run an engineering team before. Yeah, I think, definitely retrospect. One thing I'd say is, people should consider bringing on a talent person if you're a venture backed really early on. We just hired the most amazing head of talent in the valley, Camille. Who just joined us from Niantic, and, I can't tell you how much better that's made my life.

Pablo Srugo (00:19:22):
How many, but with that with that $7 million? How many people did you want to get to? How many people would you hire?

Michael Lingelbach (00:19:26):
I think it's good to get to a team of 10. I'd say on that sort of budget. I think in general, when I think about ideal composition of early stage teams. I think you should have one person, ideally the founder who has the product vision and roadmap. You want one person who's really anchoring the engineering side. If you're doing research, you want one person who's really anchoring the research side. So that could be, you know, a head of engineering, a head of research, those people should still be really hands on though. You can't have hands off people in an early stage startup. It just does not work, and then you really want a few, you know, a solid full stack engineer to a solid platform back, and engineer. Ideally, couple staff level people. And then you want I'd say you need one designer and then you should bring on a talent person early on is my opinion. Because you just spend so much time thinking about talent, and making sure you're running a clean process and you're getting the best people. It can really become like a full time is a full time job, even at the early stage. And I think, when I think about what I spent my time thinking about really early on. I wasn't spending my time always thinking about product. I was spending a lot of time thinking about operations and hiring. And I think had I figured out how to accelerate that earlier. We would probably be where we are in at least half the time.

Pablo Srugo (00:20:43):
So one year instead of two years. I mean, you're going fast, by the way. Just historically speaking, but I get what you're saying. I guess here's a question for you. Well, how do you? These people that are joining you early, these first 10? I mean, you are funded, there's some credibility, but are they leaving something to join you? Or are they kind of, you know, just recent graduate from a Ph.D. or something like that? What was the profile of most of these people?

Michael Lingelbach (00:21:07):
I think a lot of the people were pretty fresh early on. I think that's what you tend to get. Because oftentimes, even with a large seed round. It can be difficult to convince people to leave a stable job, right? A scale or a Databricks, one of these more established, very late stage startups. And oftentimes, the I think the people that are best for setting up a seed stage company are the people that did a series A to series C stint. Where they've seen completely green fielded systems. They know how to spin up things from scratch, but they also ideally have worked with a high growth team. So they know what the hours are, they know how to build a company culture. Which is also something I didn't think a lot about when I first started the company, and now I spend a lot of time thinking about. Yeah, I think a lot of the people early on were fresh, and now when you look at our team. Most of the people we're hiring are from really later stage, high growth, well-known startups. Who were ideally there at the beginning. or pretty early, but have that like are basically two timers.

Pablo Srugo (00:22:13):
And so those first 10 people. Are you focusing more on UI product? Or you building foundational model things? What do you spend your time on with that seed room?

Michael Lingelbach (00:22:21):
Yeah, I think I spent a more time on the core model work at first. And in retrospect, honestly, I think you should probably. I would not recommend most companies go out and start a new model program. I think there's just too many opportunities at the application layer right now. But, I think for a research company.

Pablo Srugo (00:22:42):
But back then, even two years ago. That was probably different, right? Or are you saying you would have done things differently?

Michael Lingelbach (00:22:48):
I think two years ago was different. I think, now it's definitely very different. Because, I think in six months the capabilities, these models are going to be good enough for most the use cases that you'll need. And a lot of it is going to be, can you build vertical specific workflows that solve specific pain points people have? Because not everyone just wants to go into like in generic chat assistant. There's a lot of work that could be done there. I do think there's some risk, because it seems OpenAI and Anthropic especially are pushing out of just being model providers into being vertical providers. I don't know if you saw. I mean between Claude Code is growing massively and is, I think disrupting some of the IDE products. I think also there's the Anthropic's financial product they just launched. Which is more focused on, you know, that Rogo Hebbia kind of market. And so I think there's things that the foundation model apps can't do because they can't do everything. And to me, that's the area to focus if you're building la new company.

Pablo Srugo (00:23:47):
It almost seems like, this is a bit of a tangent, but from a value capital perspective. The foundational model layer on its own is not actually the big prize. ChatGPT is consumer app, you're saying, you know what I mean? Anthropic is more on the development piece and I don't finance. But there's actually more value to be captured if you have the best models by going vertically integrated and owning space versus just being a model layer. I mean, I guess that's one read of the situation.

Michael Lingelbach (00:24:13):
I think that's true. Because I think you can also optimize the model for a specific use case, right? You can train it to do really good agentic tool use. You can build out the UI where you can directly access parts of the model. And I think the frontier labs have realized this. And so it's like why they're doubling down on all these products. I think OpenAI is launching something today, maybe the browser. At the time, I just hired out people mostly focused on model, and what I realized is you need to product. You need to productionize it and need to solve specific need for people. And I think that really involves getting at least one really good product designer in early. You being a really like ,the founder needs to be that product manager that understands the market pain point and  can translate that into specs for the designer. And then, you know, really strong design oriented engineers who know how to build something that's performance stable and beautiful. Regardless of what sector, I think everyone wants to use software they love.

Pablo Srugo (00:25:10):
I guess that's the key question to dive into. As you realize that you've got to prioritize, you've got to find something you solve. At the end of the day, you need to find a problem that you're going to solve, regardless of everything else that you're doing. What is the first thing that you're going after? What's the first use case? Or are you even thinking about it that way?

Michael Lingelbach (00:25:28):
Like, and specifically to Hedra. What I realized is that there was pretty good synergy between people who are pro content creators and people who are marketers. And that a lot of that content involved dialogue centric scenes. So building a good workflow that let people do this generated video. But also let them put their message into it and do it in a fast, easy way was what our first wedge was. And so for me.

Pablo Srugo (00:25:50):
It's short form video for Reels, TikTok, et cetera. Is this kind of what you're going after?

Michael Lingelbach (00:25:55):
Yeah, I think for me, I thought there were. My theory was there were a couple four things, like basically this faceless talking channel trend. There's a lot of people that create tutorials or product reviews, or, sell a product on TikTok. That's a pretty good fraction of people who make money off of TikTok.

Pablo Srugo (00:26:15):
That's easier, right? In terms of an AI gen problem. To do those than truly avatar, human-like experience.

Michael Lingelbach (00:26:22):
No, I think they're pretty comparable. We thought we wanted to make the model so it was flexible. So you wouldn't have to think, do I use Hedra or do I use this other thing? It was just like, oh, use Hedra. Because if you add cognitive overhead, right?  It kind of ruins the magic of AI. If I had to go and be like, oh, do I use OpenAI or Anthropic for this. It does kind of make it a little bit more annoying. And so you kind of just want something that does it all. So for us, it was like, okay, there's going to be these faceless channels, people are going to make UGC content. Podcasting, I was, I feel this is going to be a thing. And, that was, I think, a pretty good read. Because I ended up being  a decent chunk of our revenue. And then I was like, okay, you really want to find a product ora market where there's people that are using you casually, but then there's  a professional upsell version of that. And the nice part about these, you know, social media is social marketing is huge and growing and like podcasting is like something that companies now do to promote their own interest right or give product reviews. I think OpenAI has a podcast. And so that was kind of like my plan is like we wanted to have this PLG prosumer popular platform that solved a lot of the cognitive overhead of creating this type of content but then have a clear enterprise use case for company communications and marketing. And that was the vision behind the Hedra product, which I think panned out.

Pablo Srugo (00:27:49):
What's the main podcast use case? How are people using it?

Michael Lingelbach (00:27:51):
I mean, John Mahua has a really popular channel. Where he generates a lot of podcast content. He did this one between a Jesus podcast was, I think, the first one that was really viral around Easter, and then he did this baby podcast series.

Pablo Srugo (00:28:08):
Yeah, the baby one.

Michael Lingelbach (00:28:09):
Yeah, now he's doing another one. He's doing robots. That's probably the most viewed popular channel. Because those get hundreds of millions of views.

Pablo Srugo (00:28:19):
Wow.

Michael Lingelbach (00:28:20):
We have channels that are now just more casual. People just talking about the news or what's going on, or creating virtual influencers.

Pablo Srugo (00:28:29):
And it's all avatars, so they'll create the script or whatever, or the audio, but then the idea is it's just two avatars talking to each other, like two AI-generated. Yeah, I mean, you can do anything in Hydra.

Michael Lingelbach (00:28:39):
You can make talking animals, babies, you can do general-purpose video. So you can do your B-roll. if you want to have a car shot, or moving backgrounds, waves, whatever. But I think that's what people us, is that they can kind of just get everything they need. So they don't have to use a bunch of different tools.

Pablo Srugo (00:28:56):
I'm really worried because listen like you've been listening for like what 10, 20, 30, minutes now. Clearly you like it and the thing is the next episode is way better and you're gonna miss it You're gonna miss it because you're not following the show. So take your phone out and hit that follow button. But so the main use case for Hedra on the podcast side is people who are like creating new podcasts through Hedra versus augmenting what they're already doing for the most part.

Michael Lingelbach (00:29:21):
Yeah, I think right now, that's more of a bandwidth of our team distinction. Rather than something we've intentionally done. Because a lot of value of feature historically has been  creating net new content. But we think there's a lot of potential in AI also just changing how we like edit and interact with existing content. So we're obviously working on a lot of stuff in the background that we have all this additional capital. And thinking about ways that we know these models can operate on existing videos definitely close to top of mind.

Pablo Srugo (00:29:51):
Here's a way to make it tangible. How could I use Hedra today? Is there a use case for me and PMF Show with Hedra today? What do you think it would be?

Michael Lingelbach (00:30:01):
Well, if you want to, for example. Make a daily reaction video, where you go over tech news. It's pretty easy to set up an automation with Hedra. Where you can pull from the X API what's going on, write your script, deploy a video with you and another host. And it could just be a generated version of you, right? A lot of people do, when they do themselves animated versions. Because seems less deceptive to people. You know, when you do a real person, there's definitely a lot of use cases for that. But usually, if you represent yourself having a fun 3D animated style, or an animated style. Is a good way of, making it clear that it's generated content, and it seems like that's something that people don't have the same aversion to. Actually, people think it's fun.

Pablo Srugo (00:30:47):
I like that, actually. What's the max length on a video right now?

Michael Lingelbach (00:30:52):
Technically, five minutes, but if it's unlimited. We just haven't gotten requests for more than five minutes, really. Because five minutes covers most podcasts when you do shot-reverse-shot or side-by-side. But yeah, the model, the one advantage of our technology and our architecture is our models support infinite recursion. So we don't have an issue with long video generation, like other providers.

Pablo Srugo (00:31:15):
Okay, that makes sense. Yeah, because I wonder, just the animated piece that you said, like, I was telling you, right? I mean, we're doing this remotely, I record all my stuff remotely, but if I could do it. If I could regenerate this, and it's us two. But we're animated versions of ourselves talking like in a studio. Visually, that's just probably better. But I mean, it's an hour long podcast, though. I don't know if I'd let tomorrow, but that'd be pretty interesting.

Michael Lingelbach (00:31:37):
Yeah, I think, and we have people that are doing that now. Because, a lot of people want to create localized content or a big tragedy. Not to get on my soapbox, but is that local news is. Been very hard to maintain in the era of consolidation and media. And so we have people now that create local news channels with anchors generated with Hedra. Which I think is kind of cool. Because, you know, news and civic engagement is so important, right? And I think, that's the power generated media is, stuff that maybe. Because of economic incentives or consolidation has gone away can now come back. Because it's just so cheap to make this content now. So I think that's a really interesting use case that we've seen pop up.

Pablo Srugo (00:32:21):
And then tell me about. When did you actually officially launch?

Michael Lingelbach (00:32:25):
We launched the first version of our platform last June. It was a very different UI. You can actually still see it, if you know the right URL. Because we still have pretty loyal users that really like the old app and the old model.

Pablo Srugo (00:32:39):
What was the difference? I mean, now it's basically text box and like ad script, ad frame sort of thing. Like what was the old UI?

Michael Lingelbach (00:32:44):
Yeah, so the new UI is very Figma-y. It has different modes, and you can switch models, and you can bring content from one mode to the other. The old UI was very much an avatar-only generator. Where it was our model, and it was just the avatar, and it was script always exposed. So it was much more fit to one type of video, and in some ways, that was really nice. Because it made the product very focused, but it meant that we were more of a point solution rather than a workflow tool. And so people would always talk about like, hey, I use teacher with X, teacher with Y, and we'd be like, oh, why did you do that? We have a really good image model, you can just use that, and they're like, oh, but it's in the avatar box, right? So I have to, it just doesn't make sense. So a lot of the motivation behind our V2 launch, which was like four months ago now. Was on how can we bring this together in a way. Where people have the tools they need to make this type of content. But it feel more like a workflow that you can kind of live in. Where you can just use Hedra for all your needs, rather than using Hedra as just a point solution. I think that's a lot of the reason why we saw this big inflection point, when we launched is. We kind of became one of these go-to tools and that's really also informing what I'm thinking about for what we're building next.

Pablo Srugo (00:34:00):
Walk me through the first launch and the second launch. Let's start with the first launch. So June 2024, how do you launch the product?

Michael Lingelbach (00:34:08):
Yeah, I mean, it really came together at the very last minute. So we had pivoted off of the tech approach and we had kind of been building this new model. And, like, Suno had come out at the time.  Pretty, like, it seemed a bit out, but it was trending. And I, also one thing to put in context is there wasn't really dialogue and generative video at this point. There kind of was, but it was just really bad, it was so bad. It was borderline unusable, and so one thing I had done is I had. 

Pablo Srugo (00:34:39):
So the general video was no audio? Is that what you're saying? 

Michael Lingelbach (00:34:42):
Yeah, there was lip sync on top of existing video, but it was like three seconds. The face wasn't coordinated. It just wasn't what you needed, and so we had this model that we were developing. And we didn't have a UI or anything. And I was playing around with Suno. And I was just making really funny songs. That's 90% of my generative music use case, is making comedy songs, and so I had this hilarious song that someone had made that was trending at the time. And I chucked it into a model with a picture. And I was like, wow, it can sing. That's actually pretty fun. Someone's going to want to use this. And then so I made the call that we should just really rapidly productionize a simple version of this and get it out there. And so we scrambled to put together a basic web UI, get the model deployed, and it was maybe a month and a half sprint. And we were also making the model better. We had shown our investors the model, the really early version of the model, and I think the response was like, oh, it looks okay. It's cool, but it's not there yet, and then I was like, all right. So let's parallel track getting the model good and getting the product ready, and hopefully it will come together at the same time. And that was, really, how it is the startups. Those super late nights, people were really grinding, and I think when we gotten that first UI. I lost like two hours just playing around with it making stuff and I was like, alright. I think people are gonna like this, and so then I was like, oh, but how do you actually do a launch?

Pablo Srugo (00:36:21):
Exactly, yeah. Yeah.

Michael Lingelbach (00:36:23):
I'm like, all right. Well, you get people to retweet it, right? That's the thing, and so I went about trying to figure out. I asked one of our angel investors like, hey, can you introduce me to some influencers, you think are interested in the space? Yeah, I took a four day sprint of basically doing 10 hours a day, onboarding influencers onto the platform. Which is really annoying at the time. Because we had no feature flags, nothing. It was very much, get like manually adding people into a database and teaching them how to use the tool. And then be like, Hey, we're gonna launch at this time, make some content, right? And we didn't pay them or anything.

Pablo Srugo (00:37:04):
What was the feedback, by the way? As people were using it, those early ones. What were they saying to you?

Michael Lingelbach (00:37:10):
I think people were really excited because they were like, wow, this is kind of something that we needed. Putting dialogue in video allowed people to tell stories. That was really important. People were like, oh, I could automate my channel with this. I could make a movie with this, and so people are finding more use cases than we anticipated. And so, you know, we launch and, the launch was pretty successful. We made like a, oh, hype videos. Hype videos, I just gave our interview to Business Insider about this. They're really hard to make, you need to find someone that's good. So I texted one of my friends, and I was like, hey, who's the best hype video person that can make this video? And we put it together in a 24-hour turnaround. And I think we got into getting 200,000, 300,000 impressions on it. Which is fine, nothing compared to what we do now.

Pablo Srugo (00:38:05):
Did you use Hedra to make these hype videos?

Michael Lingelbach (00:38:07):
We did, we generated the content in Hedra for the hype video, and then. Yeah, it was that plus all the influencers talking about us, plus a couple of news websites picked us up, and that really led to an explosion of growth. At the time, we had no monetization. Because we didn't have enough time to put monetization in. Which sucks in retrospect, we could have made so much money. At the time, we were just trying to get this out there. To be first, to get mind share.

Pablo Srugo (00:38:38):
How many users do you get? In that first month, for example? Order of magnitude.

Michael Lingelbach (00:38:43):
A million or something. 

Pablo Srugo (00:38:45):
Wow, wow.

Michael Lingelbach (00:38:47):
I want to say or like, I honestly forget at this point.

Pablo Srugo (00:38:52):
But we're talking those numbers. We're not talking 10,000 or 100,000. 

Michael Lingelbach (00:38:54):
No.

Pablo Srugo (00:38:55):
We're talking a million-ish.

Michael Lingelbach (00:38:56):
No, it was a lot of users, but it was a lot of free users, and I think I've talked about this. But free users are often a negative signal, in my opinion. The people who are willing to pay for software are sometimes a totally different group than, the people who are willing to use it for free. And so we basically were like, all right, this is a thing. Let's add a bunch of features and modification, and honestly, that was a mistake. We had something that was popular, but instead of me taking the time to think about why it was popular. I was like, let's just add functionality to it.

Pablo Srugo (00:39:28):
Functionality that people requested or just functionality you assume would be the right stuff?

Michael Lingelbach (00:39:33):
Functionality people requested and functionality that people thought would be the right stuff. But I think my problem is I hadn't identified that core high willingness to pay ICP of, the pool of users that were requesting stuff.

Pablo Srugo (00:39:45):
Who are you solving a problem for versus just, oh, the people here asked for this so I'll do it.

Michael Lingelbach (00:39:50):
Yeah, exactly! And we did, we were relatively data drive., We used a stack ranking, issue report system. So it's not like we were just flying blind. But I still didn't think like, I did the bottoms up thinking but I didn't do the tops down thinking of, what problem is this solving? We got something out that was solving a problem area, but we didn't get something out that was solving a full workflow, and that was, I think a false signal to me. And, we did get monetization on, it was still really successful. We grew to a million in revenue relatively quickly.

Pablo Srugo (00:40:23):
How fast?

Michael Lingelbach (00:40:25):
A few months, five months, six months.

Pablo Srugo (00:40:28):
Nice.

Michael Lingelbach (00:40:28):
Or something. I think for me, ultimately, I just felt like there was something missing. It didn't feel like that vision I had in my head of, like, solving this pro-content marketing, content creation problem.

Pablo Srugo (00:40:44):
From a product perspective you didn't feel satisfied? Or did you see something in the usage, or the growth patterns too?

Michael Lingelbach (00:40:51):
All of it, right? I looked at all of that. I interviewed hundreds of people. I did all of our customer support. I just was looking at it and I just felt like there was something missing. I think it was like, I went and I did sales meetings and I'd be like, oh, where is this, this and this? And I would talk to our users and see them using other tools with us. And I'd look at our model and I felt like our model wasn't able to do all the content people wanted it to do. And I just felt, honestly, it's really overwhelming, right? When you get all that feedback. I think, at that time, I kind of made this. There's something that snapped in me. Where I was like, made this call that we fundamentally have to build a V2. We need to change everything. I think I had a very clear vision of what that V2 was, that would get us to our next milestone. And that we just went heads down and worked on for life.

Pablo Srugo (00:41:45):
Was there a catalyst that drove you to that certainty that you had to build V2? Or was it after so many things. You don't know what it was, but the whole of it just gave you that confidence?

Michael Lingelbach (00:41:55):
Sometimes I just feel like you have to have gut intuitions and go with it when startups. I think a lot of it, I think there is a lot of data, right? But sometimes you just have to have a vision and conviction, and get people's alignment around that. And I think for me, part of it was thinking about us. I just felt like being fundamentally just a image to avatar workflow without anything else, just was not. It wasn't solving a fundamental need, it was solving a technology need. And I think what I wanted to do was, I felt like there was an insurmountable gap between solving the technology need and the workflow need. Without me just doing a lot more work. Does that make sense?

Pablo Srugo (00:42:36):
And what was? I think it does. What was the workflow need, that you then decided? Because as you mentioned before, you hadn't done that top down. So I'm assuming that you kind of did that top down thinking. What was the workflow need that then you landed on at that point?

Michael Lingelbach (00:42:48):
Yeah, I felt like what people really wanted is. They wanted the ability to have a common asset management layer, be able to bring in all this different type of media. Be able to generate the core image, audio, video workflow. Being able to have access different models not just our model. Being able to kind of grow to trust Hedra, as being that, you know, Figma-like experience. Where you can bring in, build all the stuff you need, and that's really what I wanted to get to. And I really wanted to, narrow in on this like prosumer or marketer profile. And that's, what I brought in, a bunch of designers. We have an amazing design team now. We kind of, did a whole revamp. We tried to build, one mistake I made. Which I'll admit is, we tried to build while designing. Which can be quite challenging. If you don't get everyone alignment and buy in on what an MVP looks like. I think sometimes, you know, it can cause a lot of burnout in engineers. If they're building something and then you're like, hey, actually no, we're changing this. Or actually, this isn't the MVP. Because there's this thing that's missing, and that was very hard. Because we were training a new model, we were trying to design this, we were building, and the engineers were moving goal posts. On what was the MVP for the launch. So it was really messy.

Pablo Srugo (00:44:11):
And you were still 10 people or so at the time?

Michael Lingelbach (00:44:13):
Yeah, we were still pretty small. So a lot was falling on a few people.

Pablo Srugo (00:44:18):
How do you get the alignment? Once you decide you want to do this V2. The rest of the team and even the investors, was that relatively easy?

Michael Lingelbach (00:44:27):
I think our investors were very happy with what I had done. Because with relatively little resources, $7 million is not a little, but Sora had come out at this point and people were really excited to use Hedra. So we had stood out, and we did something other people weren't doing. I think I've built a really solid rapport with all of our investors and I love everyone that we work with. And I think the fact that we had done something unique and I had the vision for doing something unique again. And I think that really brought a lot of people in. Team-wise, it wasn't always like that. You know, I did have a co-founder and my co-founder left kind of in that interim period. We just agreed, you know, he wanted to pursue other things. He's really successful now, working in robotic space, and I really wanted to just keep pursuing this vision.

Pablo Srugo (00:45:23):
I don't know how much you can share, but I mean co-founder disputes are very seldom talked about for obvious reasons. So even if you can't give the why it happened. Even just the fallout and the outcome of all that. The more you can share, the better on, you know, the stuff around and the stuff that's not proprietary.

Michael Lingelbach (00:45:45):
I think people should talk about it more. I think people should talk about when co-founders stop working together. They should talk about what it's like to fire your first employee. Because this honestly happens to most people and if people don't talk about it. It's just going to, it's gonna mean that you're not equipped to deal with it when it happens to you, right? My co-founder and I are still friends. We're on good terms, we talk. I think we just want different things from the company. I really had this vision of pursuing what became Hedra Studio, and he wanted to go work on robotics. And I think if a person's heart isn't in the direction the company is going, they shouldn't be there, right? And we didn't, I think it's not like. I don't have any really juicy stories here.  It was just like. 

Pablo Srugo (00:46:34):
No, I mean. I guess my question is more why? Maybe just a baseline question. If you want to do robotics, I mean, this is never going to be robotics. Why do you think he even started it with you in the first place?

Michael Lingelbach (00:46:44):
Yeah, I think that what happened is that when we started out the company. We had a vision and we thought that what we were going to build earlier was probably going to be what we built later. If that makes sense, and there was a period. Where it maybe felt really far away to getting to what became Peter's studio, and Character 3. And so, I think in that time period, if you're just not. If you don't see that, and you're not really excited about that. And you don't feel like that's going to happen. There are other opportunities, right?

Pablo Srugo (00:47:18):
And then what was the fallout, in your startup when that happens? Was it kind of? I mean, it's usually a big deal. Did it impact other things? Or was it smooth?

Michael Lingelbach (00:47:28):
Oh, yeah, I mean. This was, it was definitely a big deal at the time, right? And I think, I talked to it with the team. And some people left, right? Some people did leave, and then I kind of just like rebuilt the team. I brought more people in and I kind of just had to align people. And it is a lot of faith to believe a founder when they're like, hey, I think this thing is going to be really successful. Just stick with me and we'll make it happen. Right.

Pablo Srugo (00:47:54):
What about yourself? Emotionally, were you fine?

Michael Lingelbach (00:47:57):
Yeah, I mean, I think I've always been. I think one of the biggest things about being a founder is just having that, emotional resilience and grit. You just have to be able to get through anything, right? And I think that for me, I, like a lot. I did do a lot of inward reflection, right? And I, you know, you always have these crisis of faith. Which is like, is this gonna work? You know, and you just have to push through that, right? You just have to be like, you have to figure it out, right? It's not, is this model going to work? It's, how are we going to make this model work? It's not, you know, is this product going to resonate with people? It's like, figure out what to make it resonate with people and who your ICP is, and just be able to recite that in your sleep. And know that person really, really well. And I think what happened is I just became really fired up to make it work. Whatever the cost to myself was, and I just tried to get everyone in alignment on that. And I'm really fortunate that it worked out as well as it did. But there's a world in which I wasn't able to do that, right.? Yeah, so that was the pre-series A. Pre-Hedra Studio time was, we launched a product, we monetized, we got some pretty good revenue. Not rocket ship revenue and then went through a team change, and then locked in. and had to go build our next milestone.

Pablo Srugo (00:49:22):
And are you kind of flatlining on revenue? As you're built, because you've got so many things going on. Like, co-founders leaving, you're building model, you're rebuilding B2, designing, all this stuff. Are you still growing during that time? Or are you kind of plateaued?

Michael Lingelbach (00:49:33):
We're still growing, but we weren't growing at the rate that we were. Because we stopped shipping new features. So, that was the kind of hard call I had to make. Is I had to say like, it's worth sacrificing a little bit of growth now, to build something that's going to grow a lot more in the future. And, you know, that's oftentimes a really controversial decision, right?

Pablo Srugo (00:49:55):
I think it is very controversial. Especially I would say like, at the outset. You're like, this is what we're gonna do if you were like, okay. Yeah, that makes sense, and then you go a few months, not really growing. And all of a sudden it's like, are we sure about this?

Michael Lingelbach (00:50:09):
Yeah, exactly. If you're looking at something like a next version of a product and it's totally broken. You're like, oh my god, this product we have that's working is languishing. And this thing that's super broken is like, when is it even going to launch, right? But ultimately, I think you have to do that. Because I think one of the traps that people get in is when something works, but it doesn't work well enough. One person once told me, if you are asking whether or not you have product market fit, you don't have product market fit, right?

Speak (00:50:39):
That's right.

Michael Lingelbach (00:50:41):
And so for me, I was like, all right. We don't have product market fit, right? Because I'm asking the question, and so that's what made me do that deep reflection, and deep reanalysis of the product. Building out that version two, building out that next model. Just because I was like, we need to build to the next milestone.

Pablo Srugo (00:50:59):
And so what was the second launch like? What did you do for that one?

Michael Lingelbach (00:51:03):
That was a lot more organized, because at the time we brought in our first product marketing hire and she's awesome. She's definitely taken a lot off my plate in terms of putting together the video and co-ordination. I still am very deeply involved, but it was pretty high stakes, right? Because we hadn't done a launch in a while. It was the first thing we put out. It was a totally new platform, right? Basically built from the ground up. Same core functionality.

Pablo Srugo (00:51:29):
Did you have a long runway at the time? Or what did cash look like?

Michael Lingelbach (00:51:33):
I have flexed up burn a lot, to be able to afford the trial. The product, had the launch not gone well, we would have had. I think, six months from, or more than that.

Pablo Srugo (00:51:47):
Yeah, okay. So along the line is the setup.

Michael Lingelbach (00:51:49):
Yeah, it was like the company would have been fine. Our burn would have come down from the training costs of the model and it would have been longer. But it was still pretty much, we should make this an impactful launch. Because it would be bad if we did not make it an impactful launch. So we put people on the model. People got really excited, because it was another step. It was a much bigger change in capabilities than people were expecting us to be able to do, and it was the first model of its kind. The studio product was much more full-featured. The UI was really clean, it's still pretty janky, it turned out. Because it turns out when you scale to millions of users. Things break, who knew? But, yeah, so we, put that out and we got. You know, really quickly, like mass virality, and I think not only that. Huge amounts like company inbound and the day after we launched.

Pablo Srugo (00:51:49):
Was it similar though, the influencer strategy? 

Michael Lingelbach (00:51:53):
We didn't. I think one difference of us is, we haven't really done as much paid influencer partnerships. A lot of these viral videos you see on X. There's a playbook to doing it. There's like a couple hundred influencers you hit up and you pay some money, and they retweet your thing. And we didn't really do that for a first launch. We didn't do it for that launch. We really just, I mean. I don't actually know exactly all the people we talked to. Maybe some of them, we paid to try out the software. But in general, our position has been just make something good and then put people on it. And they should talk about it if it's good. And if it's not good, don't launch it. So yeah, we did the same thing. Where we put people on it early and gave them early access, and let them make content. And that did really well.

Pablo Srugo (00:53:39):
And what was the? Like, you said, the first one, you got a million users. Million ARR within like five months. What was the outcome of this B2?

Michael Lingelbach (00:53:46):
It was a lot. So we pretty quickly hit like an eight figure run rate. Which was exciting.

Pablo Srugo (00:53:53):
Wow. That's awesome.

Michael Lingelbach (00:53:54):
And that's been really exciting too, is also seeing the enterprise adoption. Like, you know, we keep. I think we signed an enterprise contract every couple of days now, pretty big ones now too, and that's without outbound sales. That's the other crazy thing is we haven't really built it. People are like, oh, what's your go-to-market team? But now it's a little bit different. Because we're billing out, we're hiring our first AEs, we're hiring more product marketing.

Pablo Srugo (00:54:23):
How many people are you today?

Michael Lingelbach (00:54:25):
We're 23, I think, as of this week. So we're still pretty small but we generally have tried to keep the team lean, and now we're thinking a lot about. As we scale and we get ready for our next launches. What does the team structure look and how do we like? How do we start building an org back to scale? Because I've historically been a really hands-on founder, and I think we want to keep that. But also now, I can't do first-round screener calls for a lot of candidates anymore. Because there's too many people that are applying. So we have a head of talent now, and, you know, I can't review all the copy that goes into the website for everything. So, that's why we're bringing on product marketing leader, and same with, you know, for go to market. I can't do every sales call, there are just too many calls now. So that's why we're hiring out AEs, and that has definitely been a transition going from a series, like a C company to a series A, and now beyond kind of company. Is how do I scale my time and still make sure that the org maintains excellence. And fortunately, we just hired out amazing people. I often will wake up and work that I was worried about doing the next day is just done, and that's kind of a crazy nice feeling.

Pablo Srugo (00:55:46):
How did the Series A come about?

Michael Lingelbach (00:55:48):
Yeah, so we've worked with a16z for a long time, since our seed round. And I think one of the differences about A16Z from a lot of other funds, is just how big the support infrastructure is for founders. What I'll say is they're not going to build your business for you, right? They're not going to fundamentally replace your role as a founder. But it's kind of like just, you have access to so many resources. Between the talent team, the go-to-market team, just crazy opportunities if you know where to look and ask. And a16z is just there to make you succeed. We've had a really good relationship with them. They were there for us through the whole process. I think after they saw the model and the launch. Because I only told them about it a couple days before. Because I wasn't sure if it was going to work. They saw the initial early traction and we chatted. And we're like, let's just. How do we figure out a way to continue this partnership, so that we can just get back to building? So, I didn't really go out and try to raise money from anyone else. We just kind of came to an agreement with our existing investors, and I've wanted to work with Matt for a really long time. Borenstein, who's on our board. He's one of my favorite investors I've ever met. He is hype, he just really understands deeply AI. He's been doing AI investing even before it was hype-y and just totally understands the different value propositions, and positions of companies. And the core technology differentiation. And then Martin Casado, the GP, and Justine Moore have just been such amazing resources from the a16z Infra team. Even before they led our Series A. So it just was like, it was always what I wanted. And, it would just happen to be the business got to the point where, it made sense to do a Series A. And that was actually pretty early. We had, I think our revenue only tripled at that point. It was just really exciting to see everything come together.

Pablo Srugo (00:57:46):
All right, man. Well, listen, let's stop there. I'll ask the last few questions that we always end on. The first one is, you mentioned with V1, you were not sure that you had product market fit. Which means you didn't have product market fit. What about now? When did you first feel like you had true product market fit?

Michael Lingelbach (00:57:59):
I mean, the first couple of days after launch. We definitely got really impressive numbers. It was like, wow, this is so much more money, and then I was like, all right, this could be novelty. People would just be checking it out. What's retention going to look like? And then it was like month one. And I'm like, okay, that was cool, people are retaining. But maybe it's still just a fad. Other models are going to come out and then it was like month two. And then I'm like, all right, okay the numbers keep going up this is great. Maybe we have product market fit and I just kept having this like natural skepticism. Because of like, all right there's gonna be another shoe dropping, right? And I think that, it was kind of like. I kept wanting to find a reason why there was a problem with things and the business kept doing better. And so that was really exciting and validating. Because I had gone through this dark period a little bit. Where I was questioning myself last year between these big launches. Yeah, so I think it was probably month, two or three after launch. Where I was like, okay, I think we've solved a core need, and now it's about. It's not that we're not going to do big crazy things again, but we just have something that works now. And that just makes so many things easier. Everything from recruiting to fundraising to, you know, thinking how you're thinking. Talking to people on the team about what the position in the company is. And I think that's just done. It's just made building a high growth company really easy. So yeah, I think it was probably like month two after launch.

Pablo Srugo (00:59:45):
And then you've touched on this a little bit, but what's one moment you remember specifically where you thought. Things might not work out, things might just fail?

Michael Lingelbach (00:59:56):
Yeah, I mean, I think there were a couple times. So right before I raised our first round, I had no idea if we were in a term sheet, right? I mean, you're a PhD student. you've never built anything beyond papers in your life. Asking people to trust you with $7 million to go build something feels like a lot. I know there's a survivorship bias, because you see these people raise money, but there's so many people who don't raise money. I could just not make any money and go back to school. And then go live a really successful life, going doing research, like Google or Nvidia. But I feel like I missed out on this opportunity to build something. So that was one time. I think, you know, kind of in this like after the first. Actually I would say there was the time when we did the first pivot. I felt like when we had to kind of redo our whole tech stack and that felt to me like, oh, I just wasted so much time. I wish I just had this idea earlier. I wish I'd bet on this approach before, and that made me feel like a really bad founder. Which in retrospect, you talk to so many people and you realize this happens to every single person. And people tell you this. It's not like people don't admit this. They're like, oh yeah, we pivoted so many times, and you hear that. And then when you do it, you still feel bad, right?

Pablo Srugo (01:01:28):
I think you just put so much weight on your shoulders, as the founder. It's almost inevitable.

Michael Lingelbach (01:01:33):
Yeah, because, you know and that's honestly, it's good, it's important. Everything is on you, right? But at the same time, I think admitting, when you're wrong or like, not even when you're wrong. But like when you didn't predict something, it's just so hard to always predict something. And there's a lot of luck and timing involved. That you have to, just roll with the punches. And, do what the best thing is for the company at any given time. You know doing that, going that first sort of like technology pivot and making the bet on the first version of Hedra was definitely a time. And I think in that time between, the first launch and building out these features. And realize that we needed to do deeper product and model rework to really reach that next milestone. As well as, you know, when my fat like co-founder left. That was another time where I was like, okay, I have to go bark on this like, really big quest. And there's going to be a huge. I think there's gonna be a huge payout at the end of the road, but still seems really far away. And I think, what I'd say to, you know, people listening to this other founders. It's, you just have to get to that next thing, right? You just have to have that faith. You have to think really hard if it's the right thing, right? Don't just have blind faith without conviction. That you've got through research and deep understanding of your ICP. And of your product and of the market. But when you have that, you know, people are gonna look to you, right? To be inspired, to be able to go along with you, and if you're, like, outwardly questioning. If you're outwardly not having that strength to bring people along with you. It's hard to push through those times. And I'm just really thankful that I had amazing investors that supported me along the way, amazing advisors. We had such a strong core team at Hedra. That I've always just felt like I had. Sure, I was maybe doing a lot of the pushing and the trailblazing, but also there were a lot of people right alongside me that were doing it too.

Pablo Srugo (01:03:25):
And then one last question is, what would be one piece of advice you'd give an early stage founder?

Michael Lingelbach (01:03:31):
I think that it's really easy to come up with ideas that sound good, but that people are not willing to pay for. And I think I've said this a lot that free users can be really false signal. That just because someone uses it for free doesn't mean it's something that solves a core pain point. I think really understanding what your value proposition is and putting up paywalls early, and trying to solve a problem. I'd rather solve a problem that a few people cared a lot about, and we're willing to pay like a really high amount for. Like, is this something that people will pay $20 bucks a month for at minimum, right? Like a ChatGPT subscription, right? Or is this something that people are like, oh, that's cool. But you throw up a $5 paywall, and they won't. You want to find that big problem that we can solve for passionate people. Because getting that wedge in that core retention and adoption is really important. Because you can always expand, right? You can broaden division. Like you said, Granola is just like a recorder, right? Why is it so popular amongst founders? It's because, it's actually really annoying to always remember to bring up my note taker and get consent, and all that stuff. And Granola has made it a one-click workflow. Where I can play the consent message or record it and get my notes, and write my things. And that solved the pain point for me. To the point where like, yeah, I'll pay $10, $20 bucks a month for it. And I feel like the quicker you as a founder can understand the market, understand what's missing, understand a person. To the point where you're like, okay, I can build something that person's really gonna care about and I can do it better than anyone else. The quicker you'll be able to get a successful business, and there's just no barrier anymore to making a good MVP. Software as a service has never been easier. There's so many AI coding tools. There's so much engineering talent that's hungry. You really just need to give people clear vision and build to execute on that well. So that's my advice.

Pablo Srugo (01:05:28):
Perfect, well, Michael. Thanks so much for jumping on the show, man.

Michael Lingelbach (01:05:30):
Yeah, thanks so much for having me.

Pablo Srugo (01:05:32):
Wow, what an episode. You're probably in awe. You're in absolute shock. You're like, that helped me so much. So guess what? Now it's your turn to help someone else. Share the episode in the WhatsApp group you have with founders. Share it on that Slack channel. Send it to your founder friends and help them out. Trust me, they will love you for it.