202 | The Ultimate AI Showdown: Comparing Vibe Coding tools with Mark Kashef

0:00Welcome to part two of the Ultimate AI Showdown. If for some reason you missed part one, go back to episode 200 of this podcast. We hosted Lisa Adams, and she was doing the ultimate AI showdown, comparing tools for doing data analysis for businesses. It's absolutely brilliant, so go back and check that out. And now to part two, comparing VibeCoding tools with the amazing Mark Cash. In the next few years, AI technology will change our world dramatically. Whether you are a business executive trying to catapult your business forward, or just somebody who refuses to be left behind and want to advance your career, this is the show for you.

0:45I'm your host Isar Maitis, a serial entrepreneur and an AI enthusiast. You'll hear invaluable practical tips from innovative business leaders, AI practitioners, and some of the brightest AI minds in our world today on how you can leverage AI in ethical ways to advance your career and grow your business.

1:08our next topic today is maybe the most exciting aspect of ai development that exploded in 2025 and that is vibe coding now those of you who don't know what vibe coding is and you never heard the term before first of all the term was coined by andrej carpathy who's a big name in the AI world, who's been one of the founders in OpenAI. He left OpenAI a while back, but he's still followed by a lot of people. But what it actually means, it means using AI development platform and explaining to it in simple English or other languages, if you prefer, what is it that you want to develop, and the platform will develop the thing for you.

1:51Now, when I say develop, for non-developers like me, that's a pretty broad term. Well, depending on the tools you pick, it does the design, it does the development, it does the QA and solving bugs, it does the deployment, it creates an application that you can actually use. And again, not all the tools do all these things, but just like we spoke before, you can mix and match these tools to create applications that you can use. Now, the cool thing about it, people like me who are not coders, who are not computer engineers can develop applications for anything that you want. If you are a software engineer, you now have rocket fuel in your back pocket to do a lot more stuff a lot faster.

2:35Now, the interesting thing here is that if you know what you're doing, you can do magic. If you don't know what you're doing, this could be really, really scary. But the cool thing is if you know English, or like I said, many other languages as well would work, you can write applications that will solve either basic or complex problems in your business. But there's a wide range of tools right now that do a huge variety of things, and they each have a specific target audience in mind, and they still have pros and cons. And so our guest today, Mark Kashif, comes from a background of data science.

3:15He is a data scientist, a software engineer, but he became a vibe coding ninja. He shares priceless videos on YouTube on how to use AI tools to do really incredible stuff. A lot of it around vibe coding, but not only that, but he's definitely somebody to follow if you want to go that next step in your AI development. And he brings together both worlds. One and a deep understanding of how software actually works and how data needs to be analyzed. And on On the other hand, he's really, really good in explaining it to monkeys like me so I can follow and do the same stuff and develop my own applications and do some cool things with it.

3:54And so from my perspective, Vibe Coding is the next version of superpowers that AI gives us after we learn how to chat and learn how to build some basic agents. Now you can use it to actually build applications that you need. And why would you want to build applications? because every application you use today has between 20 ,000 to 172 ,000 functions when you need four. And yet you use the really complex application because that's what you have. Well, that's not the case anymore. You can create your own applications that do exactly what you do and connect exactly to the data that you need and display it in exactly the way that you want consistently.

4:33And so you can start creating applications as of today. and we're going to do this with the amazing help of the amazing Mark Kashif. And so, Mark, welcome to Leveraging AI. Thank you so much. That's a very humbling intro. The bar's way too high from where I actually am, so I appreciate it. That was underplaying it. Like, what you do is really amazing, so I appreciate it. I appreciate it. Right on. Should I share my screen? Yeah, let's get started. That's why we're here. All right, cool. So, hey, everyone. I'm going to walk through three main tools right now when it comes to vibe coding. There are, if you look at probably now 100 copycats of these tools, but these are three of the most powerful ones.

5:17And this basically builds on top of what Lisa just showed in the last segment. So she showed at the language model level, at Gemini's level, at Claude's level, how to build these applications. These are basically what we call a vertical agent that's on top of those models. What that means in plain English is Bolt, Lovable, and Replit here in order. They've all built infrastructure on top of these language models and tried to give as many tips and tricks and guides behind the scenes to make these into junior software engineers. And that's why you can Vibe Code with these in a lot more powerful way than you can with just Cloud or Gemini or anything else because they have focused on marrying software engineering principles and infrastructure with language models.

6:04And just to give a visual, instead of just using nerd speak, if I had to give you a visual of what this looks like, imagine you have Lovable, Bolt, and Replit. And behind the scenes, behind the curtains, you have the real star working away, which is Claude. And for most of them, similar to what Lisa mentioned, Claude is one of the best, if not the best right now, at writing code, especially production-level code, for software engineering. So behind the scenes, this is all powered by a brain, and right now all of them are using Cloud4, which is the latest and greatest version of Cloud. If I were to give you a quick comparison before we actually go into and see how they work and how they operate and the types of apps they can make in literally minutes, this is a quick comparison.

6:50So Bolt is really good at being fast and speedy, and the actual code it writes is pretty well thought out. Lovable is the most beautiful. So if you look at the design, it wins design every single day in terms of beauty. In terms of ease of use, it's also really easy to use as well. So if you are starting out completely from scratch, Lovable is the easiest way to go from zero to idea on paper that you're really excited about. Bold, you have to work a little bit more on the prompting side to get there, but Replit is one of the most robust. So imagine Replit as having its own repository of tools at its disposal that are all Replit's infrastructure.

7:29What that means in plain English is Replit actually used to be like a cloud provider before, but now they've kind of built on top of their database tools, their server tools. So Replit can not only build apps for you, but it can natively integrate databases, everything, without you having to worry about going the next step and learning new tools to connect to these other providers. So Replit is one of the most robust, but it also takes sometimes four business days to finish. And I'm exaggerating, but from the time you send a prompt, it could be 15 to 20 minutes before you actually get your first iteration.

7:59So if you're less inclined to know what you're doing or how to prompt, you'll get a way faster feedback loop with bold and lovable and really get your skills upskilled. Once you're ready you can graduate to try to work with Replit because then you'll be able to plan out your prompts a lot more effectively and have that quicker feedback loop you're looking for. And then this is the last chart I'll show before we actually go into the platforms and show you what I'm actually talking about. So in terms of MVPs, which stands for Minimum Viable Projects or Prototypes rather, Bolt and Lovable are exceptional for you being in the middle of a meeting.

8:37Most of the times when I'm speaking with clients and I'm trying to visualize what their dream outcome looks like, I will use Lovable on the side. I'll take the transcript of my Fireflies, dump it into Lovable and say, try to visualize what this client's actually looking for. And I'll have that run on the side while we're having the conversation and they'll be able to show it and say is this kind of what you're thinking so the power of being able to use that speed is very impactful and similar to what Isara mentioned you you want to know when it's time to bring a knife to a gun fight and when to bring a nuclear bomb to a gun fight and it really depends on which type of tool you want to use so if you want to build something highly robust in a quick amount of time repli is probably the best way.

9:19If you want to build just something quick, a mock-up to get the ideas flowing, Lovable is a great option. So without further ado, let's actually take a quick preview of what these tools look like when you log in, and then I'll show you some apps I already built out. And I ran it ahead of time just for time's sake, but for Bolt right now, this is what it looks like. You'll notice all of them have a big glaring text box in the middle of the screen where you can ask away your dreams in terms of what you're looking for. You can attach an image by the way to all of them and say here's a website I really like or here's a design I really like.

9:52I think you can paste up to four to five images, send it with a prompt and say try to recreate this to the best of your ability. And then one key feature here is this button if I say something really poor like build me an app to automate my manufacturing business with a surveillance dashboard. I'm going to misspell it on purpose. You can see here they have something called enhanced prompt. So if you're getting started, this will take your very vague prompt and make it more production level ready. So this will also help you become a prompt engineer in a very compressed short amount of time because Because you'll start reading and understanding the flow, how it expects to be spoken to.

10:36Now, is this perfect? No. And I'm going to leave as a little teaser at the very end here. I'm going to show you how you can use perplexity to make you a prompt engineering deity without actually learning the fundamentals yourself. But until then, you would just send this and then it would actually pop up the first response. One more thing just to add one short thing, because you said about attaching files. What I do every time is I have like a brand guidelines, fancy PDF. I upload that and then it knows how to grab my logo, my colors, my font and everything. And you don't even have to say anything.

11:10You literally just upload it there and then it figures out what to do with it. And it works on all the different applications in a very solid way. Exactly. Yeah, no, that's great, Adan. Thank you. And on the lovable side, same thing. You can upload files in terms of images as well. You can connect the whole workspace. You can connect to something called Superbase. This is basically a database that a lot of SaaS platforms use to store data, but also enable functionality, which I'll get to shortly. And this one's pretty straightforward. You send the prompt as well, and you'll get some form of user interface.

11:43And the last one is Replit, where this one's slightly different. So Replit, on the left-hand side here, you can actually click on this, and you can select exactly what kind of platform you're trying to optimize for. So if you want to build a modern web app, you can just click this and it will know behind the scene, it'll get a cheat sheet injected behind the scene, that that's exactly where you're going for versus having to guess. If you want to just pick auto, based on what you write in natural language, it will guess what you're trying to build. This is the only one that I'll try to run live just to show you how it's different from the others.

12:14So if I use the prompt we just made in Bolt, okay, and you run it here. like I said this will take a while to get to the point where it the interface is there but one thing it's exceptional at is it plans out loud and it plans a prototype before it even builds the prototype so it tells you basically this is kind of what I'm thinking this looks like is this kind of where we're at or should I pivot my thinking before I even build the first version and the reason why it's so thoughtful you can see right here it's now building a visual preview of a prototype of my prototype to see if we're in the right direction.

12:51It's giving you a list of all the items it's planning on adding as features. And what's also cool is even if you didn't come up with a feature yourself, it tries to suggest features that could be richer. So Replit is, like I said, can be a much richer experience overall. And then it always asks for your permission, which I love. I feel like an actual boss to a junior software engineer where you have to approve and start the project in order for it to actually go and build it out. and once this finishes running we should get some form of interface here that will display you'll see right now production dashboard there's something loading here it's building these components as we speak right in front of me there we go we got some modals here and by the end of this I can say like this looks kind of boring or I wish it was something else you know and once I click approve and start then it'll be on its way it'll probably take around 10 to 15 minutes because one thing replit likes to do is it tries to make everything functional not just build a mock-up it tries to actually make the databases work it tries to pop up a modal that says you says you want to use open ai great go get me your api key you want to use stripe go get your stripe key so it's very comprehensive in its responses so that's what i wanted to show as a preview before we actually look at apps.

14:09Now I'm going to show you the prompt that I've used. I'm going to show you two sets of apps. This prompt is eight pages. Is it because I'm so inherently gifted and smart and intelligent that I wrote an eight page prompt? No. I actually dictated a very small brain dump to perplexity and I gave it a prompt that I'll show you shortly and it helped me build a very comprehensive prompt and the crux of the prompt is build a comprehensive AI powered IT operations command center that combines real-time infrastructure monitoring cybersecurity threat detection and gamified collaboration in a modern interactive web app and then it goes through a bunch of requirements here that again I would love to take credit for but all of these eight pages of requirements and designs I just approved by going back and forth with perplexity so I took this exact prompt and I threw it into each platform to give you a sense of how it would look in different platforms.

15:08So numero uno, Bolt. I purposely just showed the code instead of the web app. In Bolt, it took around probably five minutes or six minutes, and we came up with this, which is actually pretty beautiful. It's responsive. It has pretty much all the elements I asked for. You can notice here that this part looks really similar to what we just saw in Replit. There's a reason for that. a lot of these vibe coding apps use certain what are called libraries like a normal library you think of but let's say in the coding world where you have a series of frameworks or where all of these platforms like to get icons from to populate these icons here and it put together not only one tab but if you click on multiple tabs you'll see here security center you have security alerts obviously this is all fictitional data that you can make production data but it starts out also drafting out what the profile would look like etc and this is only one shot now with vibe coding I like to say that you're only as good as your last prompt so the most important thing when you vibe code is you'll notice this is my very first prompt is every prompt I send to these platforms the same no the first prompt for any of these platforms is kind of like building a house so if you had a construction team working for you and you told them the blueprint of the house right if you don't tell them at the beginning that you want a pool in the middle of the courtyard of your house and they start planning the building the house going back later halfway through the app and telling them oh I want a pool even though there's already two rooms there in a kitchen and the plumbing's already been put out that that becomes a tricky problem to go through so if you don't want to pivot halfway and now you have to do a change order like you would in real life and they have to go scrap a bunch of code you start getting what's called spaghetti code where now you have all these types of codes on top of each other and you go into this endless loop where you're not accomplishing what you're looking for you're getting constant errors and if you're not as tech savvy you can't even read those errors and you keep clicking try fix praying that it knows what to do and then you eventually say this vibe coding stuff sucks bye to prevent that from happening you want to be as thoughtful as possible before you even kick off your journey.

17:26And then the likelihood that will be more accurate will be much higher. So that was bold. So to pause it just for one second and connect to what you said before about perplexity, I do the same thing. I have a project in Claude, which is I did a lot of research on what needs to go into product design. And then I add that as the input into that thing. And then what I do is say, okay, this is my idea. I want you to ask me multiple questions to punch holes and figure out the details because I don't know the details. The other thing is it has all the things you need to think about when you design a software, like the user interface.

18:03Does it need to be mobile or not mobile? Like all the gazillion other things that you want to be able to do, and then it just asks me all these questions. Now, some of them I don't know. And I said, well, I don't know. What do you suggest? And then it makes a suggestion, and I can pick from its suggestions. But think about the AI as your software design assistant that most likely knows about software design way more than you do, even if you're a software engineer like Mark, definitely if you're somebody like me. Now, I have the benefit. I ran software companies for almost 20 years. So I understand.

18:37I don't write code, but I understand how software is built and how it works. But if you don't, it's still fine. You can still have that conversation back and forth. And only after I have a long conversation and then I have a very solid prompt, again, like Mark said, explaining that needs to be a pool in the middle of the courtyard, then I go into one of the tools and actually let it run. And the results are significantly A, faster, B, better, and C, like Mark said, for people like me, significantly less frustrating because I don't know how to fix it. And I'll say one more thing about the whole getting stuck in the loop of not being able to fix.

19:12When you're doing this five or six times and it doesn't solve the problem, just ask it to go in a different path. Because you don't understand how the code runs. You know what? Forget it. Let's start this component from the beginning. And then just that part starts scratch. Instead of trying to fix 15 different things, your chances of getting it done is significantly higher. Exactly. That was amazing. You know, one last thing, sorry. Most of these tools have like an assistant that doesn't actually write code that you can just say, OK, let's brainstorm this for a minute inside the tool itself. Right.

19:47I do this in Replit all the time and then it doesn't write code. It just helps you kind of like troubleshoot and come up with new ideas on how to solve the problem. And then you go back and try to edit. Yeah. You stole my thunder. Yeah, I was going to show that next. So awesome. Perfect. No, you're good. So I was going to show the app first in Lovable and then Replit and then show that functionality. So in Lovable, that exact same prompt I showed you, if I put this back, you'll see visually, although it's somewhat similar to what I just showed you, it is ever so slightly prettier. It focuses on the little things like the size of the font.

20:18You'll notice Bolt here. It kind of put a lot on one screen. And I have a bit of overload, cognitive overload when I take a look at it. So even though this is pretty, there's a lot going on. There's titles, there's subtitles, there's headers. So lovable is very lovable in terms of the designs it creates. And as you go from tab to tab, you'll just notice that it's very easy on the eyes. Let me see if this one works. Let's see this one. And you'll see, by the way, if there's an error in the code, you'll always see this pop up that says try to fix. So either you can look at the logs, if you're a nerd like me, or you can click on try to fix.

20:51And it will try to figure out what is happening in the underlying code. And then last but not least, this is what Repl.it put together. So Replit, I will say, will always be slightly less pretty than both those platforms. But in terms of functionality, you'll see it jumps ahead. It asks permission. It's asking me right now, how does the application look and feel to you? Are all the features working as expected? So it's literally waiting for my feedback to tell it, yeah, you did a good job or not a good job. The other ones are kind of like, here, take it. Here's my report. Let me know if you like it.

21:22If we go back to what Azar was mentioning, all of these tools have something that's called chat mode. In chat mode, I can actually ask a question like, what do you think we should add next? And then instead of actually executing and writing code and making the application more bloated, it just talks to me. I'm talking to Claude now, but I'm talking to Claude in the context of my project. So if it comes up with an idea or a plan, I can read through it. If it looks logical, all I have to do is click on implement this plan right here. And then now it will go and execute it. But at least now we're on the same page.

22:02I noticed that a lot of times that I will spend 70 % of my time, even in lovable right here, it will be called chat right here. If you click on chat mode, I'll spend 70 % of my time here and only 30 % of my time actually writing the code. because I'll spend five, ten messages to make sure we are 100 % on the same page before we go and write the thing or do the thing. Then Replit has the same thing as well and then Replit's just really good at dealing with feedback in general. All of these have also visual editors so if I go to here I can click on this button right here and then I can go select a specific element and you'll notice on the left hand side it's now referring to this exact element.

22:44So instead of praying that I send the prompt and it knows exactly what I'm talking about I can say no this one this is what I want to change same thing with lovable you can click on visual edit and I can say this button this buttons ugly and I don't even have to send a prompt sometimes if I want to just change the size I could just do it here without a prompt and that's how a lot of these platforms differ ever so slightly you can't do this exact functionality in Bolt, but in Lovable, you're allowed to be specific. I can even, by the way, Google a specific font name, paste it here, click save, and it will make the change in the code, which is very powerful because it saves you from having to guess how to tell the thing in natural language.

23:25And then in terms of Replit, I don't think there's, is there a prompt helper? I don't recall that myself, but at the very beginning, you can get some help going back and forth on flushing out the prompt. And one thing I just want to show you on Replit, again, small nuance, is you'll notice it actually takes physical screenshots of the web application while it's processing, whereas the other ones are solely depending on code. So where I say it's more robust is it's not just looking at its code, it also is taking screenshots like I would send to OpenAI or Claude and saying, analyze and see what's wrong.

24:02Let's see before the user catches it, let's catch it ourselves. And what I mentioned before in terms of being more robust, you can notice here if I go on database, deployments, all this stuff, I go to all tools, all of these tools are pre-built in Replit. So if you want a database, if you want the ability to use Google Sign-In very easily, all that's built into what's called Replit authentication. They've kind of made one whole ecosystem of tools. If I go back to Bolt and Lovable, if I want to enable something like Google sign-in, I'm going to have to go to something like integrations here, and then I'm going to have to create what's called a Superbase account, which is like a database account, and I'm going to have to connect to that Superbase account, then give it some credentials, then it can start to build functions that allow me to do authentication.

24:48So that's just one small microcosm of using these tools, the differences. These can get me a prototype immediately without me having to think about anything in depth. Replit will allow me to go much deeper, and And another cool thing about Replit is they have a mobile app. So if you're on the bus, if you're on a plane, if you're waiting in the airport, I could have built this on a mobile app and just prototyped on the go, vibe-coded while waiting for my delayed flight, and then I'd be on my way. So that's just one more idea. And the last thing is on the Replit side, I've used Replit as a developer for years to deploy applications on a server.

25:26The fact that it's all built in, again, in that one ecosystem, that means that they're really good at cloud, they have secure cloud, they're already trusted by hundreds of organizations, and now they're just marrying their whole ecosystem to cloud. So that's where I'd give Replit the edge on robustness. Now being mindful of time, I'll see how much time I have left. I'll just show you the hack in Perplexity. If you're looking at all this, you're like, I'm still stuck at how you wrote that prompt. I'd rather unblock that for you before we end up. So in perplexity, and you can use whatever language model you wish that's connected to or hooked up to a web search.

26:02So you can, I usually use deep research, but just for time I'm just going to use a very quick search and what I'll do is the following. I'll just dictate in natural language so you can see how poor I am as a prompt engineer. So I want you to act as a prompt engineer with a specialty in product design as well and I want you to look up all the documentation, watch YouTube videos on the platforms Lovable, Replit, and Bolt, and I want you to come up with a universal prompt that I can create for my e-commerce business that would work for all of them and build a beautiful, minimalistic e-commerce app that lets me manage my Shopify orders programmatically, just using my natural language, typing in some form of text box, and I want you to output that prompt in an unformatted code block in Markdown so I can just easily copy paste it.

26:54So in this case, in plain English, I just said go and watch YouTube videos, go and read documentation on all these apps, and it will know that it's using Cloud behind the scenes. So in a way, it's also going to research that. And when I send this, you're going to see what it's going to do. It's going to go through lovable documentation. It's going to go through replet documentation. It watched the ultimate guide on, oh, I don't have to go back. My bad. It watched the ultimate guide on using Replit and Bolt. So let me just try that again. Worst case, I'll just do it. Oh, there we go. So let me go back to what it did.

27:31Go to sources. So it went to Reddit. Went to YouTube videos. It watched videos on Shopify integrations. It went through GPTs. It went through GitHub, which is basically a repository for different apps. and then it came up with, after doing all that, this prompt. And you can see here, this exposes my lack of genius because it did all the hard work. Build a beautiful, minimalistic e-commerce web app that connects to Shopify store and allows me to manage orders programmatically. And then it gives a list of all the different requirements. Clean, modern, minimalistic UI with responsive design and basically a laundry list of all the things we want in this app.