Chief AI Officer at GenAIPI on building a million-dollar business with Replit

We spoke with the Chief AI Officer of GenAIPI, who built a million-dollar business in months using Replit as their entire tech stack despite having no development background.

Our conversation explores how non-technical founders can leverage AI-assisted development to rapidly build and deploy production-ready applications, the learning curve challenges, and why speed of iteration provides a competitive advantage against traditional development approaches.

Key points via Sacra AI:

At a high level, how is Replit used at your organization today? What are the top few use cases?

Replit is used in many ways in my organization. My organization is called the General AI Proficient Institute, or GenAIPI.org. The entire website—every single functionality from the assessment tests, email systems, admin systems, courses, certifications, all the automations, everything—100 percent, including the database, is run through Replit. This means it was built in Replit, we deploy in Replit, we manage it in Replit, and we update in Replit. All of our analytics are there. Everything happens inside Replit.

Of course, we are connecting with APIs to other systems like Google Analytics, Facebook pixels, SendGrid, Stripe, and things like that. But the bottom line is our entire tech stack lives inside Replit, and we have no developers on staff. We have just hit about a million dollars in revenue this year with a product that was started in February.

We also use Replit to do all kinds of quick iteration and testing of new ideas. We can whip up a new MVP, feature, or idea in sometimes minutes, sometimes a few hours, or a couple of days if it's really complex. The bottom line is it allows us to iterate and test things at a rate that was previously unfathomable.

What's the biggest misconception you hear about Replit in business environments especially when you're talking to other founders or technical teams?

One of the biggest misconceptions is that Replit can't be used to create a finished product—that it gets you 80 percent or 50 percent of the way there, and then a real development team has to take over. That's what I hear, and I think that is a complete misconception.

The reason I think that is because I am actually making tons of money with my business, and our team is doing incredible things. We have finished products live with customers, including enterprise-level customers. I've personally trained businesses on how to use Replit, and I could point to a hundred different companies we've trained that are using Replit to deploy real technology—fully finished and operational.

Could it be better? Sure, of course. But is it good enough to make money? Absolutely. This concept that it's just good for ideas and then you have to hire a development team to finish it—I think that's a bunch of baloney.

Let's dig into something you just hinted at. Which is who's actually building in Replit. Who's hands on day to day? Is it the same people who sponsor or pay for it, or are those different groups?

In my company, it's the same. I'm the CEO, so I'm paying for everything, but I'm in there building actively every day and training actively every day. I started the company by myself as a solo person, and now I've grown the company to the point where I have to have employees just to deliver on some of the products that we're doing. The human interaction part is critical and growing, so I have to have more people. But I'm the one paying for it, I'm the one using it, and then, of course, my employees are as well.

You mentioned training other companies too. At those other organizations where you've seen Replit succeed, who tends to be the primary builder? Is it the founder again? A technical generalist or someone else?

It's amazing how many founders come and take my training and courses to learn how to do this stuff. So many founders are like me. My background—I founded a company back in 2016, and it was a tech company. I am not a developer. I don't know how to write a single line of code. So I had to raise initially $500,000, then a million dollars, then $5 million. I eventually raised $13 million.

I had to do that because before I even knew if my idea could work, I had to raise $500,000 so I could hire developers to build the idea, and then I could see if I could sell it to somebody. I thought I could sell it to somebody, but you don't really know until you actually build it. You spend all that money, and it's risky. You're risking other people's money.

But I had no other option. I didn't know how to code, and I wasn't going to spend the next few years trying to learn how to code. I just went, raised money, and made it happen. Now with Replit, I have the ability to just think of an idea and start making it. By tomorrow, I can go and see if someone wants to buy it or if someone likes it.

So many founders are just like me. They have an idea, a new feature they want for their existing product, or they want to launch something new. Or they're entrepreneurs who haven't launched yet, and they want to build something. I'm really finding a ton of CEOs, founders, and entrepreneurs who want to do it themselves. It's not necessarily the technical CTO. I think the technical CTO is kind of threatened by Replit. They want to use something more custom and fill in the gaps in their own way.

But the founders that aren't technical are saying, "No, Replit can take us through the ideation, the MVP, and let's prove it out." At the very least, they'll use it to build something and say, "This works the way I want it." Then they can hand that working product to the dev team or their CTO and say, "This is what I need it to look like."

I think it's so cool that founders now don't have to say, "This is kind of what I want. Go build something, and then we'll check on it later." Instead, it's "Here is exactly what I want. No ambiguity. This button should do this. This button should do this. This is the value I need to create here." I'm really seeing a lot of companies where the leadership gets their hands dirty with Replit, and they really see what's going on.

That's a fantastic breakdown. Especially appreciate the point about founders creating functional prototypes to hand off rather than just specs. Let's shift to how you chose Replit in the first place. When you were getting started, did you consider alternatives like Lovable, Bolt.new, Zapier, n8n, or even building a custom stack on Amazon Web Services? What tipped the decision toward Replit?

Replit was one of the first ones that I found. I learned about it from a LinkedIn post from somebody who was talking about Replit and how they had just watched somebody build something in 5 minutes. I thought, "Wait, what is that?" So I looked it up and really dove into Replit. This was a little over a year ago.

We dove into Replit and started playing with it. After I had spent a couple hundred hours building projects and playing with things, that's when I really started to ask, "What else is out there?" Lovable, Bolt, Base 44, and these other systems. I went and looked at them and tried them, and I just didn't like them.

Replit is the winner in my opinion. For someone like me who is a non-developer, it's easy enough to figure out, and I'll qualify that in a minute. But it's easy enough to figure out while being powerful enough to be really flexible to push all the way through to deployment. That's why I love it. I'm a big fan of Replit and would never consider anything else at this point.

Got it. Sounds like Replit struck a balance between being approachable for non-developers and robust enough for full functionality. You mentioned you'd qualify the ease of use point. What's the learning curve like for non-developers entering Replit? Were there barriers at first?

One place where Replit is working on improving—and they're even turning to me a little bit to say, "Maybe we can work on this together" because my training is really effective and efficient—is the learning curve.

I was a tech CEO for a decade, so I've sat in whiteboard sessions with developers, CTOs, and VPs of engineering talking about why we can't scale something and why we need to invest in certain technologies. While I can't develop and write code myself, I understand and am comfortable around that language.

Because I understand my way around software, when I clicked on the tab to open the database and looked at the tables inside Replit, that wasn't scary to me. I think a lot of people without that experience don't have that base knowledge of how software works, how the internet works, how applications connect to different pieces, and how you can run into scalability and cybersecurity problems.

The way I overcome that and help other people I'm training is by doing a 10 to 15 minute portion of my seminar where I teach them about coding fundamentals. What's a front end? What's a back end? What's an API? What's a database? What does Postgres mean? What is authentication? How do tokens work? Just simple high-level things that only take a few minutes to go over.

I don't teach how to code. I just teach what those pieces are for and how the puzzle fits together. By teaching how the puzzle fits together, they understand what they're doing and can really start to dig deeper into Replit and do some really fun stuff.

That's unique about my training—I'm helping make sure they really get it, not just blindly talking to an agent saying, "Build this thing," without actually knowing what's happening behind the scenes. I think it's important to at least understand at a high level why you're doing something and why the agent is building and structuring the app in a certain way.

What would you say is the single greatest strength of Replit for your business needs? And on the flip side, what's the most significant weakness or limitation you've run into?

The advantage is easy: it's speed. If I have a problem or a new feature or something I want to whip up, I can just talk to it as fast as I can talk or type, and it's almost done.

I recently launched a new university program for BYU that I'm working on to help students better understand AI. Gen AIPI up to now has been focused on enterprises, not so much on university students. But I agreed to do some things there, and I said, "I just need a new section of the website." So I went to Replit, told it what I was doing, and boom, I had a new section. It took me a couple hours to whip it up.

To be able to add something like that same day within a couple hours is just unheard of. In a traditional dev environment, that's taking weeks or months. When I'm competing against a company that's still using traditional development, I'm going to win 10 times out of 10 because I can move faster.

One thing that's difficult with Replit—and any AI coding platform is going to run into this—is sometimes you get stuck in loops. Sometimes something goes wrong, and if you're not a developer and you don't know why it's wrong, then you can get stuck and discouraged.

I've learned a lot of tips and tricks about how to push through those problems. Sometimes I've been stuck on one problem, some little thing that should have been simple for the agent to do, and I spent 3 hours on it before finally figuring out that it was working on the wrong part of code or on some duplicate app it created.

Knowing to look for those things is important. That's another place where beginners are going to get tripped up—they're going to have to go through some of the same struggles, and many people aren't going to stick with it. They're just going to say, "This didn't work. It sucks," and give up.

Learning how to talk to the agent is a skill. It's an acquired skill. You have to learn how to do it and how to push through those problems.

On the topic of supporting apps in production, how well does Replit actually support your needs once something is deployed? What's solid versus what's lacking when you're running real users through live systems?

The basic deployment needs are there—you've got databases, production and dev databases, which is a new thing that Replit can do now. It can handle two different environments, which is fantastic. It can handle deploying custom websites and domains.

Replit is trying to sell domains through replit.com through the interface, but they don't provide all of the DNS settings necessary for that to actually be useful at this point. So I'm still using something like GoDaddy or some other domain provider to handle DNS and things like that.

You also have to understand how systems talk together, like setting up email or Stripe. Those are pretty complex systems that are important for deployment. Many systems you build inside of Replit have some sort of authentication, some sort of login that has to be set up. If you forget your password, you need a reset password mechanism. Even for that simple functionality, you need an email system that can email the user with their password reset.

Right from the start, you have to be able to set up more complex systems that will allow you to deploy to the public and put real systems in place that feel complete. If you put out something that has a login system without a "forgot password" functionality, you're going to be in trouble.

Knowing that you've kept your entire stack within Replit so far, have you ever considered migrating to something like Amazon Web Services or another hosting provider? If not, why has that been the case?

I have considered it. But the reason I have not is because so far I haven't run into a real reason why I need to. I have hundreds of users and almost a million dollars of revenue in the past 6 months with Gen AIPI. I just haven't had the need to migrate.

Quite frankly, it's just so much faster and more convenient for all to be there inside of Replit. If I don't have to leave Replit, I'm not going to leave Replit. I haven't run into a functionality where I said, "Replit doesn't have this ability, I'm going to have to push out to something else."

I also think connecting to external systems is pretty easy. If it's well documented, then the agent can understand it well. Connecting to AWS buckets is not a hard thing to do. I would have zero fear of whether I'd be able to complete that task within Replit.

I might head in that direction at some point if I feel like I need it. But up to now, Replit has been able to provide everything I need inside of it. For payment systems and emails, there are some external systems you have to connect to. But in terms of basic databases, storage buckets, and the types of things you need to deploy an app, Replit takes care of it very well.

Let's go a bit deeper on that infrastructure layer. Specifically, how would you rate Replit's infrastructure support? Things like hosting, scaling, authentication, and deployment. Where does it shine? And where does it fall short for your production work?

I really don't like Replit Auth. I hate the Replit Auth system. The main reason is because I don't have any control over the branding and that whole experience, and it just looks and feels bad. I always say don't use Replit Auth—just set up a standard JWT auth and build it custom. I don't have branding control over everything else, and I just don't like that. That's probably the worst thing that I always avoid.

I mentioned before the hosting element. Hosting's fine. But when you want to push to custom domains and set up custom DNS, and have custom DNS for email settings, Replit doesn't offer all of the DNS settings you would expect. You can only add text records and maybe an A record or a CNAME record. You can't add other types of records that are important for DKIM records and SPF records—things that are going to be important to ensure you're not landing in spam inboxes. Replit doesn't have those capabilities yet. So I'm still managing DNS outside of Replit.

The auth system I'm not a fan of. It's not hard to work around, you just have to know. But I think a lot of people probably run into that and think, "I have to use this Replit auth system." They don't even know there's an alternative because they don't understand what an auth system is and how it works. That's something that could be better taught or more explicitly explained with better options.

How far has Replit spread across your organization? Which teams or functions are using it? Which ones haven't?

My company is unique because I was the first and only employee originally. I started the company with Replit, and now I've hired other people—we're about 10 people now. Everybody's using it. Everybody's been trained on it. If you want to work for us, you have to understand how to use probably about 15 different AI platforms, but Replit is the software platform and you have to be able to build stuff. So everybody in our company is using Replit.

Outside of your company, you've trained other organizations. Have you seen Replit spread broadly or is it usually siloed in specific teams?

I would love to see it spread more, but usually, it's just the people that I've trained. Companies usually put their head of marketing, head of social, COO, and other leadership through the training. Sometimes they'll do the whole company. There's one company I work with where they put 50 people through my training, and all of them have Replit accounts and build stuff. It's amazing.

But most companies I work with, I find that it kind of stays at the leadership level, and they're not pushing it down as fast as I'd like them to. I'm encouraging them. We're still learning and working with these companies, consulting with them, trying to help them really adopt AI. I'd really love to see it spread a little bit more.

How do you measure return on investment for Replit? What signals matter most when you decide it's been worth it for your business?

When I first built Gen AIPI, I did an analysis after I finished and launched the first version. By the way, I was able to launch the first version in 3 days. I started on Thursday, worked about 35 hours over those 3 days (12-hour days), and launched it. I got my first customer a couple days after that, and a couple months later, I was approaching $200,000 of revenue.

The amount of money that I spent in Replit to build and launch that was somewhere around $140. It was almost nothing for a very complex system. Because I've raised money and run teams before, I knew how to analyze what this would have cost in 2020 without any AI, with a traditional team.

What that revealed was that it would have taken a traditional team 12-18 months to do what I did in 3 days. Because of that time and the amount of people needed, it would have cost $2-3 million to launch Gen AIPI, and I was able to do it for $140.

If that isn't a perfect explanation of how the world is changing, I don't know what is. People say developers aren't going to be replaced. Well, I've built what could probably be sold for close to $10 million, and I did it without a single developer. That's significant.

Being able to build something that significant at that scale with so little time and money investment, which I also equate to risk—if you have to spend a lot of time or a lot of money, there's a lot of risk. If you don't, there's not a lot of risk, so go try things. Replit has dramatic ROI—it is the best ROI of any product I've ever touched in my entire life.

You had a magic wand and could improve one thing about Replit to increase its value to your team, what would you pick first?

I would love for Replit to just work a little faster. I know it's already working fast, but I have to type something in and I have to wait. I have to type something in and I have to wait. And I don't like waiting. What am I supposed to do during that waiting time? I could sit there and read what the agent is doing, or I can move to another task, or I can read a book if I want.

But bottom line is I want to be pushing forward, moving forward and going faster. I don't know if that's just more processing speed needed on Replit's side. If the agents and LLMs like Claude—if those are underlying technologies that we're waiting on responses from APIs, maybe that's out of Replit's hands. But even faster speed and more control over that would be amazing.

There are plenty of little things that would be awesome too. For example, I'd love to be able to work on two projects at once because I'm waiting. Instead of having to sit there and wait, I want to work on two totally separate projects. Right now, if I want to do that, I have to have two Replit accounts and two computers doing it because it uses port 5000, so it can't run two projects on the same computer or they lose connection to the project.

Within Replit, are you using specific agents or AI integrations like Claude, Langchain, or OpenAI for your workflows? How do those work in tandem?

I use ChatGPT quite extensively at the beginning. A lot of people will go into Replit and say, "Build me a weather app," and then let Replit get to work. I think that's a bad prompt—a terrible prompt, actually. Replit can do it, but the problem is you're leaving a lot of gaps and forcing AI to fill in the gaps for you.

What I'll do is go into ChatGPT and say, "I want to make a weather app. Let's talk through it." Then I'll ask ChatGPT to interview me about what I want this app to do. What is the brand going to be like? What are the colors going to be like? Why is this weather app not just another weather app? Maybe we're going to focus on extreme weather events and make it feel exciting. We're going to look for records and different things. Here's my unique selling proposition. Here's why people are going to come to this app and be excited about it.

I'll spend sometimes an hour or two hours working through that strategy, the differentiators, coming up with the right name, the right business flow, and all these things. I teach this in my course—really building that strategy. Then I'll ask ChatGPT to put together a business plan that has the monetization strategy, the vision, the mission, descriptions of everything, user workflows, and all this stuff. It'll be 2000 words long, quite substantial.

Then I can take that and paste it into the Replit prompt. Instead of saying "Build me a weather app," I can say, "Here's my whole business plan." I'll also create a user journey document, a branding guide document, and other documents. I'll grab all these documents, drop them into the initial prompt, and later structure them inside the project so Replit knows how to reference them.

This allows me to remain the boss, the creator. So many people just let the AI do things, and then it doesn't work or it's not exactly what they want, and they say, "This sucks." Well, of course it does—you didn't tell it what to do. That would be like hiring a developer, saying "build a weather app," then locking him in a room. That's not helpful. He would want to interview me, ask questions, and we'd write a plan before he starts to work.

You have to treat Replit like it's a person. Give it tons of information and context. You can give it even more context because it's an AI and doesn't care—it can read 50 pages really fast. I use ChatGPT for extensive planning before going into Replit. Then I use those documents as guardrails inside Replit so it's following my plan and strategy. When it's building a button or a menu or a workflow, it understands how that relates to the monetization strategy because it has access to the monetization strategy. So few people put that kind of information inside Replit, and that's why they get bad results.

Do you ever combine Replit with automation tools like Zapier or n8n in your workflows? If so, how do you decide where those fit versus keeping everything inside of Replit?

I like to do what I call "surf" on other people's technology. So yes, sometimes n8n or Make or Zapier are the better answer for certain workflows. If I want to connect Google Drive to Buffer or something like that, there's already a great workflow in place, so why try to recreate that?

I'm a big fan of using certain elements inside of Replit and sometimes saying, "This other system can do it better." And sometimes I say, "I really want better control over this." I'll also strategize with the agent. I'll go into plan mode or talk to Claude and say, "Let's talk about this for a minute. What's the best way to do this? Here's what I want to accomplish."

The cool thing with Replit is you can just talk about desired outcomes. I can work on that strategy and say, "If I need to accomplish this, what's the best way to do it?" I'll give the agent permission: "It might be easier to use an external service like n8n or Make for one of these automations. But if you think we can do it here, let's do that. I want you to help me decide what's the better way. Give me some pros and cons. Maybe this will be much easier because we've already built the infrastructure for it. But if we haven't, then let's use some of these other tools that are pretty user-friendly."

How do you handle updates and iteration cycles once an app is live in production? Things like managing versions, reviewing changes, rolling back if needed?

I'm just deploying when I'm ready to deploy. Once you deploy an app, I can keep making changes inside of Replit, and until I know it's working, I'll do really extensive testing before I deploy the new version. If I add a new feature, workflow, or something, I ask a lot of questions to the agent. It's like a QA process or unit tests.

I'll say, "Let's check. How's the login system? How's cybersecurity? Are we still compliant? Are we still optimized for SEO?" I'll go through and do a pretty in-depth QA process.

I also connect to GitHub so I can push versions up and make sure I have backups. There was a famous story not too long ago about a CEO whose Replit erased their entire database, and they lost tons of data. It was tough, and they eventually figured out how to get it back, but it took a lot of work with the Replit team. They posted about it on LinkedIn, and the damage was done.

I think that CEO is probably at fault more than Replit was because they didn't put the proper practices in place. I ask Replit pretty often to just make a backup of the database: "Please just make a backup." And there it is, ready to go. I'm pushing to GitHub and making database backups just by asking the agent to do it. That gets put into the file system and pushed to GitHub, so I've got all that if I ever need it.

You just have to do a lot of testing. Occasionally I'll get into a spot where I need to roll back because something didn't work. But I've gotten to a good point where I know what's going on. If it starts to go in a really bad direction, I will roll back immediately. In terms of rolling back an actual deployment, I've only had to do that once, and I quickly fixed it and moved on.

I do use the rollback feature in the chat sometimes. I might try something and let it go crazy on a complex task to see how it does. Usually it ends up breaking something if I don't give it a lot of guardrails. Then I'll just roll it back and say, "Now I see what you were doing. Here's really what we're going to do." I like that approach where I try something and then back up to do it more intelligently.

Any final thoughts or additional points you'd like to share about your experience with Replit or the text-to-app space in general?

I really want to talk about the difference between Base 44, Lovable, Bolt, and Replit. One of the things that Replit really excels at is how little it requires you to leave the platform. That's so important for the CEO that I'm training. If I have to train them on 4 or 5 or 6 platforms, they're going to get discouraged and probably not use it. But because I can train them on Replit, I can just say, "Look, you can find all this stuff. You just open a new tab, and here are all these options."

It's so much more powerful to be able to stay inside one platform. Base 44 is good, but it's relatively simple and not as powerful. It can't go as far. I would feel nervous deploying a good enterprise-level app from Base 44. If I'm just doing a basic website, then it could work. But because I'm trying to push real enterprise-level functionality through there, I need something more powerful. Replit is the winner for me—it just gets better results.

What formats or resources worked best to get people activated and productive with Replit? Was it templates, workshops, documentation, or something else?

I do live training, a lot of it on Zoom and sometimes in person. All of the training is proprietary to Gen AIPI, my company. I've learned how to train people so they can get it very quickly, taking them through great training steps and helping them understand that they can do anything with AI if they just follow the instructions.

All you have to do is talk and read. You don't even have to learn a new language for it—you speak in English, great; you want to speak in Spanish, great; you want to speak in Morse code, great. AI can understand it all.

Then I have them start building things where they say, "Wow, this is crazy. I can do this." Then I start filling in the gaps, introducing new parts of the system and helping them understand how they get charged, how they can optimize, how they can be more efficient, how they can debug, and how they can push forward until something is finally working.

My trainings are anywhere from 8 to 12 hours that I'll do with companies, sometimes over 2 days, sometimes all in 1 day, sometimes over a week. It depends on the deal. I love working with these companies and helping them figure out and gain confidence with AI.

Once they understand this general framework and gain that confidence, it unlocks so many things, not only in Replit but other AI platforms as well. They're able to be productive because they get over that fear of "I don't know how to do that" or "I'm not good at sales/marketing/product." You can use AI to backfill where you're not great. Use your strengths, and have AI backfill the rest. Once you learn how to structure those prompts and discussions and how to use different AIs to work together, it's so powerful what you can accomplish.

What about the opposite side? Where do you see the most friction or breakage in onboarding or retention for Replit? What's common feedback from clients or individuals when things don't click?

I think a lot of people just don't have the patience to work for 35 hours on something. It took me 35 hours to get the first version of Gen AIPI launched, and many people stop at 4 hours and say, "It's pretty good." I'll follow up with them and ask how their business is coming, and they'll say they haven't finished it yet. They probably haven't touched Replit in a month or two or four, and they just gave up because they didn't finish it.

I think a huge percentage of people using Replit are entrepreneurs. The right type of entrepreneur can do this, but a lot of people want to be entrepreneurs without putting in the work. Replit definitely makes it easier and can reduce the amount of time, money, and therefore risk involved in starting something up. But they don't have the actual entrepreneurship skills needed to take it forward.

Part of entrepreneurship is having this drive and passion that gets you to push through the hard times. But even people who do finish a product and get to 100% often get lost because they don't know how to be entrepreneurs. They don't know how to sell, how to market, how to run ads, how to ask people for money. A lot of that is because they didn't start the project by determining who the target audience is, what they're going to pay, what value they're going to provide, and why they're actually doing this.

People fail with Replit for three reasons: 1. They get in and feel it's too technical, so they can't really understand the architecture and get lost. 2. They start to figure it out, but they don't have the drive to actually finish something. 3. They build something, and then they're lost on what to do next—how to actually make money.

These are three very different groups with different solutions. Replit would do well to provide front-end basic tech training, entrepreneurship and sales training, and general motivation training. They need Tony Robbins in there to help push people through and actually make their dreams come true.

Entrepreneurship is usually about having some dream, some vision, some impact you want to make on the world, and you need software to push that mission forward. Most people make something and won't even post about it on LinkedIn or Facebook. They don't talk about it. They make it and then wonder how to get customers when they're not even talking about it. These things can be taught, and if Replit can figure out how to overcome those hurdles, they can explode as a company in a very good way.

Wrapping up, anything else you'd like to add that we might have missed? Anything you'd emphasize?

Let's talk about training a little bit. I don't think Replit has great training available. There are some tutorials and some people teaching (obviously Gen AIPI is teaching these things), but they're not doing what they need to do to really push the product to a non-technical audience.

That's something they're trying to work on—figuring out how to get this to a broader audience, how to get people to stick around, how to get them to push through the problems and insufficiencies of AI. I've been a tech CEO, so I understand when you're releasing a product that might not be perfect. It's okay if there's a bug. If something isn't working, I'll push through it, work around it, and figure out how to work within the quirks that Replit has.

Replit could do well if they could figure out how to provide more training and get people to actually use the training. Maybe it's in the onboarding or first-time usage. Maybe it's a button on the main profile that says, "If you need help, click here," which takes you to places like Gen AIPI that provide that training and other resources. I think that's going to be a critical element for Replit to make it to the next level.

If you could design the ideal starter template or project for a first-time user, especially a solo entrepreneur inside Replit, what would it be? What would it do? And what teaching moments would it include?

I think building a CRM is a fantastic, relatively basic thing to do. There are relationships, databases and tables, login and auth, probably some sort of landing page or marketing page, and you need some email integrations and different things like that.

Most people have a need for a CRM in their life. Maybe it's just their own personal Rolodex, or maybe they have a business and want to keep track of new leads and opportunities. Being able to understand how a deal and its data connects to a company, how that company connects to a person, creating a Kanban board, moving deals through stages and categorizing—this type of platform is relatively easy to understand in terms of its usefulness.

It would be a fantastic template for a first-time user or even a more advanced user if you want to get into more advanced things. It's a general-purpose piece of software that a lot of people could get tons of value out of as they think through how to build it, and it would allow for a lot of teaching opportunities inside the platform.

What specific elements do you think would add the most value in that training template? Like real-time tips, guided walkthroughs, live examples—what would help users learn most effectively?

I'm a big fan of real people teaching, or at least something that looks like that. Get a video of somebody talking through something and showing where to click, what to do.

One strategy that works really well for teaching is showing people how to do it wrong, and then showing them why it's wrong, and then showing them the right way. "Most people do this, but what if you do this? This is why this wouldn't actually work very well. A lot of people are tempted to do this because of X, Y, and Z. But if you actually do this, because you're thinking ahead a little bit, this is something that's going to work really well."

Those types of teaching moments where you show and tell with an actual voice—whether it's AI-generated or a real human walking through—I think that's super important and super useful.

Any final tips for unlocking unexpected value from Replit based on your extensive experience?

I'm a white water kayaker. I love to kayak down big things—Class 5 rapids and waterfalls and all kinds of stuff. When I do it, because I've been doing it for a couple decades, I make it look relatively easy. People will see that and ask, "How do you do that? How can you get good at that?"

You can watch as many Instagram videos and YouTube videos of amazing kayaking as you want. It looks awesome, and you can learn how to roll and do all these things. But until you actually sit in that boat with water pushing up against the edges, feeling how the paddle interacts and how you can manipulate eddies, jet ferries, streams, and all these things—until you actually get in there, you can't become a better kayaker.

I think that applies exactly the same way with Replit. You've got to get in there and just work it. I've spent thousands of hours inside of Replit, so I know how to use it really well. Even if I'm the best teacher in the world, if I'm sitting down with somebody who's brand new, they're going to struggle for a bit no matter what.

Figuring out how to inspire people, make it easy, make the onboarding simple, and providing resources right when they need it would be so powerful in helping people get more time in the boat. They need to spend time doing that. Teach them, have them do it, check it, teach them some more—back and forth. If you could create an automated system that did that inside Replit, it would be very powerful. But the reality is, to get good with Replit, you just have to spend time in the boat.

This transcript is for information purposes only and does not constitute advice of any type or trade recommendation and should not form the basis of any investment decision. Sacra accepts no liability for the transcript or for any errors, omissions or inaccuracies in respect of it. The views of the experts expressed in the transcript are those of the experts and they are not endorsed by, nor do they represent the opinion of Sacra. Sacra reserves all copyright, intellectual property rights in the transcript. Any modification, copying, displaying, distributing, transmitting, publishing, licensing, creating derivative works from, or selling any transcript is strictly prohibited.