Scaling A Startup To $500 Million | Expert Advice From Co-Founder, Super Agent (EP# 48)

Ismael, thank you so much for being on our podcast. I'm so excited to talk to you. You're a multi-time founder, multi-time CTO. You studied chemistry in college. You cut your teeth in e-commerce in Sweden before you joined Qvideo. Qvideo? I may have pronounced that incorrectly, but—

Qvido. Qvido. It's like liquid, but with a Q.

Perfect. You were the founding developer there. And then, you know, from there, it's just like an incredible last 10 years of creativity. You founded 4 companies back to back. KeyCrunch, Bitcall, Mersenne, where you built the company for, I think, over 5 years, and now SuperAgent, which you founded in 2023. It's backed by Y Combinator. What's the journey here? You're— how did we get to now? A lot of companies. Tell us the story.

So basically, you know, I was always interested in science, technology, math. Math was something that I really loved and enjoyed. Ever since I was a kid. What I really didn't like was school, and I ended up, you know, I was so good at math that my teacher basically told me to help him teach instead of learning. So I usually just went around and helped my, you know, fellow students with their math, and I actually got my own little booth where people like students could come and I would help them out. This was like in senior high school. And so when I graduated, I didn't really know what I wanted to do. I wasn't at all interested in like entrepreneurship or building companies. I didn't know. I'm from a small city, you know, so there were everybody worked at the local paper mill basically where I grew up. So, you know, I moved to a bigger town. My brother was studying engineering in Gothenburg, Sweden, which is a small like industrial town on the west coast of Sweden. And I just, you know, worked at some cafes. And, you know, I always— what I remember is I always worked near the, you know, engineering university that we have here. It's one of Europe's finest engineering universities. And, you know, I would pull like night shifts and walk past the dorms and like lights were on, people were studying and I hated studying. So for me it was like, what are people doing? And then one day it just, you know, dawned on me that perhaps they're doing something right and I'm doing something wrong. And so I, and that's, you know, and so I had to pick like, what do I like? I like math, you know, chemistry was always very easy for me because of the math. And so I just decided to enroll in a bachelor, you know, bachelor program for chemistry. And when I did that, I started meeting a lot of, you know, other students at the university and I met this guy and he was, you know, he talked about like he would basically run a mentorship program for, you know, underprivileged kids. So he would go out to the poorer parts of town and help people with math. And I was like, dude, that's what I did, like my whole senior high school time. So I'm up for doing that. And, you know, I think that's where the entrepreneurship really came in. We started a nonprofit. It grew. I got to travel all around the world. I was like in Southeast Asia. I was over in the US, you know, and we went to the worst schools in the worst areas in the worst part of the world. And I met over, I think, 100,000 students. And this little nonprofit grew and it grew so fast that, you know, I wasn't, you know, I always wanted to do something more with it. And I immediately understood that we cannot build something huge if it's only nonprofit. So I tried to build like a European Khan Academy, but I wasn't, you know, successful in getting the other founders, so to speak, on board. So I ended up getting ousted from that organization because I was too profit-driven, basically. And that's where the, you know, that's where the entrepreneurial journey really started. And I actually dropped out of university because I was doing that thing full-time. And that was my first year at university. So I never graduated, but I met a lot of awesome people. And one of the CEOs of a big company, Volvo, actually, the car manufacturer that was supporting us with money, he told me like, dude, you should come and work at this place. So I went there and still doing, you know, still trying to entice students to study math, chemistry, enroll to the local university, stuff like that. And but again, I felt like it wasn't really what I was. I wanted to do something bigger. That was my whole goal, like to do something really big, something impactful, something that impacted a lot of human beings. And, you know, my first year I met this dude called— his name is Alan, and he went on to graduate, and I was doing a bunch of other work and stuff like that. And he went to Google, he worked there, and then he called me one day and said, dude, we should start a company together. And I was like, I'm all for starting companies. At that time, I had moved to Germany. I was working for a German company within e-commerce, and he was basically working with AdWords at Google. So, you know, he had an idea that, you know, when you have a lot of products in your inventory, let's say you are a big e-commerce player, you have like 10,000 products, it's really hard for you to advertise for all of those products. Usually you only advertise for your best sellers or a specific category. So he had this idea that we should build a software that could automatically create ads and, you know, follow up the advertising and basically capture the long tail of, you know, the advertising for a specific product inventory for a for a user. And I thought that sounded great, you know, and we started building a prototype. We got a couple of customers, we got some investment like $50K. And, you know, we were, we was our first company. I knew nothing about pricing, nothing about sales. What I knew was that I need to get, you know, revenue in order to survive. So one day I just took my phone I left the office at that time. We were around 2, 3 people in the company and I just, you know, went online, grabbed the list of 10,000 e-commerce companies in Europe and started cold calling them. And, you know, the first 2 days was, you know, brutal. I got nos all over the place. I was like, this is not for me. But, you know, I thought to myself, if I can manage to get a 15-year-old to study math, I should be able to sell this product to someone who really wants it, right? And at that time, Google had launched something called image ads or Google Shopping ads, as they called it. So it was like you could get your image on the front page instead of just text ads. And so I just called people and said, dude, do you want your image of— do you want images of your products, like on Google's first page when they search for, I don't know, you know, shoes? And the first one said yes. And then I started, you know, basically creating a pitch and it went well. And I hired my first, you know, team member to the sales team. And eventually we grew to, I think it was around 8 people. And we had, you know, in a year we got like 500 customers, Europe's biggest e-commerce players basically. And that was everything I did. I watched Wolf of Wall Street like every day and, you know, bought beers to the sales team and I was really amped up, you know, and I loved sales. That was, you know, I went from a very technical background to doing sales. So I really love that and it went well. We sold that company, was called KeyCrunch. That was my first company, real company, and my first success. Prior to that, I was involved in a startup called Quido, which did basically, we bought and sold account receivables and it was a trading business more or less. That didn't fly. So, but KeyCrunch flew. So it was really amazing to go through something, you know, building something from scratch and selling it and getting a bit of a bank, you know, from that as well. Life was great. And I thought like, let's do the next one. And since I was calling a lot of people, I wanted to, you know, automate that process. So I built a little dialer in my browser so I could just click a number on any webpage. It will call through voiceover IP, like Twilio or some API that we used. And when we got acquired, the sales manager of the acquirer like loved that solution. So I was like, okay, perhaps we can take that solution and build a company out of that. And that became Bitcall, right? And so we started that company, we raised some money, you know, tried to sell, we got a couple of customers, but it didn't really, you know, fly this time. And I tried everything. I tried sales, I tried calling people and, you know, the voiceover IP back then, this was like 10 years ago, the quality was shit. People were like, I can't call anyone, you know, they don't hear what I'm saying, blah, blah, blah. So that didn't actually work out that well. We struggled for, I think it was 2 years to build that company and it just didn't work out. So that was a lesson for me that like not everything is going to be super easy to scale super fast. And so that initial success was actually luck, I would say. Some of it was luck and I didn't have that luck. The second time around. But I also knew that I could never go back to doing anything else. Like, this life is for me. I'm going to build products all my life. But I didn't know what to do specifically after Bitcall. What I did know that me and Alan had been working together, you know, from university all the way up to KeyCrunch. And then we did Bitcall that failed and we found, you know, loyalty, which is super important in a founder. I think that's probably the biggest lesson there for people starting companies is that find the founder that is like a brother, you know, that, that will do basically what a family member would do for you. That, that should be the goal, right? So loyalty is super important and we had that and still do. And so, you know, I got contacted by a company and they were running an e-commerce and they asked me to do, you know, some due diligence on that company because it wasn't going that well. And I basically made a PowerPoint like a management consultant and I wrote like in the PowerPoint like this company is shit. What you need to do is throw out the management and replace the management. And then I was about to send that in and Alan told me like, okay, so if you give them, you haven't given them a solution, you just basically pointed out the obvious. And I was like, okay, so I'll add a new like a bullet point to the presentation. And I just wrote that me and Alan will be the new, you know, management basically. And so I sent that in and I got a response and they were like, yeah, let's go. And so we just, you know, from zero to one, we were all of a sudden we were met. I was the CTO. He was the CEO of a, you know, 100-people e-commerce company in Germany.

Was this Mersen?

Yeah. So the company was named something else, but we had to have like a consultancy company in order to be able to do that work. So we started Mersen as a consultancy firm.

I see.

And that, so that was the idea, us two just doing something until we got the next product idea that we wanted to, to, to invest in and work with. And, but, you know, we took over that company. It went well. We did a turnaround on that company. The owners were super happy and they just, you know, gave us more and more stuff to do. And we were like, we were only two people. We need to staff up. And the owners were like, yeah, go and staff up. You know, that's your part. And so we did that. And out of that grew like Mersenne became a, basically a software development consultancy firm for bigger tech companies in Europe. So we had companies like Klarna, Trustly, a bunch of companies that we were working with and it was really profitable. You know, the business model was profitable. I had a huge network of developers, really seasoned developers all around Europe and Asia and also in the US. So we were able to, you know, assemble a stellar team and that business was really profitable and profits is something that, you know, blinds you when you do consultancy because it's, you don't, you know, you're profitable day one and it just accumulates over time. The problem with consultancy is scaling it from 15 to 1,000 people is really hard work. That's not, it's probably harder than scaling a PMF product to that level. I would say, I don't know. I haven't scaled a PMF product to that. Size yet, but I would guess it's equally hard, right? And so when it's equally hard, you really need to think about, is this what you want to do the rest of your life? Do you want to wear suits and go to these meetings and do stuff that people tell you to do, but you don't really think is going to work? You know, and I didn't want to do that. So one day, me and Alan, we had, you know, that's the thing with loyalty. If you work together, you know, for a lot, many years, you don't even have to tell each other that you're you're tired of what you're doing and you want to do something new. So he took a week off, he went to Vegas to play poker. And we have, by then, you know, we had seen, you know, some embryos of AI. We had seen GPT-2, we had seen these early image diffusion models, and I was playing around with it like crazy. And everything I could think about was this new technology that was emerging. And I wanted to do something with it, you know, I didn't know what, but I wanted to something. So Alan went out, he went, he flew to Vegas, he played, he loves poker. So he was playing Texas Hold'em and I was stuck here back in Sweden and I didn't really know what to do. So I thought to myself, like, I'm building like KeyCrunch, our first company, the advertising optimization automation company. I'm going to do that, but I'm going to do that for social media ads and I'm going to use AI to do that. And I started building, it was, you know, really hard to build. The quality of the models were crap, basically. I made a landing page, or we made a landing page, and that landing page got 100,000 signups in like 4 weeks. So I applied to YC, got rejected, and we were really struggling technically with the product to get a product out that would actually work, you know. And while I was doing that, I was also building a framework for basically orchestrating all of these different models together. And I call that framework SuperAgent. It was like an internal framework that I had built. And since everything I was using was open source, I thought, yeah, I can open source the framework and get feedback from other developers and probably get help with building out the framework. And I had this idea that building with AI should be super simple. It should be equally simple or even more simple than building a regular web application like you did in the Web 2.0 days. And so I open sourced and it got huge traction, the open source framework as well. And I created Discord server, you know, the community blew up. We had a bunch of contributors and we were on all the, you know, VC basically product maps that were out there and we got a bunch of users, and you know, it really felt like we had something. And I applied again, or we applied again to YC, and this time we got in. And so this is, I think, yeah, approximately fast forward a year ago, a bit more, one and a half, one and a half year ago.

Why do you think you got in?

I have no idea. I have no idea. In our cohort, it was the, I think the, the acceptance rate was 0.8%. It was less than 1%. So I have no idea. I actually think the other idea was better than this one. You know, I saw a clear path on how I could monetize that one, which I didn't do with the open source framework. I have never built an open source company before. But I think it was luck again, you know, just luck. So we got in and here we are. So that's like the backstory. What I've learned is that building products, it's a game of statistics. If you try enough times, you'll make it. If you try enough times over a long period of time, you will definitely make it. And if you don't give up, you will definitely make it. I think that's like the gist of everything that I've learned during my, I don't know, 13 years building companies, right? So I think that's it. I think that's it. I actually did a chart once with all my successes and all my failures. And I had like, 3 successes and 10 failed products in there. So some, some were like, you know, 2 months, some were 1 year, but they still failed and we still kept going and we'll probably keep going until we're 60, you know.

Thank you so much for the story. I think what's, I mean, what really stands out to me is the breadth, right? Like what's so interesting about your story is, and I'll just play it back a little bit, like You start as this maybe sort of slightly rebellious kid who doesn't like school, who's naturally gifted with math, who has maybe kind of a stroke of luck with a teacher who's like, hey, come teach. This is a great way to fuel curiosity. This is a great way to actually care about something. And then that leads to this nonprofit which catches fire. You probably go through some degree of like emotional loss when you're ousted from that experience and when the founders don't see eye to eye on what this thing could be. And then boom. Like boom, boom, boom, success, failure, success, failure. And it's across so many different industries too, right? It's like e-commerce, it's ads, it's SaaS, it's AI now. I think it's just, it's impressive kind of the breadth of your curiosity over the last 15 years.

Yeah, I like to learn new stuff. You know, I indulge in learning new things. That's one of the, probably one of the things I love with ChatGPT is that you can ask it any question and you can learn anything on the fly. That wasn't possible before. So I think that, you know, The amount of learning is, or the speed that an individual can learn today is like 10 times the speed.

That's insane. Yeah, it's insane. The moment in time that we're living through. I am interested in that for a second, and that's when do you feel like you started to be really aware of the exceptional moment in time that you, you live in? When were you very aware of that?

I think it was 2022. 2022, something like that, you know, late 2022. I think that was the moment where I felt like this thing, whatever this thing is, it's going to change something. I'm not sure exactly what, but it's going to change. And like every day since that, since that, you know, insight has been just fulfilling that same thought, you know, it accelerates very fast, faster than anything I've been you know, I have seen before. I've never worked with crypto, which I would guess would have some of the same feeling for people that were in that space. But this feels like enormous, you know.

So let's talk about SuperAgent. So you kind of talked a little bit about how this all happened, but maybe take us into like what's really the core insight behind this product at this point.

The core insight, I mean, to start off, SuperAgent is still in its infancy. We have a couple of enterprise customers. We are really happy and working closely with them in order to achieve our goals and find a scalable product, what we call product market fit, right? My learnings so far has been if you want to build in the AI space, you really need to think about what value you bring to the table with your product and what value other players such as the model providers bring to the table, right? Because in the end, if you don't bring anything new, any new value to the table, you will just end up competing with huge companies with a lot of capital and a lot of brain, right? More brain than you can accumulate yourself, most probably. And that is a death trap, I think, for startups. So my journey with Superagent has been to find a place where we can provide value to the end user and find our spot in the market, basically. And I think that is the thing that I've learned. That's the biggest learning that I've got from Superagent and also a bunch of learnings from YC. I think I've learned a lot during the, you know, 3, 4 months that we did the batch about what it takes to build a company. And most probably, or even more rather, what not to do and how not to think when building a startup. That's, I think that's the value that YC.

Yeah. Share some of the lessons about what not to do. And then if you could follow on with that, I'm curious, like, what have you done to try to discover where your real value lies tactically, you know, literally on day to day?

Yeah, so I think that, I think the second part of the question, you know, comes a bit into pricing as well. But basically, YC tells you not to die. They tell you that if you are able to survive, you know, not reach PMF, but just to survive for a certain amount of time, let's say between 2 to 4 years, your, you know, success rate increases drastically. After those, if you're able to survive, the success rate over time accelerates basically or increases. And usually there are a couple of, you know, a couple of things that make companies die according to YC. And I would agree with those. The first one is that the co-founders split up. So finding a great co-founder will definitely help someone you work with, someone you trust, someone who's loyal, someone who's your brother or sister, you know. Figuratively speaking. And so that's the first thing, the, the co-founder, the team is super important. The second thing is hiring. What ends up happening is that you raise a round. We raise the seed round. Some people raise a lot of money, some people raise a lot less money. And what could happen is that when you raise your money, the money, you think that you've made it and you start hiring a bunch of people for roles that you're not ready for, you know, you're too immature to be hiring for those roles. And all of a sudden the money's gone and it's really hard to raise a new round if you haven't made any progress. So they tell you to keep the team nimble, small, you know, and I believe wholeheartedly in that as well. Keep team small in order to just survive until you reach some kind of product fit. So So I would say those are the two, two learnings. The third last one I would say is that if you are trying to build a tech company, if it's B2B SaaS or whatever it might be, it's all about the probability of success. So every decision you take can either increase the probability or decrease the probability. So let me give you an example. If you want to build a company from Europe, a huge company, right, that decreases the probability of success. So you could just pack your bags and move to let's say San Francisco, and that would increase the probability of success. And your role as a founder is to increase that probability of success. And if you actively know that a certain action would increase the probability of success and you still don't take that step, that, that would be really bad. You know, that's a bad sign, I would say. So what we are doing is that we are relocating to SF basically. In order to be able to increase the probability of success. If you look at it purely mathematically, which I like to do logically, I think that would be the third learning that I got from YC. When it comes to finding product-market fit, I think AI is very special because so much is happening and it's in its infancy. And if you look back on the internet when that was in its infancy,. And if you look back on what products actually made it out and how did they start. So let's take Google as an example. Like there's a bunch of companies that were hyped in the beginning of the internet boom. None of them are, are here today, basically. But there is this search engine that started off as a search engine for basically campuses, you know, at Stanford. And now it's one of the biggest companies in the world. And then there is this bookstore online. You know, that started off selling books and now it's, you know, the cloud of the world basically. So I think that you should be very careful, careful. And again, try to survive. Just try to survive is one thing. The thing that I've learned. And the second thing is to constantly iterate on your product constantly and get feedback from users. One way of getting feedback is to actually try to sell your product to a user.. I find that to be very valuable because if you try to sell your product and nobody wants to buy it, you know that you need to do something different. You don't— you perhaps don't know exactly what, but you know that the thing you're doing now is most probably not going to work later on either. So you know that you need to do change. And if you iterate fast, let's say you iterate every 2 weeks or every month in the beginning, to try to find whatever takes you to the next level, right? The next level might be 10 paying customers. Great. Then you've done something right. Now you got to get 100 paying customers, right? And the thing you did perhaps to get 10 customers might not work for getting 100 customers. So you need to iterate again fast in order to not die again, right? So iterating fast, shipping fast, shipping every day. Is something that every founder needs to do all day, every day if you're building products like us, right? And I think that that is probably the most valuable lesson I've learned. So in that case, like pricing doesn't matter that much because you're just trying to sell something, right? It might be cheap, it might be wrong, it might be whatever, whatever. You're just trying to get from 0 to 10 and then from 10 to 100. Then later on when you have 100, perhaps you'll, you get some MRR, right? And you are at that point where you want to raise another round to build, you know, more MRR. Then I think it's very valuable to start thinking about the business model more in detail and specific pricing plays that you can make that makes sense for the customer and makes sense for the product and the company, right? And I think that that's where you can get leverage from that, whatever change you might do to your pricing or to your business model. Because if you only have 2 customers, I mean, if it's not huge customers, the pricing won't help you that much. The one thing you should not do is to give it away for free. I don't believe in that. I don't believe in free. If it's just B2B SaaS, you're selling to another company, they don't use free stuff. They are used to paying. So they should pay. That would be instinctively how I would approach it. There are, of course, cases where that's not true, but in most cases, I've found that that is true, right?

I think those are very, very valuable lessons. Can you take us into maybe a real story recently, if you can, where you tried to sell Superagent? Maybe you sold it successfully, maybe you got a no, but what was the process like? What was the feedback? How did you iterate? And describe a little bit too, like how you talked about pricing when you were trying to sell it.

Yeah. So I can take our first enterprise customer that we succeeded in closing. So we were developing a framework for other developers to build apps, right? But we got a lot of inbound. We got inbound from, you know, a bunch of different type of users. One of these users was a guy that's working in, in, he's the head of compliance at a big healthcare company., right? 20K employees. And he reached out to me and he said, I did a LinkedIn post, you know, showcasing one of the features of SuperAgent. And he reached out and said, we need to, we need to have, we have a bunch of tasks and we are overwhelmed with all of the tasks that we need to do in the compliance department. And he said, like, I'm sure AI could do like 90% of these tasks. And what I did was that I just sent him the documentation page and was like, here you go. You can go and build your own AI agent. To do whatever task it is you want to do. And I never heard from him again, you know, and because my mindset was that my ICP is developers. It's not compliance people at some, you know, it's not C-level people at a big enterprise. It's developers that want to build stuff. 6 months passes, the batch ends, you know, and we are just trying to survive and iterate. And we have by then built, I think, 3 different products, you know, that we tried to sell. He, he emails me again and says like, that demo you did for me, I really, really need it. You know, I really need something. So I took a meeting with him and from there we were successful in building a product that they could use and it's now being implemented across the whole organization basically, which is a huge, huge thing for us, of course. And one thing when it comes to pricing, What we noticed was that it depends on your product a lot because in our case, we had to, you know, we had to really think about what is, what is the value that we are providing to, to that customer. And in reality, what we are doing is that we are replacing human beings with AI agents, right? So our pricing strategy was that, okay, so how many human beings are we going to replace? And we found a number. We did the business case, we found a number. Of human beings that we thought that we could replace, or, or other type of costs that we could, you know, get rid of, and we just charge them for that, you know.

So, so to be specific there, you're saying like, let's say we can replace 10 humans, 15 people, 15 people. Yeah. And let's say we pay each person anywhere from $50,000 to $150,000 a year. We're going to come up with a number, and are we going to charge that customer that number? Are we going to charge charge some percentage of that number so that we're—

Yeah, so we did. Yeah, exactly. Exactly. So we did a percentage, but I'm quite confident that you could charge the whole amount depending on what type of product and value you actually, you know, and what type of role you're replacing.

I want to dig into why you think that, but so let's— you come up with this pricing strategy. How does the customer respond to that?

They loved it. They loved it. Because the thing is that when you go to an enterprise customer, and they try to sell it within the organization. Two things happen. The first thing happens— the first thing that you need to do is to find the champion internally at that company that really loves your product. And we were successful in creating that champion by shipping a product that that champion used every day. That champion was successful in recruiting other champions internally., right? So we had a whole team of 3, 4 people that were using our service every day. And I think that what ended up happening was that when we were discussing the contract, they needed to do a business case internally to show higher-ups, the CEO basically, what value they could extract from this and how this matches their, you know, budgets and the existing costs that they have, you know, budgeted for.. And in that discussion, it was obvious that, you know, they weren't looking, they weren't seeing AI as something that they wanted to add on to their existing software stack. That's not how they look at AI. I think, you know, the previous iterations of B2B SaaS, you would buy like a software here, you would buy a software there, you would pay a couple of seats here, you would pay a couple of seats there. You might do an enterprise deal there. It didn't replace anyone. And the thing is that the whole promise of AI is to augment your team. You listen to Sam Altman, he's talking about AGI. What is AGI? It's basically human being doing work for you. So you don't have to bring in other human beings to do that work. He uses this analogy that he believes that there will be a billion-dollar company built up by one or two people. People leveraging a bunch of AI as, you know, their augmented workforce. So I think companies view AI like that as well. They are looking to optimize perhaps their internal cost structures or, you know, other, other, other things than just adding more, you know, adding a new software into their stack. I think it's very different this time around from, you know, Web 2.0, uh, B2B SaaS. And I think that that plays a big role in how you price it as well. Um, and I think that's a huge opportunity for founders. There is no other technology that allows you to, you know, um, augment a whole team. It might be the customer success team, it might be the sales team, it might be the compliance team or the law, uh, legal team. It allows them to do so much more with so much less. And somewhere there, I think there is a price. You need to think about that when you price the product because they are thinking about that as well, because that's what people have promised them with AI in the, in the first place, right? So I think that's very important. And I think that differs, that it's something different with AI products in general from regular B2B SaaS. Like Slack, for example. If you're selling Slack, you can charge seats. Nobody will ask any questions. But if you're selling human beings digitally, how are you going to charge for that?

Yeah, I find this to be fascinating. To me, I have many questions as opposed to lots of answers on this front. And so I'll kind of walk through some of where my head has been. Let's take, for example, what you— the story you just told, right? And maybe to kind of go back, it sounds like in pursuit of what is our value, what this customer helped you realize is our value, at least initially, is for chief compliance officers and really large companies who have an enormous volume of compliance tasks that could actually be done by agents, AI agents.

Yes.

And in so delivering that value, we could basically replace dozens of people depending on the size of the organization.

Yes.

Therefore, we should charge some percent of what the organization was paying all those people to deliver a cost savings. Do you see that?

That's a naive way of doing it. But yes, that's how we're thinking.

That's how you did it initially. And I think, I think for anybody who has paid attention to this podcast or been a founder, like one of the only things that's true is your pricing will likely change. So like, I don't anticipate that this is your pricing forever. Exactly. Maybe, maybe, maybe to that end, like, how do you see this evolving? Because I mean, I've been talking to founders building AI products who are going after usage-based billing, but then they're challenged by that because it's not predictable to the buyer and the fuel costs are coming down. I've talked to AI founders who are saying, look, I'm just selling seats because honestly, it's super low friction to the buyer. And like, they could go build a lot of this in-house. It's a buy versus build thing. And I need to reduce friction. And so seats makes a lot of sense. You're describing real labor savings and pegging that as your, as your price metric. Tell us a little bit about where you think this is going.? And maybe we can talk about it through the lens of the price metric itself, because I think that's like one of the fundamental things to get right as a founder.

Yes. I think the fundamental question you need if you're building with AI, you know, uh, I think the fundamental question you have to ask yourself is how are you going to charge for reasoning? Because that's what AI does, you know?

Okay. Go deeper.

What is the, Yeah. So again, I, I'm just, you know, I'm, I'm trying to take it back to first principles. Like AI is here to give us reasoning. It can augment a human being's reasoning. It can replicate the reasoning. It could even do it better than a human being, right? So what it, how do you as a founder that's building all of this reasoning, how do you charge for that? How do you charge for that reasoning? OpenAI charges through tokens, right? But you might use 2 tokens to do a task that's worth hundreds of millions of dollars to solve. So the, you know, the, I don't believe that that's the optimal way of charging for reasoning, right? Again, you know, the amount of tokens you use is irrelevant. It's irrelevant. It's the task that you are solving. That brings the value to the user. So the question I ask is, what are they paying today to get, you know, that becomes the next, like if we extrapolate reasoning and how you charge for reasoning, how do human beings charge for reasoning today? It's through salary. So you buy my reasoning by giving me, let's say, $300K a year and a bonus and a car. And healthcare, right? That's how you pay for reasoning today. And then the question becomes, how will you pay for digital reasoning in the future? And I don't have any answers, but I know that it's the reasoning part that's the valuable part. The rest I don't believe will be that valuable in a couple of years. And I think founders need to think about it in that way as well, because that was the promise of AI. The promise of AI was reasoning. It was not to replace UiPath with a new workflow builder. That was not the promise of AI. That's not why people are spending billions on GPU clusters. That's not why we have energy, you know, issues with all of these big companies that are hyperscaling their data centers to replace UiPath or some, you know, workflow management thing. That's not the endgame. The endgame is to deploy reasoning.

I've now done 50, 50+ of these conversations and you were the first person to make this point. So it's very original, it's very unique. I'm glad you make it because I think what it does is it points to maybe the new order, whereas the last order of selling B2B SaaS was selling automation, right? And like you were really, you're really trying to say, hey, look, the value is automation. Things will be easier, things will be faster, you'll get more efficient. We'll figure out how to price that. Maybe it's via seats, maybe it's via usage, maybe it's via platform platform fees, whatever it is. You're suggesting that there's a new paradigm here with all the companies that will be built in this next wave. And the paradigm is the value is reasoning. And it's very difficult to know how we will price reasoning. And our best proxy is salary.

Yes, that has been my thinking so far. It might be wrong and we might never get there. You know, there's a small chance that this whole thing will just you know, we've reached the peak. I don't believe we've reached any peaks at all. And I believe that we will be deploying reasoning, you know, on a very large scale across all of humanity in 5 to 6 years. And I believe that founders should be building around that. And yeah, so I'm not sure like how I've tried at all. I've tried seats. We've tried seats. People pay. Yeah, they pay for seats. That's not a problem. It's something they, they're used to. Usage? Sure, they can pay for usage. I'm not sure any of those are the right way of doing it, actually. I think that there is more to it than that this time around because the product has changed. It's not automation, it's reasoning.

This has been so fun and I hope that we do it again. The closing question that I always like to ask is What is the most formative moment in your career? Like it was truly an inflection moment that you view it as changed everything for you.

I think as a founder, formative moment is that you need to be, as a founder, you need to be very flexible in the way you think and what you work on. And you need to be able to rapidly shift because the world is rapidly shifting. It's shifting more rapidly for every decade. And if you are going to survive, you need to be able to throw out something that is actually successful to build the next thing that could be even more successful. And I've done that many times, many times. And I've seen founders who just won't do it. They just won't do it. You know, they will never abandon an idea. But my formative moment was when I understood that the idea in itself is nothing. Until it's proven. So, you know, you should focus on proving the idea. And if it proves to be wrong or false, you should view it as a scientist and just move on to something else, prove the next thesis that you might have. That, that is, that has been for me very formative. And I've lost a lot of colleagues that weren't able to do that switch, weren't able to, you know, do that context switch. And I think that is one of my strengths that I've built in these 12, 13 years or whatever it is that I've been building.

I love that story.

So I think that's it. Yeah.

Yeah, I love that story. Ismael, thank you so much again. Where can people learn more about SuperAgent and connect with you?

So superagent.sh and you can connect with us. Our email addresses are all there. You can connect with us on Twitter or X, sorry, and Discord. We have our Discord channel. Everything should be on the webpage. So it's just, you know, Superagent. And I post regularly on X. I think more people should share what they're building on day to day. And yeah, just reach out. Happy to help.

Thank you so much. This was so much fun.

Thank you for having me.