Using A.I. to Become a Superpowered Indie Hacker with Przemek Chojecki of Contentyze

Courtland Allen: 7s

What’s up everybody? This is Courtland from IndieHackers.com, and you’re listening to the Indie Hackers podcast. On this show, I talk to the founders of profitable internet businesses and I try to get a sense of what it’s like to be in their shoes. How did they get to where they are today?

How do they make decisions both at their companies and in their personal lives, and what, exactly, makes their businesses tick? And the goal here, as always, is so that the rest of us can learn from their examples and go on to build our own profitable internet businesses.

One of the things that makes this show possible is that we’re in the middle of an explosion in the power of the individual. You can set out on your own, build something cool, get rich, and change your life, quite frankly, without needing a whole bunch of other people to help you out or a whole bunch of investors to invest in what you’re doing, because there are so many tools now that give you more leverage than ever.

You can do this without even knowing how to code, and if you do know how to code and you know where to look and which tools to use, you have insane amounts of power and leverage. My guest today, Przemek Chojecki, is a great example of this.

He’s embracing artificial intelligence in ways that I don’t think most people realize are possible for the solo indie hacker. Hearing your story gets me excited about how much you can get done as a solo founder and what the future holds for indie hackers.

(end of intro) Do you think AI is overhyped, under-hyped, or it’s at the right level of hype?

Przemek Chojecki: 1m 24s

I think it’s both in the sense that it’s overhyped because it’s still not doing everything it’s supposed to do. It’s presented in the media right now, especially if you look at popular media, but it can definitely do much more than we think in general that it can do.

So I think AI is still in its early days. It only started 12 years ago, maybe 8 years ago in 2012 and some competitions around computer vision. Then it’s gaining this momentum where we have better hardware, more data. So it’s not hyped enough for what it will be able to do in a couple of years.

Courtland Allen: 2m 4s

I figured you would say it was under-hyped, because you are super into AI and you’re doing some cool stuff. From my perspective, AI has always been a big company game. When I think AI, I think Google, I think Facebook opening up these expensive research labs and hiring PhD’s, and spending tens of millions of dollars to collect terabytes of data to train their models.

And then here you are, a solo founder indie hacker. You’re not spending a ton of money. You’re not taking a ton of time, and so far you’ve been able to use AI to your advantage in a way that’s good. So briefly, when did you start your company, Contentyze, and how much revenue is it generating?

Przemek Chojecki: 2m 41s

I started Contentyze in January 2020, so this year, nine months ago, and I’m generating $4,000.00 to $5,000.00 a month. Most of that is still coming from the content we generate.

That’s a funny model, because we generate so much content on our platform that we are going to monetize some of that through advertisement, sponsorships, affiliate links. And some of the revenue is coming from paying customers in the platform, but that’s still a minority. I want to flip that and have more of the income, more of the revenue being generated by the SaaS platform itself.

Courtland Allen: 3m 22s

What’s cool about making five grand a month is that, as you said, you just started this year, so that’s good, and you’re now doing this from Warsaw, Poland, I believe. I looked it up this morning. Your cost of living there is three times cheaper than it would be in a place like New York City or San Francisco, which is the cool thing about being an indie hacker.

You can live somewhere cheap. You’re doing business online. You’re earning dollars as if you’re living in the richest place on earth. So I imagine you’ve got to feel good about how things have gone so far.

Przemek Chojecki: 3m 49s

Yes, definitely. It’s very interesting, also, for indie hackers from that perspective, because I think it’s much easier to start working for someone remotely, and there’s an expectation that you work remotely because we’re in the pandemic, maybe there’s a second wave. Maybe there’s not. But in general, probably things won’t go to normal and that’s a good bet for the near future, because there’s no vaccine in sight.

Looking at the statistics in virus places, it’s probably going to be much more spread. So taking that into account, building a team anywhere you live is very doable. You can hire people from around the world.

Your hiring costs are coming down, because you can look for people in places where you have more potential bang power with your dollar, like Warsaw, but also going to the east, like Ukraine, or going even more east like Russia, China anything like that. So there is a lot of potential for that right now, growing your digital online businesses.

Courtland Allen: 4m 52s

This is what I look for when I’m looking for people to interview, when I’m looking for things to talk about. What are the new trends, the new platforms, the new technology that’s enabling solo founders and individuals to do much more than they ever could before?

Nowadays, as you’re saying, it’s easy to use all these tools to build a business no matter where you live. You don’t need to live in Silicon Valley. You don’t need access to a ton of investors. You don’t need expensive tech. You could do this from any place on earth. You could be a digital nomad living in Bali, charging people hundreds of dollars a month for your SaaS product.

I think the other thing you’re doing that’s particularly high leverage, that fewer people have caught on to, besides the remote working thing, is using AI to your advantage. People are sleeping on AI, I think, because it seems so inaccessible and so expensive, but it turns out that’s not quite true anymore. Maybe it won’t be true in a few years.

You’ve done some cool stuff that I think we’re going to get into in more detail, but just briefly to give people an idea about some of the cool things you’re doing with AI, number one, you’ve got this company, PetaCrunch. It’s an AI media company. You’re interviewing founders. How big is the team there at PetaCrunch, and how many founders are you interviewing, exactly?

Przemek Chojecki: 6m 3s

Basically the team is one person and that’s me. The rest of the team is AI generated. They’re AI journalists and they’re fully scripted, contacting the founders on behalf of themselves. Basically, they managed to interview over 1,000 founders in three months. So it’s a huge volume of interviews done over email with founders from around the world.

Courtland Allen: 6m 29s

It’s crazy.

Przemek Chojecki: 6m 30s

Yes. That’s an exciting part. I started doing PetaCrunch even before Contentyze, and then Contentyze came as the next level. I’m doing the platform which can help you leverage content ideas.

PetaCrunch still working. You can still go to petacrunch.com. Right now it’s working as a side business for me, because I’m no longer actively using those algorithms to scout for the new startups to take them on, but rather you can still buy an interview, buy a press release from PetaCrunch and have this publicity there.

But Yes, PetaCrunch was the starting point for me to go into Contentyze and build the whole SaaS platform to generate content at scale. I love writing. I love automating. I love optimizing. So that was my dream come true.

Courtland Allen: 7m 18s

It’s that perfect combination of skills for what you’re doing. You’re a writer. You are a productivity nerd. You’re into AI, and now you’re doing AI generated content.

You’re putting me to shame, because I’ve been interviewing founders for five years, four years at Indie Hackers and I’ve done 500 interviews on the website, and you did a thousand interviews with your AI journalists in just three months. So we’re going to talk about that. We have to.

Now your current business, Contentyze, is also similarly impressive. How many writers do you have employed at Contentyze, and how many articles and distinct pieces of content are you putting out every month?

Przemek Chojecki: 7m 48s

That’s a great question. I have no writers, or almost none, because I sometimes take contractors to generate articles for me, just for comparison. So I have only developers and those people are helping me write up to 100,000 articles per month at the current rate.

Courtland Allen: 8m 6s

What?

Przemek Chojecki: 8m 7s

So it’s something absurd.

Courtland Allen: 8m 9s

A hundred thousand articles a month.

Przemek Chojecki: 8m 10s

Yes.

Courtland Allen: 8m 10s

I thought you were going to say a thousand or something. OK, so mind blowing stuff. You’ve got no writers on staff. We’re going to get into this. We’re going to get into how you do this. But maybe the best place for us to start is back where you got your start. I believe you were getting your PhD. You were in academia. And that’s where you discovered AI for the first time.

Przemek Chojecki: 8m 27s

Right. So I was doing a PhD in mathematics. That wasn’t that far from artificial intelligence. Then I went to Oxford for a research fellowship, and at some point I decided that, OK, I want to quit academia and go into business and change from pure mathematics to machine learning, basically because I was fed up with academia at that point.

I wanted to have more impact on the world, be able to have something tangible, have some faster feedback loop on what I’m doing, not having to wait years to see the outcome of the research that I’m doing, but be able to do something and then within weeks, within months, or even faster sometimes, be able to see what I created and how well it’s performing.

So that was my thinking when it comes to my collector (ph), I mean, collected (ph) logically. I think being an entrepreneur is a cool thing. Going from mathematics to machine learning was also natural, because I love automating and optimizing stuff. So being this productivity nerd, I wanted to do something more in that domain and coding was natural in that respect.

So that was a couple of years ago. I went from academia, and I started starting my first startups. Both of them failed. I had two startups prior to Contentyze, prior to PetaCrunch, and they failed for different reasons, but both were a great way to learn about being an entrepreneur.

Courtland Allen: 9m 58s

Most people, I think, find it very difficult when they’ve got a lot of momentum and skills going in one direction and it’s safe and they understand it, to just throw all that in the trash can and say, “You know what, I’m going to be a completely different thing.”

Going from academia, where you’re working on your PhD, you’ve got other friends I can only assume who are also working on their PhD. You’ve got the social circle. Did people think you were crazy to leave and try to start a company?

Przemek Chojecki: 10m 23s

Oh Yes, totally. I think it’s super hard, because especially academia is such a bubble that if you go out of academia, it means that you’re some kind of a loser for not getting another job in academia. It’s a bubble, so it’s hard to go out.

For the first two years after going and deciding that I want to go out of academia, and I was battling my thoughts about whether I’m a loser or not for going out and whether I should stay, but I was hard convinced that staying in academia would make me miserable in the long run, that I had to quit.

Still, that was a continuous process because it wasn’t like one day I go, “That’s it, stop,” and I quit, but it was like, OK, maybe let’s do less research and try to find a consulting gig, try to get into business.

So this process of switching from academia to business took me three years. Right now I’m fully an entrepreneur, 100%, but that took me literally three years after deciding that I wanted to quit. So it’s not easy, and only right now I don’t feel like a loser being an entrepreneur and choosing that path, but back then, I was insecure about switching from this well-defined career that I had and I could have going for decades.

You have this very narrow path when you do research, become a post doc, become a professor at a university, and then go on, teach others. Then you’re 60, 70 you have a tenure track job. And that’s it.

Super interesting intellectually, because you keep on thinking about very interesting problems, but in the end, your career is very linear. You can’t do crazy things. I wanted something more. That wasn’t that appealing to me. I was doing that for 30 years of my life, and then I decided that, OK. Let’s try something different.

Courtland Allen: 12m 22s

Yes, and you can certainly do lots of crazy, innovative things as a tech founder. But you hopped from one bubble into another, because tech is its own bubble. What did you do to come up with the idea for your first company? You were out on your own. You’d never started a company before. What are the first steps that you take?

Przemek Chojecki: 12m 37s

I was doing those different consulting gigs, trying to help others with AI, and at some point, one of my colleagues, over drinks, he offered me a stake in his company. It wasn’t formed at the time. He wanted to have someone from a more technical background, him being a businessman before, to do something together.

So what we did was we formed a company, and that was the story of Bring.ai. And our goal was to build the last mile delivery platform, something like Postmates or Global, to deliver products within a city within 90 minutes.

So you could buy something online and we could deliver that for you, or for example, take your envelope from one place to another, deliver your keys that you forgot from work, something like that.

We underestimated the costs involved in the startup. For both of us that was our first technological startup. And we miscalculated how much we should pay drivers, how much should go to marketing. Those kinds of business which are Uber-like are very capital intensive. We weren’t prepared for that and we were bootstrapping everything. So we couldn’t make it, even though the growth was nice.

Courtland Allen: 14m 0s

How long did it take you to figure out that this was too expensive, and it wasn’t going to work?

Przemek Chojecki: 14m 4s

Basically a year. After one year we were running out of cash and we decided to pack things and that’s it. We were on the slope. We had a couple of weeks of runway, but we didn’t have any investors. We were not advanced talking to investors, and we decided that we didn’t know whether we can make it work. We decided that, OK, that’s it.

Courtland Allen: 14m 29s

How did that feel? You had all this pressure on your shoulders having just quit your job, your research path in academia, and now your first company fails after a year you have not much to show for it.

Przemek Chojecki: 14m 39s

It felt bad in the beginning, but after we decided, it was like taking the weight off your chest in a sense, because first of all we had the experience. It was a great learning curve that we had as a startup.

I wasn’t expecting much. I was expecting that to be like going to business school. I don’t like being able to learn and entrepreneurship at the university, and you have to become an entrepreneur to learn that. So I was looking at this experience as a forming experience of becoming a founder, becoming a CTO, CEO, and that was that. I wasn’t that unhappy about that happening. I was sad for some time.

Courtland Allen: 15m 29s

I think one of the interesting things about your first business is that until you start something, you have no idea what it’s going to be like. You can read all the stories in magazines or listen to podcasts about founders and come away with almost any impression.

You might think, “Oh, I’ve got to be an expert. I could never do this.” You might think, “Oh, it’s so easy. I’m going to knock it out of the park on my first business,” But until you try, all you have are thoughts and hypotheses about what it’s going to be like. You don’t really know. What would you say surprised you the most about being a founder for the first time?

Przemek Chojecki: 15m 58s

That it’s not about brainstorming new ideas. You have to get your hands dirty. For example, either start driving around. You don’t have customer support in the beginning, so you have to respond to every single mail that’s coming your way.

There are many different layers of problems that appearing from customers. For example, if something doesn’t work on the platform, it’s your responsibility to fix it as soon as possible, especially as clients are paying for the experience for the platform.

So there were head-on meetings with customers, customer service, and building the whole customer experience. That was a struggle in the beginning, and I think that’s the hardest part in building any customer-facing application.

Courtland Allen: 16m 50s

OK. So here you are. You’ve learned a few lessons. You’ve got some experience under your belt. You’re sad about your company not succeeding, but you’re not down and out, obviously, because you started a second company after that. Tell me about that process. How did you come up with the idea, and how did you get started?

Przemek Chojecki: 17m 4s

The second one was Bohr Technology, and the idea was that we used quantum computing to optimize processing, logistics, and transportation.

Courtland Allen: 17m 13s

Very ambitious.

Przemek Chojecki: 17m 15s

Yes. So that was super deep deck. That was fun. We did the company with a different colleague of mine in Toronto who created the structural app. That’s a well-known accelerator in Canada. It’s also the only one with this quantum computing stuff, so we were happy to be there. We spent a couple of months in Toronto.

That was going good. We secured pre-seed funding. But there were two problems back then. We actually brainstormed the idea. Before, my cofounder was as lawyer. He already had one small contract with one of the companies in Poland about developing something in the space. So it was natural, in a way, that we started brainstorming and there was this idea for the company that we could create something in the quantum computing space.

To some extent I was also too naïve about the technical difficulty of the problem, so I said yes very quickly to the whole thing, because I wanted to do different things. But it was very challenging later on.

I had this PhD in mathematics, but still learning about quantum physics and how quantum computing works, that was challenging. I was happy that in the end I learned a lot about quantum physics and quantum computing.

The idea came naturally. We moved to Toronto at some point, but there was this problem that first of all, we were super early to the market and it was hard to convince an enterprise to buy our solution and trust us to find something like a solution. So the selling process was going very slow.

On the other hand, we started having many differences in how we wanted to operate, how we wanted to go with the company with my cofounder. So those were the two main reasons that we decided to split. There was not enough of a track record to progress, so it was a natural decision that, “OK. We should stop here. We can’t make much more of it right now.”

Courtland Allen: 19m 17s

I’m curious about this product, because you’re doing quantum computing and I think practical quantum computers aren’t even a thing yet. So what was it that you were selling to companies or trying to sell to companies?

Przemek Chojecki: 19m 29s

There are no quantum computers which are better than classical computers yet, but you can already experiment. So our sales pitch was that you should prepare for quantum computers right now and experiment with them and build some applications on top of them, which you can do right now because later on when the real quantum computers come in, you’ll be left out of the game because you’ll be too late already.

You won’t be able to acquire talent as well, because this is a small niche and there aren’t so many talented players being able, for example, to build quantum machine learning algorithms for European vacation. These things exist and that was our sales pitch. We were selling to R&D departments of large corporations and that was a hard sell, especially in the beginning.

Courtland Allen: 20m 18s

Yes, I bet.

Przemek Chojecki: 20m 20s

That was B2B but B2enterprise.

Courtland Allen: 20m 25s

So you were having trouble selling this prototypical quantum computer. You started having cofounder issues, which a lot of people don’t realize but are often the number one reason why companies die, because your cofounders don’t see eye to eye. You disagree and you try to go in different directions which obviously doesn’t work when you’re a fledgling company that’s having trouble with sales.

Usually, when a company’s doing well and you’re crushing it and making a ton of money, you and your cofounders get along splendidly because it’s obvious what you should be doing. But when there are cracks in the surface and your product isn’t selling the way that you wanted and it’s harder to build than you thought it would be and it’s taking longer and you’re running out of money, I think that’s when the strife starts to show up.

Przemek Chojecki: 21m 1s

Yes, it’s a real test of your relationship with your cofounder. So we didn’t pass this test, this relationship test, with my cofounder. It’s one of the main reasons. The second main reason is probably lack of capital and underestimating costs, so I went through both of them with my first startup and then the second one.

Finding a good cofounder is a crucial thing, if you want to have one. So that’s a funny thing. For example right now I’m raising seed rounds for Contentyze. I’m talking with a lot of investors. I had a couple of calls when one of the investors asked me, “OK. Do you want to cofounder? Do you need someone from more of a sales background?”

And at this stage I’m saying, “No, I’m fine soloing the whole thing, because it’s more interesting from my perspective and I think I can do it.” If there’s any lack of skills that I will have on board and I’m happy to hire someone and I could give them a small portion of equity, but not on the cofounder level.

There’s still this thinking that you should have a cofounder and the best startups are the startups run by two people or more, but maybe it’s not true. I think the statistics are that both are equally successful in the end. You have solo founders which are very successful.

Courtland Allen: 22m 24s

Yes, my hierarchy is, if you can have a cofounder and you know that you and this person have complimentary skills and a history of being able to get through conflict, then that is the best situation.

Right below that, rather than having a cofounder you’ve never worked with before and you’re not sure how you can handle challenges, it’s better to just be a solo founder. If that means you’ve got to shrink the scope of your ambitions, you should shrink the scope.

But I think nowadays it’s easier and higher leverage than ever to be a solo founder, because there are so many tools and things are so cheap. It’s not as much of a disadvantage as it was back when people were first starting to write about this stuff 10, 20 years ago.

And then below that, all the way at the bottom, is working with a cofounder just because you heard you should, but you guys have never worked together, you’re not tried and true. You haven’t tested your relationship, and your company’s going to be the first and true test. Often that doesn’t work out very well.

This is in the arc of my career as a founder as well. I always started things with cofounders. Sometimes it went well. Sometimes it didn’t. But with Indie Hackers I worked by myself, and later on I brought on my brother as a late cofounder.

I know we have years of conflict resolution because we’re twins and we fought all the time growing up. So if we disagree about something, we know we can resolve it cause we’ve done it a million times before. I’ve seen the same pattern with a lot of other companies, like Stripe, founded by two brothers. I don’t think it’s a coincidence.

Przemek Chojecki: 23m 41s

I have a brother, and from my perspective it’s also scary to some extent. I trust him and I think he’s an extremely talented person, but on the other hand I’m scared that that will influence our family relationships and maybe make it worse. Weren’t you scared of that? How is your process going there?

Courtland Allen: 24m 0s

I’ve fought with my brother for so many years that there are no fights that we could have that are going to influence our relationship. So maybe that’s what it comes down to.

If you and your family have had a lot of conflict, maybe you’re a little bit more sure. But if you haven’t, if you’ve always had a peachy relationship, then maybe you don’t want to risk that. But for me, it was clear. We can get over anything.

Przemek Chojecki: 24m 17s

OK. I’m pretty sure of that as well. We can probably get over anything, but having a family dinner and then awkward silence, because someone fucked up something on the business side.

Courtland Allen: 24m 30s

What’s been interesting during COVID is, my brother’s in Brooklyn. I’m in Seattle at the moment. My mom is in Atlanta. We’re all separate. I have no plan to fly to Atlanta and see my mom and get COVID in the airport and get her sick or anything like that. But she misses us. So my brother and I have these daily Zoom calls. She’ll dial in. She always asks, “What’s the Zoom URL today?” and she’ll just listen to us talking.

So there’s good and there’s bad, but I think it’s cool to bring the family together. I’m a big proponent of stacking things in your life. There are all these articles you can read online about, “OK. What are you going to do in your life?”

You can have a fulfilling career. You can have good relationships with your friends. Are you going to pursue hobbies? Are you going to spend time with your family? What are you going to do? You don’t have time for all of it. I think you do have time of all of it if you can find clever ways to combine two or more things.

For me, Indie Hackers is very much a fulfilling career. It’s family time, in a way, cause I’m with my brother and now my mom all the time. And it is a hobby, because it’s not just one thing. I get to spend a lot of time writing code, doing design, hosting a podcast.

I think almost anyone running a business, if you design it from the ground up in terms of choosing what you’re going to work on and who you’re going to work with, you can stack a few of these things and in the same 8 hours a day instead of getting one of these boxes checked, you can get two or three of these boxes check and live a better life. I’m a huge advocate of working with family if you think you can make it happen.

Przemek Chojecki: 25m 51s

That’s a great thought. I’ve never thought about it that way, but I like stacking things together in the sense of trying to work on something which I’m passionate about and finding projects that I would like.

So even if I fail in the end, I will have the pleasure of going through the process and enjoying the whole process, instead of thinking that, “OK. It’s really shit right now, but maybe in two years it will be something fun,” which I think is important when it comes to entrepreneurship, because there will be many failures along the way and you should expect them to happen over and over again. Unless you do something which is exciting, you will be miserable otherwise.

Courtland Allen: 26m 33s

Yes, and apparently that’s what you did, because you failed at your first two companies and yet you weren’t discouraged. You kept going, so you must have been having a good time along the way.

Przemek Chojecki: 26m 41s

I think it’s something with my character, that I even like when it gets worse and you get tested, because at least something’s happening. So I had this amazing chat with my friend who is a young film director. He’s also ambitious here in Warsaw.

He was saying that he has this motto recently, the worse it gets, the better it gets as well. Right now he manages a crew of 40 people, and he’s 27 right now. From my perspective it’s complete chaos because film crews probably are not as organized as startups or companies where you have actors, you have some copywriters, and having everyone fill their role is challenging.

I like this as well in being an entrepreneur, that you have that ability to test your skills in battle. It’s like going to war. You can practice before, for example, by reading books, by doing some small exercises like writing a business plan or whatever it is, but in the end, you have to test yourself in battle and that battle is the market. In the end, everything is decided by the market and by other people, by customers who either like your product or they don’t.

Courtland Allen: 28m 3s

You’ve been in battle a couple times. You’ve tested the market. It didn’t particularly go well, but you’re still determined. You’re enjoying yourself despite the hardships. What are some of the lessons that you took with you after the failure of your second company and going into your third company?

Przemek Chojecki: 28m 17s

Probably the most crucial thing is that patience is key, and I have to think long term when it comes to building the product, getting the customers. The other thing is I’m executing better on ideas. I think in the end, entrepreneurship is about executing ideas, and thanks to the previous failures, I’m able to execute on ideas much quicker than before.

I have this feedback loop much shorter, I’m able to build apps, build certain features quickly. I think that’s crucial in the end, because one thing is brainstorming and thinking what can be done next, but in the end, someone has to execute on the idea.

It better to be quick, because the quicker you are, the better you can position yourself on the market. That is a crucial finding from the two previous failures. Another one is I can probably do solo work with the entrepreneurial route and I’m fine doing that.

The third one is, I’m not scared of failures. If Contentyze were to fail, I’m fine with that in the sense that I know I can find new ideas, new markets, new products that I can invest my time and my money in and I’d be OK with that.

I should focus on enjoying the process and being the things that compound, in the sense that they compound assets, they compound knowledge so I can, even after the failure or success, I can come out stronger than before. I think that’s the key.

Courtland Allen: 29m 54s

So you’ve got this mindset that you’re in where you’re like, “You know, the idea’s important but execution is what matters.” And after these two failures, you realize you’re not afraid to fail, and you also realize that you’re comfortable being a solo founder.

You don’t necessarily need to work with anybody else, which I can understand after having this blowup fight with your second cofounder. How do you take all these lessons and move forward to eventually start PetaCrunch and Contentyze?

Przemek Chojecki: 30m 16s

Basically I was thinking what to do next. March 2019 was the moment when Bohr Technology ceased to exist. We decided that we’d close everything.

I took some months off to go on holidays and think about stuff. But then I decided that I want to do something in content because I was always a huge fan of literature. I was always reading a lot, writing a lot. I have published five books. One of them is almost entirely generated by AI, so that was done later on, beginning of this year.

I was doing a lot of writing before coming to Contentyze, and my idea was that in the first place I want to build a tool for myself, to boost my own performance. While I was doing that, I first tested that on PetaCrunch.

That was the first initial project where I was thinking how I can automate the media industry, how can I make something purely autonomous in the media market. Then I went on to thinking that, OK, I can build this platform where these kinds of algorithms can be shared with other people and they can use this well to create content. That was a year later.

Courtland Allen: 31m 28s

Walk me through exactly what that looked like, how you set it up, just so we can have an idea in our minds of what it looks like to build an AI-powered company as an indie hacker.

Przemek Chojecki: 31m 37s

After my previous business failed, this one with quantum computers, in the first place I wanted to talk with other founders, especially the ones who raised a lot of capital or had great product or something like this, how did they do that? What’s the recipe? What did they do right? What went wrong? What kind of advice did they have?

That was the initial idea. Then I had, after thinking a little bit about it, I decided, “OK. Maybe I can do it at scale and ask 10,000 founders at the same time.” That went into being PetaCrunch. I did it for myself to learn about, how do you build a business?

Courtland Allen: 32m 20s

that’s almost exactly why I started Indie Hackers, to be honest. It was like, I want to understand what other people know that I don’t know. Let me go talk to all the most interesting, successful founders and say, “Hey, this is an interviews but I really I want to mine you for information and figure out how you came up with your idea, how you made your money, how did you find your first customers?” But you did it at a way bigger scale than I did it. I think I’ve interviewed 500 founders on the website in the last four years. You’ve interviewed more than that in just a couple of months.

Przemek Chojecki: 32m 46s

Yes. That’s true. But that’s the power of AI. But I’m sure they are not as high quality as what you did on the platform, because to some extent, those were early days. It’s mostly scripted conversations.

But I managed to get interesting people on the platform. It wasn’t me who was doing all the talking, it was these AI journalists who were contacting people. They managed to get to talk with CEO’s of startups who raised $100 million or something like that. I won’t name names here, but you can check them out on the platform. It’s cool. I learned a lot in that process, and that’s what led me to Contentyze in the end.

Courtland Allen: 33m 30s

Explain how that AI works, because I’m curious. Who does it choose to reach out to? How does it find them? What does it send them? At what point do you come in and do some human intervention, if at all?

Przemek Chojecki: 33m 40s

Sure. That was super simple, because we were scraping the data of the startups who got funded in a particular amount, for me it was between $1 million to $100 million, and sending them emails. That’s it. Then you if you got a response, you’d send them the questions. That’s it.

Courtland Allen: 33m 57s

And what part did your AI do, all of that?

Przemek Chojecki: 34m 0s

All of that? No. All of that was done by AI. I only intervened when someone asks some nonstandard question. Like in the beginning, someone asked one of the AI journalists, “Are you AI? Prove to me that you’re not, and I’m happy to answer your questions.”

The AI was requested to tell a joke about anything. So that was the human intervention. I went online. I searched for a joke and I emailed this person this joke, and that was it. And then AI took over from that place.

Courtland Allen: 34m 33s

Wow. So tricky. So then you would just send them a bunch of questions, I guess, and then they would answer your questions and you’d publish the interview?

Przemek Chojecki: 34m 40s

Yes. Unfortunately the last part also needed the human intervention, and that was all of the interviews were in different formatting, in different styles. I had to make them more appealing visually, I had to do it myself unfortunately. So that was the painful part.

Courtland Allen: 34m 57s

I think that’s the beauty of AI though. Because I think people think of AI as this all-or-nothing thing. If the computer can’t do all of it, it might as well not do any of it, but that’s not true.

As a founder, if someone could take off half of your workload, if you hit a button and suddenly this bot is going on Crunchbase, finding all these founders who got funding, emailing them, initiating an interview request and you’re only intervening to polish it up at the end or to tell people jokes when they want proof that it’s not an AI, that saves you a tremendous amount of work. It’s more than worth it, even though the AI can’t do a hundred percent of it by itself.

Przemek Chojecki: 35m 28s

Yes, exactly. It’s the same with Contentyze. We’re not promising you get a perfect article on whatever topic you think about, but what we promise is that you’re going to save a lot of time, whatever you’re writing, because you can just make the headline. You can submit the whole article, and we summarize that for you.

It’s like that with any kind of AI right now, because we don’t have any perfect AI doing everything for you. It’s mostly about automating some small stuff so that you can be more creative yourself.

This is how I view AI in the end. AI is just a tool to boost your productivity so that you can be more productive by not focusing on repetitive, boring tedious work you would have to do every day otherwise.

Courtland Allen: 36m 14s

So why not keep working on that? It seems like it was working. You’re much more efficient than dopes like me who are doing interviews by hand. Why not keep working on PetaCrunch and make that your main thing?

Przemek Chojecki: 36m 22s

No, it’s still working. What I decided to do is I didn’t want to spend my days editing the interviews. Right now you can still go on the platform but I switched the pricing model. PetaCrunch is making money because you can submit your interview for $99.00 and be featured on the platform after some approval.

It’s fully working. It’s something in the back of my head because I’m mostly thinking about Contentyze, but from time to time I’m thinking “How can I automate even more of PetaCrunch so it’s fully autonomous? It wouldn’t require anything from me, and it would be a fully working media service.” I would like to have TechCrunch but done by AI, no human intervention.

Courtland Allen: 37m 14s

That sounds fascinating. That’s one of the coolest business ideas I’ve ever heard. I don’t know how good these AI journalists can be. Maybe they can do interviews. I don’t know if they can write entire stories.

Przemek Chojecki: 37m 25s

The whole point is that this is what we’re trying to do with Contentyze. In the end, Contentyze is the missing part in the fully autonomous PetaCrunch. The huge subject is the whole synthetic media landscape. By synthetic media, I mean exactly media which are run by AI or algorithmic scripts, just no human intervention in the middle.

Synthetic media is, for example, these articles generated by AI but also Twitter bots and many other things that you can think of. For example, videos can also be fully generated by AI. They are not as good yet as those recorded by humans. But this is another direction that synthetic media is taking.

Courtland Allen: 38m 9s

So when you started Contentyze, was that where your head was at, “I want to turn PetaCrunch into this full media organization, and I just need this one missing link, so I need to go build that first and then maybe come back to PetaCrunch”?

Przemek Chojecki: 38m 20s

Yes. That was my motivation back then. And then doing that, I remarked that maybe I can make the whole platform public and have other people use the platform. Then this is what happened, and I’m happy interacting with other people talking about Contentyze, because I’m getting a lot of ideas from customers, users. It’s great. I wouldn’t be able to do it all by myself.

Courtland Allen: 38m 46s

So tell me about the first month with Contentyze. You come up with the idea. You’re like, “I’m going to build an even more advanced AI system that you can give it a prompt and it will write an article for you.” What’s the first thing you do after that?

Przemek Chojecki: 38m 56s

Exactly. So the first step after I had the idea, I drew all the dashboards that I wanted to have. I drew the architecture interactions I would like to put in place. And I came back with my friend who is a great item back-end developer to do that back-end stuff for me. We culled it together, some parts. That took us two months to do it. It was great.

It wasn’t as good as I thought in the sense of like how UI and UX is designed and everything. But then I decided, OK, it’s cool enough to keep working on that. I have a lot of fun doing that, and so I started looking for contractors who could take me to the next level.

Courtland Allen: 39m 36s

And what exactly was it that you had built with your developer?

Przemek Chojecki: 39m 40s

That was a platform working in the cloud, but the only thing you could do was, you have a place to put your prompt, to put your headline or a short sentence, click on generate, and get a text out of that.

Courtland Allen: 39m 54s

So I could say, Ten Ways to Come Up with Business Ideas. I would write that in a text box and click generate, and then your AI would write an article that’s ten ways to come up with business ideas?

Przemek Chojecki: 40m 5s

Yes. Basically a draft for the text. So with that version we got up to 100 users, so definitely there was some interest in the platform. It was growing super slow and I was thinking of a ways to make it more appealing, like UX design, product design, and how to make it much better.

I know that I have the technology in place to make it more appealing, but the whole product side was what was missing. So I went home and started looking for people to do content for me, design more things in the backend, add different machine learning models and stuff like that.

So I decided that I could grow it into something much bigger by myself. It was a good decision. In the end we finished the second iteration in June 2020, so after three months and from then on, we had this great group of users.

Courtland Allen: 40m 58s

So AI, to my mind, is a little bit intimidating because it seems like something that’s super expensive to do. It seems like you need to have a lot of experts, a lot of PhD’s, and it’s also going to take a ton of time to get it right.

And yet here you were with a small, scrappy team, somehow bootstrapping this and funding yourself, and somehow getting it done in three months. How were you able to be so efficient and so productive with such a small team and in so little time?

Przemek Chojecki: 41m 25s

Well to be honest, it’s part of my own expertise. I didn’t have to look for someone who’d design the whole infrastructure. I did it myself. And then I needed people for very particular pieces that I know that either I can’t do myself or maybe I can do it myself, but it will take a couple of weeks.

If I get an expert, he could do it in one week and a couple of days or something like that. So I was able to optimize my own time and resources thanks to the fact that I’m able to understand all the pieces coming into the platform, so I was lucky with that.

I mean, I didn’t have to take any black boxes. I could understand every single algorithm that I was using. In the worst case, I could quote myself the whole thing, so that was on the good side.

I’m feeling confident that I could do the whole thing myself, so even if I couldn’t find a person to do it, I could always do that myself. That was a great thing, because that gave me a lot of freedom to pursue the whole thing.

Courtland Allen: 42m 30s

Would you have been able to do this by yourself, let’s say two or three years ago, or are there new tools and platforms that are making what you did this year possible?

Przemek Chojecki: 42m 38s

Well I couldn’t do that two years because of the breakthroughs in AI machine learning and what’s new on the market. I couldn’t do that for another reason, because I wasn’t that good in programming back then that I am right now.

I’m still learning because computer science is always learning. There’s always something new. There’s new research coming so you have to learn more, but no definitely, Contentyze couldn’t have existed a couple of years ago. It’s cool that the timing is perfect as well.

Courtland Allen: 43m 13s

Give me a sense of some of the breakthroughs, because I think this is what’s enabling indie hackers to do things that they weren’t able to do before. And unless we are ex-PhD’s who have our ears to the floor of what’s going on in AI, we don’t know about these breakthroughs. So what’s new and what’s possible for an indie hacker to do today that they couldn’t do a couple years back?

Przemek Chojecki: 43m 32s

If you don’t have this technical background, you can still do a lot of things because you have a lot of open source models, models which are readily available from IBM, Google, Microsoft, Amazon. They all have models which are easy to connect to through API.

That drives down all the cost. For example, say I want to build an application recognizing cut faces, or maybe recognizing a particular dress and then searching for that dress on the internet. Right now, with open source tools from one of those big tech companies, you could probably do that under $1000.00 or $2000.00, if you can hire people from Fiverr or Upwork to do particular beats of the technical stuff for you.

It drives the whole cost of building the solution down, and it’s truly great. So the first thing is that technology’s much cheaper to do because there are so many ready-to-use, off-the-shelf components, and on the other hand, because of the whole world going to the remote work, it’s super easy to find people on Upwork or Fiverr to do that for you. So you can iterate very quickly on your ideas and it doesn’t cost you much to do that.

Courtland Allen: 44m 57s

It’s fascinating that it’s not so much that you’re trying to build out an impressive team and compete with Google or Amazon or Facebook, but rather they’re releasing a lot of their work for free on the internet. And even if you don’t understand it technically, you can hire someone who does and package it up and build these solutions and bring them to people or industries that just don’t have a lot of sophistication.

Maybe they’re doing everything in an Excel spreadsheet or maybe they’re hiring people to repeat a bunch of manual tasks and you could whip up a quick script and sell it to them for 30 or 40 bucks a month that’s going to enable them to get a lot more done, assuming you have the domain expertise to spot these problems and figure out the inefficiencies that companies are going through and match them up with the right algorithm or the right model with AI that can solve that problem.

Przemek Chojecki: 45m 41s

Yes, definitely, especially that there are many industries that are still using Excel spreadsheet as their resource and even without any technical background, you can be a broker for technologies connecting that AI with a particular domain, and if you’re an expert in a particular domain, like logistics, transportation, maritime, whatever it is that’s your passion or your background, your career background, then you can find solutions which are easy to sell for a particular market. I think that’s great.

That’s definitely doable for indie hackers to do it. If, for example, you have your corporate job but you’ve had enough of that. You want to quit. Then the easiest way is to look for ways that AI can automate some of the processes.

It’s not even that hard, because it’s mostly thinking in terms of processes and the things you want to automate. Because AI, in the end, is just a tool for automation. It’s nothing magical right now, but if you can find processes which are repetitive, tedious, boring then you probably can create an AI machine learning algorithm who will do that for you.

If you have no technical background, then probably the most practical advice would be, think about this process and describe it well, and then start writing to different software houses specializing in AI data sciences machine learning, asking them for a quote about the particular solution that you have in mind.

Then with that quote in mind, you can go back to these particular industries, add a margin on top of that of 20, 30%, and maybe start selling from that point on. And I think that’s the best way to build something technical without having any technical background at all.

Courtland Allen: 47m 36s

That’s super fascinating. I want to get into more detail on that a little bit later, because there are so many different types of AI models and technologies that you can use to do different things, and I’m not sure everybody knows what the options are.

What’s the difference between the natural language processor and a recommendation engine and whatever else is out there. But before we get into that, I want to understand how you were able to finance what you were doing.

You said you started doing this in January 2020, you started Contentyze, and it wasn’t until June 2020 that you had your second iteration out, which I can only assume is much better than the first, more accurate, more impressive.

But you’re hiring during this time. You’re presumably surviving and eating and paying your rent during this time. How are you funding yourself? How are you making enough money to keep this operation going?

Przemek Chojecki: 48m 20s

Basically I have some savings. That was part of it. But also I was sustaining myself through consulting. I was able to do freelance consulting. And when I was starting, I was consulting for up to 20 hours per week and from there it went down.

Right now I still have consulting projects, but they are no more than five to 10 hours per week. I had this approach that whatever I make through consulting, that goes to me, and then whatever the company makes through content, subscription, that gets reinvested into the company itself.

So I had this luck of being able to do things on the side with consulting, so I didn’t have to quit any kind of a job but reduce the amount of clients that I had to focus full-time on Contentyze. Thanks to the growing revenue of Contentyze and all the content generated, I was able to cut down the hours and cut down the clients that I didn’t need anymore for my consulting business.

Courtland Allen: 49m 24s

I love that approach of working a job or doing consulting so you’re generating revenue. And then in the early days not just saying, “OK, well I’ve got this job but I’m going to squeeze my company into nights and weekends or whatever few hours I have remaining.”

You’re like, “No, no. I’m going to take the money from my job and I’m going to use that to hire people. Or I’m going to take the money from my savings. I’m going to use that to hire people so I can grow my company faster.” So then you’re not really trading anything off.

Yes, you don’t have a ton of time to work on your business, but you’re generating all this money from your job, and now you’re pouring that into making your business grow faster. I think most people don’t do that. They either do one extreme or the other. They either quit their job cold turkey and say, “OK, I’m going to burn through all of my savings so I have no income anymore,” and that’s super scary. I’ve been there. It’s stressful and it sucks if it doesn’t work out.

Or other people will do the other extreme which is like, “All right. I’m not going to quit my job but I’m also not going to take any risks whatsoever. I’m not going to spend any money hiring anybody. I’m going to do it all by myself.”

I think the sweet spot is exactly what you did, and especially if you have a consulting job, you can gradually reduce the contracting and consulting work you’re doing down to five days a week or four days a week or three days a week as your business ramps up its revenue and becomes more and more successful, so it’s never this huge, risky thing.

Przemek Chojecki: 50m 32s

Yes. I also think that’s the best way. Of course, it’s not always possible that you can simply reduce the amount of previous work, although sometimes definitely it is, especially right now probably you can ask your boss, your employer about reducing your time from full time to part time and maybe start with that.

I’ve heard many stories about people doing that, just saying openly that they have something they work on the side, and they want to take more time to do that. I think that’s a great approach as well.

But definitely if you’re a freelancer, then you can already start doing that, and you can think about how much income you need each month and then reinvest everything else into your business.

But I think that’s the most tricky part, because probably many people could do something like that, but it’s super scary to invest your own money, especially in uncertain times like a global pandemic for example.

Because when you’re investing in business, you don’t know whether you’ll have that back, when or if at all. But in order to make money you have to spend money, and that’s the only way that works. So you have to spend either on building the product or marketing, sales or whatever it is, in order to make everything faster.

Courtland Allen: 51m 49s

Yes. And even if you’re not technically spending money and you’re just investing time, well your time is money. The time you spend on your company could be time that you’re working another job or doing contracting and getting paid. So I think that insight is a hundred percent true. You have to spend at least some money to make some money.

If you’re currently working a job and continually earning money, it makes it much less scary to start investing it and spending a little, compared to quitting cold turkey.

In your particular situation, Contentyze started making money on its own, not from paying customers but from you taking advantage of your own AI writing well before you had paying customers. What did that look like, exactly?

Przemek Chojecki: 52m 26s

Even before Contentyze, I already started working on some kind of side income, because of all the situations with startups, failures, and then I quit academia. The first startup and up to probably the middle of the second startup, I still had a position in academia which was doing research. That was fine because I still had a couple of research papers going in, so I had this super stable position and I could do startups, because even if I fail, I can go back to academia.

After my second startup failed, there was no way of going back to academia. I had no side income, so I had to work on consulting more in order to feel more secure. That something more started with Medium.com, my favorite blogging platform. I publish a lot on Medium.

At some point I noted that you can turn on monetization and it works the same as it works on YouTube, meaning that you get paid for how many views there are, how many people are engaging with your content.

I started getting a couple of hundred dollars in the beginning, and that was mid-2019. So that kept motivating me in the first place. I mean, it wasn’t enough to sustain myself, but it was already enough to think about this as a direction, and what to do next.

So I was developing my writing, writing for others, writing for myself. Then I discovered affiliate marketing and that was a great discovery. There’s this idea that you can promote some products, put the links to this product, and if somebody buys something through this link, then you get paid some percentage of that.

So I was writing about data science books, machine learning books, different technological stuff, and I started putting those links, not to be ordinary links but to be affiliate links. Suddenly my income grew to like over $1000.00, $2000.00 in a short time and I said “OK, that’s great and I can keep on going.”

That was the start also of Contentyze. I was thinking to myself that if I can do it myself and reach the level of $1,000.00 per month and maybe if I start to employ AI and do various content on a much larger scale, then maybe I can do much more than that.

Courtland Allen: 54m 49s

And how much more did you do than that when you started employing AI to write content for you and put in affiliate links?

Przemek Chojecki: 54m 57s

The highest I’ve gone is probably $5,000.00 per month, but then I was fully focused on the content. I took the decision to leave it as it is so that started to fall off, but then focused more on the platform. Because I felt in the long run, I think I can make and generate more income from the platform, the licenses and doing everything around Contentyze than the actual content.

Courtland Allen: 55m 26s

So you’ve got this fork in the road where it’s like you can either generate a ton of content and try to make money from affiliate fees, which is cool and very innovative, but also probably not as big of an opportunity as building out this Contentyze platform so other people can pay you a monthly fee or something to come and use your platform to generate whatever articles they want.

Przemek Chojecki: 55m 45s

Exactly. I decided that I want to stay at this level of a couple thousand dollars per month with these affiliate links and that’s fine, because that gives me freedom to reinvest all that into hiring people, to buying ads on Facebook, or whatever it is that needs to be done with business, and I don’t have to worry about going through my savings and having nothing at the end.

But then I decided that I don’t want to develop more of that side, because that also takes a lot of time and I just want to build a better product. With a better product, if Contentyze doesn’t work well I can always go back to that part.

Courtland Allen: 56m 22s

It still fascinating to me, even though it maybe doesn’t have as crazy a potential as this other product you’re building, but for a lot of people, a couple thousand dollars a month is super meaningful income. The fact that, for you, it’s more or less like you’ve got an AI doing it, how much manual effort do you have to put into these articles to keep generating a couple thousand dollars a month in revenue?

Przemek Chojecki: 56m 43s

Probably five hours per week, because I have to curate some of them and for some of them I have to set up ads to make them more popular. So I would say up to five hours per week right now. There was more time to be put into that to create all those digital assets, but right now to sustain it at this level, it’s not that much time that I need, which is great.

If you look at that from the perspective of the lifestyle business, I definitely could go into that direction working five to ten hours per week and then drinking coffee, hanging out with friends or playing video games for all the other time, but for me, I would be so demoralized after one month.

I had this period in my life where I was trying to do something like that, but I couldn’t. I have a hard work ethic and I have to work to feel fulfilled and be happy with my life. So I can’t do things like that.

Courtland Allen: 57m 45s

I think a lot of people hit that point and it’s a checkpoint. They want to get to this point where they feel stable and they feel supported and they feel like they have some sort of machine that’s running that can generate income and they’re not going to go broke and starve.

But then once you get there, the same thing happens. You’re like, “OK, well this isn’t enough money. This isn’t as ambitious as I’d like to be. I do want to spend my time working on things that are hard and challenging,” so you don’t stop there. But still it’s hard to get there for so many people.

Walk me through an example of what one of these articles looks like. What’s the process? Do you have to come up with a topic idea yourself, or does the AI do that? And then how are you getting people to read these articles and click the affiliate links? What’s the whole process look like?

Przemek Chojecki: 58m 23s

In the beginning that was super manual, in the sense that that was the pieces of advice that I would give myself. For example, what books do you have to read as a data scientist? There’s a book for each level, whether you’re a beginner, intermediate. So everything was about data science machine learning technology.

That was working, and still you can see that on my blog on Medium. If you go there, most of the text that I’m writing, I might be writing myself or with a little extra help from AI to generate some of the drafts. But basically, I curate everything.

But then to go on with that on a massive scale, then what you have to do is you have to look at the keywords that people look for but there is not enough content, and then try to provide that content.

There are services like BuzzSumo or Ahrefs, and with both of them you are able to look for what people are Googling in a particular month, then within the particular niche, within particular regions.

And then you look at the questions people ask within the web browser, and then that’s the title of the article that you want to write to have the organic traffic coming to your website, coming to your blog post. So that’s what we’re currently doing in an automatic way. But there’s definitely much more to be done.

Again, if I were to do it properly, it would be a totally different business that would take a full-time job. Unfortunately I can’t do everything, and I had to decide what I need to do, and I decided to go with Contentyze, be more on the content generation part, rather than content research, content marketing part.

Courtland Allen: 1h 0m 5s

Right. It’s cool that you had a little foray into this, because you’re dog fooding your own product. You’re getting to see what it’s like to be a customer of Contentyze and how you might be able to use Contentyze from a customer’s point of view to try to make money, but then you decided, “OK. Well let’s focus on the platform.” What did you learn from using Contentyze yourself to make money, that informed how you built your product and how you would find and entice customers to try using it themselves?

Przemek Chojecki: 1h 0m 29s

The most important thing I learned is that UX design is super important. Unless the product is super intuitive and easy to use, nobody will use it. For me, it was easy to try, because in the beginning especially, Contentyze was already working from the cloud, but then still I was more involved with working with my own scripts in a Jupyter lab with Jupyter notebook. So I decided that it shouldn’t be like that.

If the product is going to work, then it should be more appealing to work with Contentyze than trying to code it yourself. I was iterating on Contentyze until it reached this point where right now I’m less working with my own code than with Contentyze because it’s much simpler to do everything with Contentyze rather than to code it yourself.

That allowed me to have this super short feedback loop, because I didn’t have to talk with anyone about what to change in Contentyze to make it better or easier to use, because I have myself.

I had myself and I was able to ask myself, “Is it enough or should I push a little more?” I’m trying to be transparent with myself, pretty open, so I wasn’t lying to myself about, “Oh, it’s going great,” when it was absolutely awful. I was iterating very quickly.

Courtland Allen: 1h 1m 46s

It’s fascinating how much of an advantage you get and how much more adoption you can get when you have this easy-to-use interface. There’s another AI model out there that’s super popular. I think in part it’s just because it’s so easy to use, and it’s GPT-3.

So essentially you get access to their API. You send them a prompt or a text. It could be almost any question or instructions using English, and it can generate a whole wall of text for you that it’s almost indistinguishable from a human writing it. This technology’s been around for a while. Maybe it’s not quite as good, but people haven’t used it because it’s been so hard to use. You have to set so much stuff up.

But now, when it’s so easy and it’s this easy to use API, I think people are getting super creative. Pieter Levels, I think he got access and built this thing called Ideas AI. Every day it generates a few new ideas using AI, and you can upvote the best ideas. It’s got all these business and startup ideas that it’s generating, and people are applying it in all these new ways.

So it makes sense to me with your business, figuring out how to make the interface more intuitive and easier to use would increase your revenue and increase the number of people who are interested in using it and understanding, like, “OK. Here’s how I can use Contentyze to get ahead in my business or to do something else that no one else is doing.”

Przemek Chojecki: 1h 2m 57s

Yes. This is something I learned and I keep on learning, is that it’s not always about technology. It’s often about how you present the product or the product design.

This is something which of course Apple got well, because they don’t produce the best laptops. They definitely don’t have the best components, but what they do well is the whole experience of using a MacBook and how intuitive it is, how easy it is to use, how well-performing the software is.

That’s something super important to remember especially for me, as coming from this technical background, I have this inner thinking that building cool technology will solve all the problems, when it’s only halfway true and maybe even not halfway, because technology itself is nothing. It’s just a tool that you can use well or not.

Selling the thing or presenting the thing is more important in the end, getting the technology into proper hands to make it appealing to the particular audience you have in mind. This is the challenging part for me, but it’s something that I keep on learning, and this is great to keep on iterating about and keep on iterating on product design.

Courtland Allen: 1h 4m 15s

So where are you finding these customers? Because even if your product design ends up being super slick and it’s great, most people don’t know that you can generate a lot of content and articles and product descriptions or whatever with AI.

I think you probably have to do a lot of evangelizing and educating people and convincing them that, “Hey, this is possible, and this is a new way of working.” How are you finding customers for Contentyze?

Przemek Chojecki: 1h 4m 38s

Right now it’s natural. It’s funny because we generate so much content that we have organic reach plus social media. We try to explain as much as possible how Contentyze works, what the natural language generation market looks like, and what kind of tools you can use.

It’s like you said, it’s all about educating potential users, clients, and I still believe that it’s a long process. I know we have just started, but this education process might happen that you will see one video of Contentyze right now, maybe another one in half a year, and then you come back to us in one year, and that’s fine. But it’s writing a lot about what you can do and how you can make your life easier by automating some parts of the writing process.

Courtland Allen: 1h 5m 28s

Give me an example of a customer or two you have who’s using Contentyze and what they’re using it for.

Przemek Chojecki: 1h 5m 35s

Sure. So for example, there are many people in real estate using Contentyze to turn spreadsheets into actual plain English explanations. So for example, one person has a compounder which aggregates data about rent prices in cities in the U.S., with how those change from week to week, what’s the hottest neighborhood in Chicago, what’s trending in LA.

And then you get this bunch of data, large spreadsheets, but in the end his users want to have a short report. This is what Contentyze does. You can quickly implement a template which would give you this summarization of the text, of what’s going on within the particular city.

Another example might be another customer is a person who’s running a health lifestyle blog, and she generates the blog posts, has a couple of headlines, has a couple of articles she wants to base the whole text upon, and based on that, Contentyze gives her a draft of the text that she then modifies and puts on the platform or on her website.

Courtland Allen: 1h 6m 49s

Cool. So there are a whole bunch of different use cases.

Przemek Chojecki: 1h 6m 52s

Yes, and to be honest that’s also problematic from my perspective, because I don’t know where exactly to focus my attention. But it will be easier if there would be one type of customer, for example, only real estate or only ecommerce, because there are a bunch of people from ecommerce, or only marketing people.

In the end, it will probably boil down to me choosing again after which particular market I want to go, because you have to make some kind of a choice in order to make it easier.

Courtland Allen: 1h 7m 22s

Yes., that’s the difficulty with building any technology where you’re enabling people to do things that they’ve never done before. Suddenly there are all sorts of use cases and you have this very generic, lukewarm message. It doesn’t strongly appeal to any one group, so it’s hard for it to resonate with people.

I talk to Vlad Magdalin who runs Webflow, and they’re in the same situation, where they’re enabling people to build websites using this intuitive UI that they didn’t have before. People are building website for all sorts of different reasons, so I think he eventually narrowed down to three or four different target groups who seem to be the heaviest users, who got the most benefit from the product, and built out landing pages and documentation and all sorts of resources for those three or four groups.

That means excluding everybody else and not focusing on them, but I think it’s important to get that foothold in the beginning, and then you can run with it. I think a related concept here is that there’s this book, Crossing the Chasm, which was popular in the 90’s. I’ve talked it before but there’s this idea of, you’ve got the late adopters and you’ve got the early adopters. And at the very earliest edge of the early adopters you have the innovators.

These are people who are often entrepreneurs themselves and they’re working on things where they want to get ahead of everybody else and they’re visionaries. They’re always on the lookout for anything that’s new, anything that they can take advantage of that others aren’t taking advantage of, and they care a little bit less about the product being polished or full of bugs because that might be an advantage to them.

Maybe that means that they can use Contentyze before anyone else can use it, and they can pump out content and automating what they’re up to well before anyone else. So I’m curious how many of your customers are in this camp. How many of your customers are indie hackers or people who are way ahead of the curve and who want to take advantage of AI before other people do?

Przemek Chojecki: 1h 9m 6s

I think most of them, for now. Definitely people who are using Contentyze are visionaries to some degree, because it’s not as easy as I would like it to be to get on the platform.

There is still a learning curve you have to go through. So you have to be determined to at least spend half an hour or an hour learning all the possibilities within the platform to do something meaningful. So it’s still, I think, the visionaries.

But from my perspective, that’s exactly the goal of building the product which is appealing to as wide an audience as possible, is trying to make it as simple as possible as well. So whenever you login onto the platform for the first time, you right away know how to do things in a way that brings you value.

I think that’s very important. So it’s literally about having product design which is great, easy, compelling, and great to use in the end. That’s the most important thing.

Courtland Allen: 1h 10m 14s

So when I think about all these indie hackers trying to get ahead, trying to start new things, and I think about that and the intersection of AI, there are really two different approaches that stand out, and you’re doing both of them.

One approach is you try to innovate. You try to build some new AI solution that allows other people to take advantage of AI in ways that they haven’t before. So at Contentyze you’re doing that because your allowing people to generate all these articles, a massive number of articles or product descriptions or whatever that they’ve never been able to do.

That strikes me as pretty hard to do, because you need to be an AI expert yourself or you need to hire AI experts, and you’ve got to innovate.

The other usage of AI that I think is empowering for indie hackers and solo founders is using it behind the scenes. You’re not providing an AI product to somebody else but you’re using an AI product or maybe you’re even building in-house tools to help you become much more efficient as a founder.

As a team of one, you can interview a thousand founders, or as a team of one you can provide customer support to hundreds of users while you’re coding your product at the same time. Which of these two paths do you think are more promising for indie hackers? I’m leaning toward the latter, that if you want to take advantage of AI, you should be using it behind the scenes.

Przemek Chojecki: 1h 11m 24s

Definitely. I mean, probably the second one, because especially I think we going to the world of no-code, or at least minimal code, where you can implement a lot of things without knowing how to code and you can, by connecting various API’s, maybe adding some kind of a layout on top of that, you can quickly build new solutions which would cost you thousands of hundreds of dollars before, and right now they cost you maybe a couple hundred dollars at most or even less than that because you’re just connecting different API’s. It’s super important to try and understand what people might expect, and then give them that solution.

Courtland Allen: 1h 12m 12s

So give me an example of some of these API’s you can use or ways that you can use artificial intelligence to make yourself more efficient as a founder and avoid doing a lot of repetitive tasks?

Przemek Chojecki: 1h 12m 22s

The product that I’ve discovered just lately, built by another solo founder, is NAN.io, that’s a product done by one solo founder from Berlin. He recently secured the funding so the whole thing is growing.

It’s already popular within the deaf community. What it does is it’s like Zapier on steroids. It allows you to connect various things like Google sheets with emails, with Airtable, with a couple of other things, super simple to use.

You have to code just a little bit to start using that, but it can save you a lot of time, especially as a solo founder, because for example if you’re running a blog for WordPress, and you want to have statistics about your blog post, then you can connect WordPress to Google sheets, to email, and suddenly you hit some major milestone, you get a trigger on your email saying, “Hey, you just got 100,000 views.”

Then that’s being written down in your Google sheets where you manage all your statistics. So things like that which would take you a lot of coding before, you can do with a couple of clicks. That’s one example of a productivity tool using heavy AI, actually AI-based.

Another thing which is pure AI and it’s great to use is Lumen5 and Synthesia. Those are two other startups I guess, both more in computer vision. So what Lumen5 is doing is that you provide the script and by analyzing the script you get the whole video. For example, I can explain a video or a sales pitch or something like that, and it literally takes five minutes to create the whole video from your script.

Of course, you can modify it later on the platform, but it makes the whole process of building things much faster. You don’t have to spend money to build and explain your video. Of course, if you have money to find professionals to do all the automation, do all the design, then that will be better in the end.

But if you’re on budget, if you’re a solo indie hacker and you want to focus on the product, then that can save you both money and time to do that. Synthesia, on the other hand, is automated anchors in the sense that you can create this virtual persona who will talk from the script.

Again, you submit your script and then you have someone AI-generated talk in this artificial voice for your script. So you can have this whole staff of people saying things on camera for your company.

Those are small things that add up together to something much bigger, which is you can literally automate many of the things you’re doing currently by one AI startup or another or just different solutions. And that, again, drives the cost of doing startups much lower.

Courtland Allen: 1h 15m 38s

Yes, I think this is so cool and that you’re thinking in a way that’s very different than most people are thinking. Because most people have never heard of these startups and they don’t know this exists, and so when they go into particular tasks, they don’t think, “Well let me first Google this to figure out if there’s an AI startup that’s got this unlocked. I don’t have to make this video, and AI can make the video and I can just edit it. Or I don’t have to write this article and AI can write this article for me, and I can come in and snip some things around and add some content and then it’s done and it’s gotten me 90% of the way there and I just do the last 10%.”

It’s super cool that it’s not all about hacking things together yourself, but it’s about going out and paying for somebody else’s product, which is usually affordable and worth it. But let’s say you want to compete with some of these bigger tech companies. One of the hot companies right now is TikTok, obviously, and their entire main feed of all these funny videos and these posts people are making, is AI recommended.

You don’t have to follow anybody. You don’t have to subscribe to anybody. You go on and you see all videos that exist in the world and they look at your viewing habits and recommend, “OK. Based on how much time you spent watching this post, or which post you liked, here’s what we’re going to show you. Here’s what you like.” As an indie hacker, is it possible to build something that competes with that or that mimics that behavior in your app, or is that still years away?

Przemek Chojecki: 1h 16m 52s

No, that’s not possible. I won’t lie about the fact that in some use cases, you won’t be able to compete with the big players because of the fact that some of those machine learning models take a lot of data that you won’t have as an indie hacker, and on the other hand it takes a lot of capital in terms of computing power.

So probably to do something like TikTok did, because they recommend their system is really great, really unique, you probably need to spend a couple million dollars for the computing infrastructure, not counting in the costs of senior data scientists, machine learning engineers and people like that around the whole thing.

So no, definitely that’s not doable by indie hackers. GPT-3 is also probably not doable by indie hackers because of the size of the model, but there are ways to shrink it down and make it more affordable. But then you still have to be an expert in the field to do that.

It depends. Oftentimes, you won’t be able to compete with those tech giants by going in a head-on battle over who’s got the better model, because they will have more data. They will have more computing power than you, and they can keep on throwing resources at that particular problem.

But what you can do is, you can be smart about it. By being smart, I mean you can find different hacks and different small things. For example, generate this video by AI because you can use this startup technology, and use technology from the other startup. It’s like sewing your own solution from what’s available on the market.

With that, you would be faster than any big company because the whole process for them is going through different decision makers. They have to decide on budget and maybe until they start doing things it will take them a couple of months.

In a couple of months you will have a perfect working solution, sewed from those different pieces which are available as open source, as other startups. I think that’s the biggest advantage right now for indie hackers. You can definitely be much faster than any big company.

Courtland Allen: 1h 19m 7s

What do you think about the risks of AI? There’s a lot in the movies and trending discussions about artificial general intelligence, which is this idea that we’re going to create some intelligent machine and it’s going to be like Terminator: Judgement Day. It’s going to attack us, and we’ve got to worry about that.

But even besides that, the democratization of relatively dumb AI, really specific tools that do exactly what you tell them, is still risky in my eyes because it’s enabling people to do things they never could do before.

You could put out a million articles, right now, which also implies that you could create a million bots on Twitter that all look like humans and act like humans but are just bots, influencing a particular political opinion et cetera.

This has been a problem since 2016 when people weren’t even using bots to do this. They were paying people to do this. But now, it’s in the hands of an average person who’s got a few hundred bucks to generate this content. What do you think this means for the world? Should people be worried or is it an overblown fear?

Przemek Chojecki: 1h 20m 6s

I think it’s a valid question. We should fear but we should also think about solutions. Because in the end, AI is just a tool. It’s like a knife that you can use to cut vegetables or kill a person.

Ideally, you would have some regulations which would limit how AI can be used in those cases or at least some tools to prevent those especially harmful behaviors. I think especially Twitter and other social medias are getting better with catching bad players, catching people who want to spread misinformation or buy political ads, but on the other hand, it’s not going fast enough so that it won’t catch everyone.

Bad players who have more budget, more capital to hire people to do really bad things, they can still do that. So I wouldn’t be scared of average people doing bad stuff for fun or whatever is the reason, but still organized groups of people with access to capital and talent can do bad things and they can do even worse things than before. So that’s probably problematic. There was this great documentary on Netflix recently, The Social Dilemma.

Courtland Allen: 1h 21m 22s

Yes, I watched it.

Przemek Chojecki: 1h 21m 22s

That touched upon this subject. But I think it’s the beginning of how we treat social media and general direction. Because I think we’re still early in the process even when it comes to the internet and how we deal with Facebook, Twitter, LinkedIn, and other social media.

How can we get all of these things so that we don’t lose all our time watching something on Netflix, YouTube, scrolling through different social media feeds? How do we limit those things? Because I think that’s the bigger discussion.

Because in order to limit and prevent the harmful behaviors, the harmful usage of AI, we have to limit how we spend time with electronics, internet.

Courtland Allen: 1h 22m 9s

Yes. I think The Social Dilemma in particular was fascinating, because there’s this question of, “OK. What about bad actors who use these huge platforms and who might use AI to influence public opinion on these platforms?”

But then there’s the flip side. What about the platforms themselves? What’s stopping Facebook itself from using artificial intelligence to radicalize people, maybe even unintentionally, or to waste tons of people’s time getting them to scroll feeds and look at ads?

Then on top of that, you’ve got this present unbundling of these bigger social networks, where we’re seeing the rise of smaller communities. Indie Hackers is an example, but there are all sorts of communities that aren’t necessarily run by these huge companies, that are run by people who have smaller budgets. And as more and more people move to these communities, what defense do they have against bad actors?

If a country decided they wanted to build a ton of AI to flood all these smaller communities with all sorts of fake opinions and posts, do these companies have the resources to build a counter-AI? I think it’s a fascinating problem, and in some sense almost seems like sci-fi. But I think it will be the reality in another year or two probably.

Przemek Chojecki: 1h 23m 14s

Yes. I think it’s interesting what you just said about these smaller communities, because I hadn’t thought about it this way, meaning that there are a lot of discussions about how Facebook is bad and we should go out and maybe go to something smaller, but then on the other hand, Facebook gives you also the protection.

It can harness your data. Of course it tries to monetize you because your data is valuable and tries to monetize on your actions, but on the other hand, it also gives you protections from bad actors, from spamming behaviors and stuff like that, to the extent it can. But with smaller communities, you don’t have this protection at all.

Courtland Allen: 1h 23m 56s

Yes. So we’ll see what happens there. This is all fascinating stuff. I would love to have you back on the show at some point in the future to give us an update on the state of AI and how indie hackers can take advantage of it.

But for now, you’ve gone through a lot of stuff. You’ve started a few companies that failed. You’ve started a few companies that seem to be taking off. What’s your advice for other indie hackers who are listening in who are trying to figure out how to get ahead, come up with ideas and start their own profitable internet business?

Przemek Chojecki: 1h 24m 21s

I think that the crucial thing is persistence and passion for what you’re doing, those two things. Basically you should work on something which gives you a lot of fun, a lot of pleasure on a daily basis, because probably you’ll be grinding a lot.

And even if things fail, and if you lose money, time, then you should still have pleasure and fun working on the particular project. Then take your time. It takes a lot of time to do great things and build great products.

Courtland Allen: 1h 24m 57s

I love that combination of persistence and passion, because I think if you’re passionate about something but you’re not persistent, you’re just going to try one idea and it’s going to be your passion project and when it fails, that’s going to be it.

But the reality is if you’re going to go passion first, it might take you several shots because you’re focusing on what you like but you also need to do what the market likes, so might you need to keep doing it until you find the one that works.

I’ve talked to so many founders who operated that way, Pieter Levels with his 12 startups in 12 months. He was always working on cool things that he liked. But it took him a bunch of months to find one that hit it off and he wasn’t afraid to cut the cord on one, but he was persistent and kept going with another.

Or Tobias van Schneider who always works on little passion products that he thinks are interesting, and guess what, a lot of times the rest of the world doesn’t care. They don’t think it’s interesting.

But because he’s combining that passion with also persistence and releasing lots of different things, eventually he hits on something that works and he’s in this great spot where he gets to work on something that he both loves and that the world loves and is willing to pay for just like you are. So I think that’s great advice.

Przemek Chojecki: 1h 25m 56s

Thank you. No, I think persistence is key here. Because as you said, if you have just passion it’s not enough to build anything and see the traction.

Courtland Allen: 1h 26m 6s

All right, Przemek. Thank you so much for coming on the show. Thanks for enlightening us as to the state of AI and what’s going on and what’s possible for people to do.

Can you let listeners know where they can go to find out more about what you’re up to at Contentyze and PetaCrunch and anything else you might have going on?

Przemek Chojecki: 1h 26m 22s

Thank you for reaching out. It was a pleasure. To find me, either you can go on LinkedIn and find me there, or you can go on Medium and find my articles there. I think it’s [email protected]. P Chojecki, that’s my surname. Or you can Google me and find all the resources there, I guess.

Courtland Allen: 1h 26m 41s

All right. Thanks again.

Przemek Chojecki: 1h 26m 43s

Thank you very much.

Courtland Allen: 1h 26m 43s

Listeners, if you enjoyed this episode and you want an easy way to support the podcast, you should leave a review for us on iTunes or Apple Podcasts. Probably the fastest way to get there if you’re on a Mac is to visit indiehackers.com/reviews.

I appreciate your support and I read pretty much all the reviews you leave over there. Thank you so much for listening, and as always, I will see you next time.