Lessons From the Front Lines of Building an AI Startup

Welcome to Startup Success, the podcast for startup founders and investors. Here you'll find stories of success from others in the trenches as they work to scale some of the fastest growing startups in the world. Stories that will help you in your own journey. Startup Success starts now. Welcome to Startup Success. In this episode, I sit down with Ahmed Rashad, founder and CEO of Pearl Labs, a company building critical data infrastructure for the next generation of AI. We explore how Ahmed's experience at Scale AI, Amazon and McKinsey led him to identify a major gap in the AI ecosystem, why high quality training data and human expertise are becoming more important than ever, and what the future of AI innovation will really depend on. We'll also dive into fundraising in a crowded AI market and. And what it takes for founders to stand out today. Well, thanks again for being here. I think it would be helpful for listeners if we set the stage and if you wouldn't mind, you know, sharing your past experiences, your background with us and what led to the founding of Pearl. Yeah, absolutely. I. I started my career as an offshore oil driller and had a problem while working on the rig. Taught myself how to code, built a product, got acquired, became myself to a grad a little over 20 years ago. And it's just a sequence of chasing problems. It's been very, very straightforward. It's always been, I have a problem, I need to solve it, let's figure out how to solve it. Or I have a problem, I don't like what I'm doing, I need to find a way to make it go away and automate it. So let's build something to do that. Impressive. Impressive. So you must have felt this entrepreneurial drive back when you were, you said you started on an oil rig. Yeah, I wouldn't, I wouldn't call it entrepreneur. I know. I don't. Wouldn't know. If I called entrepreneurial, I'd call it basically out of a strong desire to not do the work I have to do, but I need to do, but I have to do it. So you find an easy way to do it. Okay, that's fair. You know, they say founders that solve a problem the closest to them do the best. So you're onto something, I think. I hope so, yes. So how many companies have you been involved in prior to Pearl? Let's see. I was at Oracle, McKinsey, Amazon scale than Perl. Wow, great. Okay, Good experience. So tell us about Perl and the problem you're trying to solve. Yeah, the. The problem we're trying to solve is teaching current AI Models, very nuanced niche skills. The reality is the models that exist today, they are great at generic problems or generic understanding of language or generic medicine and so on and so forth. And the more specific the problem gets, the more difficult and the more struggle, the more that the model struggles to actually respond and respond accurately. So in medical niche use cases, for example, models will make more mistakes than if you ask a generic stuff of oh, I have a headache, should I take Advil or Tylenol? Right. But the more specific it gets, the more trouble it has and the more error prone it has. And that is understandable because of how we train the models. And initially how we trained them was we give it a lot of information. And of course there is a lot more to it than that. But I'm simplifying and the models are now good at something, are good at that general. And they can understand, they can converse. Got it. So can you share? Was this problem close to you and that's what motivated you? Yeah, it was a problem that I was working on while I was at Amazon and get models to work initially. And then it was a problem that we worked at very closely at scale because obviously we did a lot of data for the frontier models. And after I left scale, I took a little break and I was thinking, I thought I took a little bit of time actually thinking about what am I going to do next? And almost every problem I was doing discovery on, it came back to do you have good data or not? That was one of the massive bottlenecks and it was a consistent bottleneck. So that's why I built Perl to actually solve that bottleneck of the very, very niche, call it boutique firm for data. Got it, got it. You know, that is where just in the work I've done with AI, that is where it falls down. You're right. When you get specific, when you need that kind of help. So can you share with us like a use case right now that you have? Yeah, absolutely. We do work across, well, a lot of healthcare, so medicine, dentistry, pharmacology, we do some science, scientific or stem fields like chemist, biology, etc. And we also do a lot of linguistics and culture. So think about linguistics, think about linguistics. Arabic, for example, Arabic is technically, when you look at dropdowns in select your language, it's one. Arabic is not one language. Right. There's standard Arabic and how it's written, but no one actually speaks with that. You only hear that on the news. Right, right. And then Arabic could technically be broken down into 22 different languages where someone speaking one language can't understand the person speaking the other language. Arabic is very, very broad and very forgiving. And the same object or the same thing can have multiple words. So you can actually construct sentences using completely different set of words to the point that someone from another region wouldn't understand what you're saying. Wow. So context switching, obviously, also, because when you're Arabic speaking, you also inject a lot of other words from other languages, especially English and French. Okay. And some Turkish as well, and you inject them. And that code switching is very confusing for models. So a lot of what we're working on right now is how do we actually make models capable of understanding, first of all, which dialect is being engaged on? It could be multiple dialects in the same conversation, by the way. So I speak an Egyptian variant of Arabic. There are multiple Egyptian variants, but I can understand someone who's speaking a, let's say a Beirut, Lebanese dialect. Got it. You're speaking in their Lebanese, I'm speaking to my Egyptian. And we understand each other. These are two languages. These are two different languages. You need to be able to tell which one is which for every word in the conversation. And then you need to be able to understand what is going on, what the overall conversation. It gets very dicey because some words actually have very, very different meanings, very different things, and some of them are used normally in day to day conversation in some places and are very inappropriate in other places. Oh, this is a perfect example of where you would need AI to get the. All these little different nuances and, and they're not typically going to do that. That's a great example. Yeah. Imagine having to do a note taker or having to build a note taker that actually handles that. Yes, yes, exactly. When did you found Perl? When did you start it? We officially incorporated in 2024. We started proper development in early 2025 and we launched in Q4. 2025. Okay, congrats. So so much has changed in the AI landscape just in that short time. How is that impacting, like what you're doing with Perl? So it's, it's, it's amazing, right, because the rate of development is just significantly faster and it just keeps accelerating. Right. So it's on a very, very steep curve and it keeps getting steeper. I think that has worked well in our favor because the need for that specialized knowledge, that specialized skill set has only accelerated. And this was a question that we got very early on, which is, well, AI keeps getting better. Why would the need for what you're doing not go away. And my mind, the hypothesis was, this is not how it works. The moment you unlock new technology, every single thing that you unlock, ten new problems emerge. Like, think about the problems we're solving today with global logistics because we invented the internal combustion engine. Right, Right. We did not have those problems when it was horses and carrots. Right, right, Good point. Yeah. Right. And then airplanes. Oh, fantastic. We have different problems. Right. So technology unlocks deeper, more sophisticated problems. And I think it's. The short answer is it serves us well because we are in the more niche, sophisticated boutique zone, or the more this accelerates, the more the need for our stuff comes in. Okay, that makes a lot of sense. Thank you for explaining that. Because you know that conversation right now around AI has shifted. Right. That things are developing so much that many startups are no longer relevant, that we're in the space because of the developments. But the way you explained where Pearl sits in the landscape, it makes a lot of sense that you will just get. Continue to bring kind of that edge to what you're doing. Yep, that's exciting. So in terms of that, so where are you in the fundraising journey with Pearl? Because a lot of the founders listening, who are, you know, working in an AI startup space, this is where they struggle. And if you wouldn't mind kind of sharing your experience, I think it would be helpful. Oh, it's fantastic. Right, right. So. So I, I believe that there are four things that you never, should never let go of as a founder. Hiring people have to interview everyone. Oh. And you're like, I can't let that go. Two, of course product. Right. Will probably be launching out there in the, in the market. Like, you cannot let that go. Three is you need to maintain a very close customer, so you need to maintain a very, very close eye in relationship with the customers. And four, you need to fundraise. No one's going to do it for you. No one's going to. A lot of fundraising is about relaying the spirit of the company to the investor. Right. And who better to relay that spirit? It's not about dumping fact, it's about relaying the spirit of the company. Knew better to do that than the founder. So a couple of few lessons that I learned from that is raise, raise when you can. It's about getting the right partners in. So don't just take money from, don't just take the highest valuation, just bring in the right money. Because whatever you get, even if you get like, let's say you're raising at a hundred now and you get a valuation from someone else at 110. If the people offering you the hundred are a better partner, go with them because the next raise is going to be at a billion. The a hundred to one hundred ten actually becomes very irrelevant if you're still raising a pre seed or a seed at whatever, 10, 20, 30 million valuation. An extra 3, 4, 5 mil isn't going to make that much of a difference versus having the right partner who could either help you grow or get in the way. I've been very lucky with my investors where they are very helpful. But I've seen other stories where it's where people just don't get along as well. Right, right. That's a great piece of advice. So what do you look for in a partnership with your investors? Like what are the certain things you're looking for? So a lot of. So investors, investors will try to help. They will tell you about the things that they can offer. A couple of things that are very important. First of all, just verify that they actually can. What they are offering is actually helpful for you. Okay, let's say they can help you hiring sales and ops people. But your product just needs good engineers that are very difficult to find. No one's going to help you find those engineers. You just gotta go find them. Right. So yes, they want to help, they can help in other areas, but it's not necessarily going to be helpful for you. So just verify that and understand that. Go in with eyes open. And frankly, the second thing is work with people you like. Work with people you like. Trust and respect. Build your relationship with the gp. Not just the associates or the, or the other partners, but build a relationship with the gp. That's good advice. You know, I've heard that communication is key in doing that. Are you pretty. Do you keep strong lines of communication open with your investors then? So yes, yes. And I, I do it in a couple of ways. So from from the beginning I borrowed a lot of this from McKinsey, by the way. Oh, okay. McKinsey learned from day one that you have to be over communicative with your team. But there's also a balance because obviously with, with McKinsey, the way my teams is that we had a check in and a checkout every day and we had a full blown checkout at the end of every week. That might be a little too much for investors. I like going in with the whole team. Handholding, right? Yes, yes. With investors. The way I do it is I publish a quarterly Update with a TLDR for those who don't want to read something long and then a full blown detail about all the key areas that we have. It's a consistent format so you can literally read it quarter to quarter and it literally maps out with just the updates. Ah, that's so smart. Your investors must love them. Appreciate that. I like that. I appreciate you sharing that because we have investors on the show who talk about they hate when founders go dark. Right. And they have all these resources that they want founders to leverage and they don't. And then they hear about a problem and it's too late and they could have helped and they like that communication and it sounds, sounds like you're, you're achieving that. It's working for you. I, I, yeah, yeah. And, and, and I mean every, the reality is from the outside it might look like it's a linear growth line. It is not, I promise you it is not. I, I know a lot of other founders and it is not. It is three steps forward, 12 steps back, 300 step forward, 15 max, then 12 back and then 16 back, then 51 ahead and it's just all over the place. Someone was asking me, how are you doing? And I told him I just had the best and the worst day in my life in the last three hours. Yeah, I think that sums up the life of a founder pretty well. So the communication, tying it back to the communication with the VCs, like some VCs are like, you're going to work with people who actually are interested in your problem. They want to be, they want to be invested and so on and so forth. I have a couple of those who we talk weekly. Right. And everything. And then others, I reach out when I need something and I let them know exactly what's my standing ask list of asks and they reach out when they have opportunities. Ah, so obviously that's a very different. Like obviously these folks don't know about. Oh crap, I had a customer delay. Right. It's going to be delayed and I'm, I don't have the money in the bank now, so I'm worried maybe it's going to go away and start spiraling. So the communication is for the other person. You cannot just use one style or one mode or one frequency of communication for everyone. The quarterly is the baseline and then you adjust or adapt that based on the person you're dealing with. Excellent. I think that's, thank you for sharing that. I think that's excellent advice for those listening a, not to just take the first check. But to think about the partnership and the investor and the relationship and then what you're doing to build that relationship and nurture it is so important. I know everybody listening will want to hear your thoughts just on AI and the landscape in general, just because you're in the midst of it. So I know it's always changing, but how are you feeling about things right now? I'm overall incredibly optimistic, like unbelievably optimistic. I think the challenge is that we have a lot of social media and we have a lot of media and it's about capturing attention and clicks and all of that. So when there, whenever there's, there's anything that could be remotely juicy news, it just comes and it, the media cycle is obsessed with it and people post and repost and add their spins to it and so on and so forth. You will hear a lot about, oh, this is the next thing. This is going to change the world forever. It's not. Right. Like you heard about this about drag, you heard about GPT 3.5, like fine, like it's iteration, caring, the next big thing, the massive breakthrough, all of that stuff. And yes, these are, these are good improvements and they're heading in the right direction, but it's not something that's coming out of the blue. Good, good clarification for people to remember that. It's actually a lot smoother then it looks like. If you look at it, it looks like it's just continuously. It's not, it's much, much smoother. It's steady pace of development, it's faster than previous cycles, mainly because we can use AI to accelerate a lot of the things that were not core to the problem, but we have to do them to get to solve the problem. Like writing code. I can write code now much, much faster. But that's the reason for acceleration. It's not that AI is growing its own consciousness and now it has three heads and so it's walking in their virtual humanity. That's not what's happening as far as I know. So, yeah, very, very optimistic. I think it's going to cause a bit of disruption. It obviously will, but I think it's going to create a lot more efficiency. Our economy relies on efficiency gains to be able to sustain itself. Otherwise we collapse. That's how we can actually command multiples on the market like we do. Right. Otherwise without efficiency gains, we're not going to be able to do that. Right. We always were betting on the future and in the future we're going to be able to do this. AI is the next iteration that allows us to do that. So that's fantastic. There's going to be some losses, of course, there's going to be some job losses and so on in the process. But again, just like the internal combustion engine. Right, right. People who are carrying people, the taxis, obviously they suffered like the saddle industry suffered, the horse industry suffered, but almost every other industry actually had significant gains. Right. And ultimately the people who handled the horses and saddles and the horseshoes and all that in the stables, they evolved into something else. Absolutely. Yeah. There's going to be a little bit of pain, but the gain, I think is significantly higher. Good, that's great to hear. Is there anything that you're worried about? There are. And this is, this is just based on my own personal stance. I think just like with any other technology, there will be good actors and there will be not so very good actors. I think a lot of not so great behavior is going to become easier. And just like a lot more people who never used to code or don't know how to code are able to build products today with very low coding knowledge. So the same way some people who like scams are going to get more sophisticated security and online security is going to get. Is going to get a little dicey for a while. We're going to make massive mistakes before we learn and adapt to the new reality. I think there are going to be. I think there are going to be a lot of new tactics that not everyone necessarily will approve of. So I think we're going to see smarter advertising for vices, for example. So there is going to be some negative impact. There are also going to be a lot of AI generated slop, which obviously is not very good for our brains, especially urban version. It's not good. Right. Imagine the volume of social media, useless social media posts that were there before AI became. Before Gen AI became super prevalent. And imagine now like it's just dull. Yeah, for sure. No, I think you're spot on on that. Yeah, yeah. And I think the last thing which is going to take some time to adapt to is AI slop in work. So I've seen this already a few times. We have something to do and someone just goes to AI and says, hey, do this for me. And they take the outp and just dump it. I think it's going to take some time to go through that because you read the first paragraph or so and it looks relevant and then you read the whole thing. It's like, this is absolute garbage. Doesn't mean anything. I think it's going to take a while to get people to adjust to what AI really is. It's not there to do your job there to help you be better and faster at your job. Absolutely. Really? Yeah. Really well said. And I actually even like how you refer to it as AI Slop because I've had people send me case studies and proposals and you're right, the first paragraph is great. And then it's like great. This is just an A. Yeah. And I think the other thing that people are starting to talk about is just like brain health. Right. Like our brain health with all that's out there and scams and things. So super interesting. You have a really good pulse on AI. We always wrap up the show with just general advice you can share to the founders. Listening. You're a very successful founder. You're right in the crux of, you know, excitement with the AI landscape. For those listening, any words of wisdom you can share with them? Don't do it. It's a trap. This goes back to your. You had your worst and your best day in the past three hours. I get it. Yeah. But if you have to do it, yes, you absolutely decide that you hate it, but there's nothing else you'd rather be doing or you love it. I don't know, maybe something's wrong with you. But be disciplined about taking care of yourself. That's good. Go to the gym, workout, hike. Whatever you do, just physically do that. Take care of yourself mentally. Take some time to read about things outside and interact in intellectual research and have a side project. They don't have to take that much but read, interact, do something outside of it. And third, take care of yourself spiritually. Like take some time and reflect. Pray, observe, be thankful. Whatever it is you do, it doesn't matter. But spiritually take yourself. Doesn't necessarily have to be faith in the specific religion, but just spiritually take care of yourself. And then last but not least, remember that it is a very lonely job, but you don't have to be alone. There are other people out there who understand. There are other founders. Create a community around you. No one survives on their own. Wow, really? That's great advice. I love that. I think the founder gets lost a lot in these startups, so I appreciate you sharing that. I know that will help some people listening. Where can we go to learn more about Pearl and what you're up to? Please, Pearl. AI, the website isn't necessarily the best, but it has good information and feel free to reach out to me. Reach out to me over LinkedIn or email. I'm always on. Thank you. It was a really fascinating conversation. I appreciate everything you shared. Thanks for being here today. I'm excited to see where Pearl takes us. That was great. Thank you so much. Kate. Thank you. You've been listening to startup success to make sure you don't miss out on Facebook future episodes, subscribe to the show in your favorite podcast player. Like what you hear. Tap the number of stars you think the show deserves in Apple Podcasts. For more tools and resources for your own startup success, check out berkeland associates.com thank you so much for listening. Until next time.