Replit Agent 3 is INSANE… Here’s 4 Ways to Make Money with It

For the past two years, the AI automation space has been split into two camps. You have tools that build apps, and workflows. And that division is about to end. A new player has just entered the arena with a move that could make standalone tools like Lovable and Nad obsolete. So, today I put Replet's new Agent 3 through a brutal head-to-head battle against its biggest rivals to separate the marketing hype from reality. And in this video, I'm going to be giving you those results. You're going to see where it wins, where it completely fails, and I'll give you a clear blueprint on how to actually make money with its current strengths. So, what makes Agent 3 a

potential game changer? Well, it comes down to two massive leaps forward that we've seen with this new update. The first is automated self- testing. So, if you think about it, we've come a long way when it comes to code generation with AI. Like, we started off with just tab completions to now we're building entire apps in just a single prompt. But these apps were always just a first draft and they were often full of a ton of bugs that you had to manually go through and figure out yourself and then go through a phase of follow-up prompting to try and get these issues fixed. Well, this massive new update from Replet with Agent 3 is designed to solve this because it doesn't just write the code. It actually spins up a virtual browser. It clicks through the app like a real user, finds its own errors, then it goes back to fix them all in a continuous loop. It's just like having a junior QA engineer built into the agent to help you identify and fix these early bugs. So, this is the first massive leap forward we've seen of being able to spin up a virtual computer, test it software, and then fix those in an iterative loop. And then the second huge leap is just the increase in the duration of the periods of which it can do this in. episodes and a massive increase in autonomy. They've gone from supposedly 20 minutes to over 200 minutes of unsupervised runtime. And the significance of this massive increase in autonomy to hours and hours of working on a software project for you autonomously is like you could be out for a walk and you have moment of inspiration and then you pull up the replet mobile app which I've just downloaded and it works really well and talk into it and just dictate an idea for an app, put it back in your pocket by the time you come back from your walk an hour or two later and the idea is that it should be functional and ready to go. So the gap between idea and actual execution is collapsing with these incredible advancements in autonomy. But the biggest strategic move here that Replet Star making is a direct play for the AI automation market competing with giants like NAT, Make. com, and Zapia. They've done this by adding their own agents and automations builder, which is essentially the same as NAT's new prompt to workflow builder where you can put in a prompt and it's going to build out an entire workflow for you. I mean, Zapia has had this for a very long time. NA10 built out their own one. And this is essentially Replet's own version where you can go prompt and explain what you want the agent or automation to do. And Replet agent 3 is doing this by building on top of Mastra, which is their framework for building and orchestrating the agents in the back end. And one of the most exciting features about this new AR agent in automation builder is their new Slack integration. So they've been able to turn the absolute nightmare of building a custom Slack bot into a simple and guided setup process, which opens up a ton of new use cases for people looking to sell AI agents to businesses. And so this rapid and easy Slack integration alone is a huge deal for anyone looking to sell internal AI tools to businesses, which we'll be touching on a little bit more later. So on paper, this Replet Agent 3 is a huge step forward. But is any of this real when you actually put it under pressure? So I put Replet Agent 3 headto-head against Lovable and Google's Gemini across a series of increasingly difficult software development challenges to see where it shines and where it breaks. So first up, I wanted

to test a standard internal tools, a CRM for a pet grooming business. And right away, the difference was pretty clear. Uh when I gave the same prompt to lovable in Google, while they were way faster at spitting out a minimal and barebone CRM in minutes, Replet took a lot longer taking its time and figuring out what it wanted to do, testing itself, but the result, the final product was on a completely different level to these other platforms. Replet's agent actually thought through the real world needs of a CRM. It built a more fleshed out and functional application because it was testing itself along the way. And it I mean it did take literally hours and so it is slower, but it's much more thorough. So for this round of building a CRM for a pet grooming business, Replet was a clear winner. But what happens when you give Replet a task that it self- testing features can't

handle. So I asked it to build a 3D game. And here the tables turned around because the automated testing is optimized for web apps and it's actually not available to do on 3D game design. So without this testing superpower, RIP agent really struggled to build a basic 3D game. Whereas lovable, which just focused on the pure code generation, produced a much better result and in a few minutes. So the winner here by far was lovable. Then I got a bit bored and

wanted to test this thing on an actual idea for an app that I've had for a while, which is an app that's going to take audio inputs, transcribe them, and have multiple AI features to be essentially your own AI coach. And I found this was where all three of these platforms still hit a wall. Now, this app was supposed to be an AI performance coach that takes in morning and night check-ins from you where you go in and you press a button and you record your voice and say, "Hey, this is how I'm feeling this morning. " And you do it again at night and do that every single day. Then you can run that up into a weekly performance review and you can click a button. It's going to take all of your daily summaries. It's going to create a weekly performance report from the AI. Then you're also able to chat to this AI coach and it's going to have all of the different transcriptions and check-ins that you've done as context so that when you ask it questions about your performance or maybe what you can do to improve, it has full context of all of that feedback and reviews that you've done. And it was with this application that I was able to find some more holes in Replet's testing abilities, which is it couldn't test the audio functionality of the app because it couldn't press the button and record into it. And it also wasn't very good at setting up the AI components that I needed like summarizing with a day or summarizing the week. Plus, it took a very long time and the final product wasn't a oneshot success. I'm sure I could have worked with it a little bit more, but it really proved to me that the highly complex and multimodal AI apps, these kinds of tools still need a human in the loop. They're not going to be able to oneshot things as you'd really like them to despite having this testing capabilities. So, for this round, there really was no winner. And so, I was pretty disappointed that out of all of the platforms, I couldn't get this to work. I even went to Claude Code and tried that and it still wasn't able to get it working. So maybe this is just a little bit out of the scope of what these kind of tools can do at the moment. So while the main app builder has its limits in terms of speed and also testing itself on certain features

the standalone agent builder for Slack is one of the standouts for this new replet update. I decided to get a simple use case for a Slack agent which is going to be able to give me real-time stock data within its Slack channel. If I'm honest, the guidance setup for this was flawless. It handled the API keys, the permissions, and everything. I've set up a lot of Slack bots and I can tell you that this is the most simple and straightforward experience of setting up a Slackbot that I've ever had. And after all the setup and prepping the app, it wasn't able to immediately deploy the app so that I could use it in Slack. And even to right now filming this, I haven't been able to get it working within Slack. So, while it showed a lot of promise initially, the actual testing and functionality within Slack, I still wasn't able to get it really working as I expected. So, yeah, it's pretty fatal error, but these issues I'm sure are going to be sorted out very soon. Um, but this is a very powerful way to deploy these AI agents in Slack, which is a place where you have millions of businesses basing their internal communications out of it. So, I'm going to give them the benefit of the doubt that these are just like launch issues. Or maybe I'm just stupid, but it doesn't seem to work no matter what I do at this point. So, after putting Agent 3 through the ringer, here's my final verdict. Let's start

with the good stuff. For standard web applications like CRM or internal tools, I think this thoroughess that it has and the self- testing ability of more regular web applications is absolutely unmatched right now. It produces higher quality and more reliable end products than its competitors. And the autonomy window is so much larger. I let it run for like 2 hours um to get some of these things done. And also, as we saw, the Slackbot integration is a genuine game changer. These things are a massive use case, but they've always been a really tricky thing to execute on. And I think with this is going to really open up the door to the explosion of AI agents within Slack. But as you now know, it is not perfect. And the bad is that it's significantly slower than the other tools. even when I'm just wanting maybe a quick prototype spat out, uh something like Lovable or particularly the Google Gemini and the AR studio was so much faster to just get something quick done and up and running. So, it's deciding which apps to build with this is important. If it's a really important project, then yes. But if it's just a quick tool that you're spinning up or a very brief MVP where you just want to get a visual of what it might look like, then this thing is going to ultimately take a lot longer and cost a lot more because Replet's costs are a lot higher than other applications and I'd probably go for some of the other options on the market. But that's potentially an opportunity for Replet to make a fast mode where hey look don't spend all the time that you usually do just get this a very quick version of this to me quickly rather than feeling like I'm going to get bogged down for 2 hours just for a basic app that I could get in 5 minutes from another platform. And so with Repent 3 you're trading at raw speed for reliability. And as the 3D game test proved it struggles with non-standard applications where it self- testing superpowers uh that are really one of its biggest strengths provide no extra advantage or even in the voice transcription use cases I was trying to use it for as well. And that brings us to the ugly. And here's the most important takeaway from this entire video is that the two halves of this platform, which is the AI app builder, like the vibe coding platform and the agent and workflow builder, they don't talk to each other yet, which is what I was really hoping for out of this. This is the most exciting thing for me because you may have seen people connecting Lovable to NATM backends and connecting them that way. So you get the front end set up on these AI application builders and vibe coding platforms and you can build the AI functionality on the back end. And I think Replet is very close to being able to cross those two over and making the all-in-one platform, but they don't have it yet. And I think they have a long way to go when it comes to uh being like an NAT or a make killer when it comes to building automations and agents. They don't have multi- aent capabilities. You can't tell it hey I want to have executive agent that is we're talking to multiple underneath it. Something like relevance AI which has really complicated AI workforce capabilities right now. I think knowing their track history of execution will definitely get there but there is a long way to go. and when it comes to their workflows, you can't trigger it by event yet. And that's kind of like the whole thing of building automations is like an event triggers something. So, very early days with it. I think there's a lot of potential here. And given their ability to execute, I'm not going to bet against them. But they do have a lot of work on their hands to catch up to where the rest AI automation marketplace is. But when they do have that, don't get me wrong, when you have the vibe coding side of things and you have the AI agents and automations and the ability to cross them over into the same application or have them standalone, I'm extremely excited for that. And I really think that's the most important thing that's happened here is that now we have the potential to have a platform that does both. So as with any

new technology release, we need to be looking at where is the opportunity for us. Where can we build a business? Where can we make money? And that's what we're going to get into in this next section. In my opinion, the single biggest opportunity for the average person to get in to make money with these kinds of vi coding tools is to sell internal tools and software to businesses. Nowadays, you do not need a technical background to start small and with small businesses and get familiar with building things like custom AI image generators for their market teams or branded knowledgebased chat apps for the new hires for onboarding. All these things can be easily built through something like Replet Agent 3. And these internal tools are incredibly valuable. Companies just prior to now didn't have the money to spend on building them. And now, thanks to things like this replet agent 3, people like you can go and sell those to them for $500, $1,000, $1,500. And we are really going to see the gradual encroachment of software into every part of businesses taking away old clunky spreadsheets or forms. And we're even seeing this in my own companies right now. I'm using it. A lot of my team members are using it. And it's like, hey, why are we using this clunky set of tools now? We can just build our own custom interface, custom app to use internally. And even people who are not technical can do this stuff right now. I've just done a whole interview recently with someone who's selling these kinds of custom software systems for $15 to $20,000 per pop because it's replacing the $3 to $5,000 per month software stack. that most small businesses are running on. At that price point, it is a bit more advanced than what we're talking about here. But this video will explain how to start it as a beginner. So, I'll link that down below for you if you want to check that out. But mark my words, building these internal tools for companies and particularly AI tools using these vibe coding platforms is one of the biggest opportunities heading into 2026 and anyone can get into it. The next one is a super lowanging fruit and that's building Slack agents for companies. So, as we saw with this new release has created much easier ways to deploy them when they work. But your offer to a company can be very simple. It's like I will supercharge your company Slack with a team of AI agents for your team to use in automating repetitive tasks. And the way this works in practice is that when you have these agents active within a Slack channel, a team member can go at stockbot like the example that I was building and say, "What's the current price of XYZ? " And it's going to be able to call a tool, get some information off the internet and pull it back in. Or for example, you might have maybe a script writing agent and within Slack, you can go at script writer and then paste a link to a YouTube video and say, "Hey, make our own version of this script. " And it's going to pull the YouTube video. that's going to pass it through a tool that you've created and then be able to pass that straight back into the Slack. So, having your Slack come alive with all of these agents in it is something that companies are really interested in right now and it should be a very easy sell if you know how to get in front of the right people. And we're only really just seeing the start of this, but it's definitely something to get good at now because when Riplet adds more features and integrations and even multi- aent capabilities where you can have one agent connected to a whole workforce of other agents underneath them, this is definitely something you want to get into now, not in a year's time. And thirdly, you can sell advanced websites for businesses. So, I'm not talking about basic brochures or static landing pages. Most businesses do have these kind of boring static landing pages that do nothing. So, in this case, you'll be going to businesses and offering them a website that has some kind of interactive tool in it. You can say, "Hey, I've just redesigned your whole website and I've also built in a custom calculator or an AI powered tool or a chatbot that will actually engage with your visitors and generate more leads. " So, and the reason this is such a good money-making opportunity is because you can identify people in your local area who have ugly websites that aren't really optimized. You can rebuild them on something like Replet Agent and go to them and say, "Hey, here's this thing that I built you. It looks way better and it has way better lead capture functionality. Would you be interested? Like, it's 500 bucks. " And you can show them the result right there. And that's what makes the sales so much easier. And the fourth and final opportunity to make money with this new tool is to of course teach people how to use it. There are two markets that you can sell this to. You can go to consumers as a complete beginner. You can go in and find out how to get the most out of replet agent. You guys watching this video are case in point. There's interest around these tools and what they can do. If you can learn just enough to get good and be dangerous and find your own way of getting incredible results out of it, which might only take a week or two, then you can start creating content on it and make some kind of course teaching people how to get the most out of repent. And the second market is maybe off the back of that you start selling that same training to businesses where they can empower their own internal teams to be able to build their own tools. whether you're selling that to their development teams, which maybe isn't the best angle if you're nontechnical, but if you are a non-developer and you've figured out how to get incredible results with this, then packaging that up into a course and going to businesses and saying, "Look, I will teach your staff how to build powerful internal applications without writing any code. " And giving that power to all staff members in a business is going to allow them to create some really awesome things from the bottom up. And now, here's what I wouldn't recommend, which is trying to build some kind of scalable SAS on top of Replet agent and then make $10,000 a month with it. really just not what these platforms are good for right now. The reliability of a public-f facing mission critical app just really isn't there. And the money is going to be in building these internal tools, the Slack agents, selling high value services through them. So stick to where these platforms are strong, but trying to make some micro SAS that's going to have thousands of users on it, payments and authentication. There's easier and lower hanging fruit you should go for first. So is Replet agent 3 a NA10 or lovable killer? Not yet in my opinion. But this is a huge step forward in autonomous app building and I think they are a platform that you should be keeping your eye on. Now, the direction of this market is clear that we're going to have these all-in-one platforms that combine app building and vibe coding with AI automation, workflow building or vibe automating is the future. And when these platforms start to cross over, that is a really exciting point for us, which I think Replet agent is closer to doing right now. So, if you're interested in this kind of stuff and want to learn how to build AI apps with these kinds of tools as a complete beginner, I have my complete 2hour course on doing that just up here. But aside from that, guys, that is all for the video. Thank you so much for watching and I'll see you in the next