How He Makes $118K a Month Selling to AI Agents

It's 118K, probably in two days it will be like 120K. That is Nevo David. He's one guy selling a social media scheduler and it is important to keep in mind that he was stuck at $20,000 a month until he repositioned to sell to AI agents. You can shift it to agent in a minute. We got 7.2 million views. And today he is breaking down his entire playbook. So by the end of this episode, you will know exactly how to make your product something that agents actually want to use and how to get it in front of the people who are paying for them. So let's get into it. All right, Nevo. Someone watching has built or they're midway through building a startup and they keep seeing this thing online about agents are taking over, SaaS is dead, and they may be understandably worried that they might be falling behind. What is the one thing that that person watching this conversation today is taking away? That SaaS is not dead. Agents are taking over, but it's going to take some time. We are very early to this, so we should definitely be there right now, especially when it's going to blow up to the rest of the market, not only for tech Twitter, basically. And I'll give you all the opportunity in building agents for SaaS. And I'll also explain what is this opportunity, basically, because people don't understand how much things they could do right now and make so much money with it. And then I will also explain how to actually do it step by step. Okay, awesome. And if you had to sort of paint out, how big do you think the shift of building for agents, not just building for humans, is actually going to be for builders who get it right in the next few years? It's definitely growing because we also see big shift from big companies doing it already right now. We can take even banks like Mercury that actually led people to selling to agents. I think it's going to be bigger and bigger, but I still don't see it in the next, let's say, two years or something like that. Because when I even go to my dad or something and ask, what do you think even about Claude? What is Claude? I know ChetGPT. So, you know, it's still going to take some time. But the good thing is actually that there is a lot of people in that early adapter niche space that do this. And they are like really underserved, basically, because we just like starting, basically. So this is like a good opportunity, especially for now and maybe in the next year. You've benefited off this a lot, right? You built Post-is, your startup, for agents predominantly. Again, we'll go into more of this in detail. But you grew Post-is from $20,000 a month to over $100,000 a month. And now it's what, doing over $120,000 a month? I mean, it's $118. Probably in two days it will be like $120. It's like $1,000 a day. But you went from $20,000 to $100,000 a month in like three months, right? Yeah, exactly. Insane. Can you show this to us? Do you have a dashboard or something you can share on the screen? Yeah, absolutely. So people will actually be, I think, a little bit freaked out when they see also how much churn I have. So this is just like how much it's showing the opportunity, basically. So as you can see here, this is my dashboard. Not still in $120,000, $118,000. Almost at $1.5 million per year. Right now it's only me, basically. So one person made it. It's pretty crazy for me. Because if we were talking about this like four months ago, I would tell you I'm planning to finish the year between 30 to 40K MRR maximum. Never imagined that maybe I should plan for 200K by the end of the year. It's a bit insane. But the interesting part here is the churn. So people might not believe. I have 14.6% churn. That's very, very high, basically. And usually when you have that much high of a churn, it means that you need to feed yourself like 10 times more customers to keep up with it, basically. And this is what's happening. Like in Post-it, in every single moment, I have around 528 trials, which is a lot because it's almost in every single moment. And this is the opportunity people can have when they are dealing with agents, basically. That's so genuine. Just after Joe, it was that February 2026? It's wild. I know we're going to talk about what exactly happened there. But you've made, you're now making $118,000 a month, predominantly through building for agents, not for humans, right? So you're breaking down that playbook today. Absolutely. Where does someone start? They're trying to build a startup for agents to get ahead of the game. Where are they going? First of all, I want to say something that is like very, like at least for me, if I just built it right now for agents, it might not like break out because, you know, I already had like I was on $21,000. It's not nothing. I already have a little bit of a brand. And I suggest everybody like start building whatever you can as early as possible because you're going to have opportunities along the way, basically. And it's a lot easier when you already have something and, you know, agents are out. So you just like do something small and you can start like you can shift yourself to agents. Basically, it's much, much, much easier. Now, the thing that happened basically is that it all started with OpenClaw. And for me, it was a no brainer. So I saw OpenClaw and I said, OK, let's make posties for OpenClaw. But it's actually I didn't start to make it for OpenClaw just because it's agents. Because for me, you know, we all saw like this OpenClaw exploding on X and we all thought like, OK, that's like a hype. Basically, it's going to fade away. I remember the first days of this. But in my mind, it was first of all, I ride every hype that I see. So two years ago when MCP came out and it was also like blowing up on X, I also did an MCP, not just today. So I like to ride like every possible hype because I know it's like much easier to push it. But when OpenClaw came out, I saw something that was insane for me because I know how hard with social media scheduler, for anybody who is building a social media scheduler, even you, is to have like a low churn. It's super hard because the problem with all these marketing products is that people are starting to use them. And then if they get discouraged, then they stop the subscription. There is not like something very hard to hold them, not to go away, basically. And when people are scheduling manually, it becomes much harder to keep them inside. They get discouraged really fast and then you try to do things like, oh, it's tricks. I'll send emails. I'll do like a lot of like different stuff. But it feels like really hard to keep up with that. Like you can maybe improve it, but like a half a percent. But when I was like on like 28, I think, percent churn, it's impossible. You can't just reduce it by like just by doing that. And I already started this a year ago when NA10 was like super popular. And I knew that if people automate the entire workflow with NA10, Make or Zapier, they're going to have like so many posts for around the week that is going to reduce the churn a lot. And it did. And this is how, like my first push. And this is the same thing with everybody that's building for agents. Because the agent does so much work, you already reduce so much of your churn. Basically. And this was like insane for me. Because not a lot of people you can see passing some amount of threshold in MRR. Because they are being stuck with churn. I think this is like the biggest thing for everybody. When you say agents do so much of the work, what do you mean by that? So I can, for example, take an agent. That can be Claude Cod, OpenClaw, Hermes Agent. There's so many. And I can like feed it some content. So, okay, I have an article there or something like that. And I can just say, okay, you see this article? Take this and start to create for me posts for the entire week or the entire month, whatever. Once they do that, that's it. Like they're going maybe to stay for the entire month because they have posts for the entire month. Compared to the other way of they are like, you know, just scheduling themselves manually. Takes a lot of hard work. The entry barrier to get the value out of Post-its. Can you give, for people who are watching who have never heard of Post-its before, what is Post-its in like one line? All right. So Post-its is an open source social media scheduling tool. You can schedule posts for social media for 30 plus platforms, basically. It's the biggest one in terms of the amount of platforms that it offers. That was like my main thing when I just started. And there is like, it's a market that existed in for like 20 years with big players like Hootsuite and Buffer really doing an amazing job. And I just like started there. And my only moat was that I was open source. So this is how I managed to get in with all the other competitors. It was nice, but not enough. That's nuts though. That's nuts though. Open source. You think, okay, if it's open source, that means you're providing the code to anybody. Anybody can set up Post-its on their own. Why are they paying you to do it themselves? Do you have the answer to that? Yeah, of course. So there's a few things. First of all, I mean, you also need to understand like how much would it actually cost you to self-host this? Because you would probably need at least, let's say, a $5 a month Hetzner server, let's say, or like super cheap. And then you will need to approve all the providers. That's like a very hard thing. You know, like some of them like TikTok, Meta and stuff. It's like super hard to approve, basically. So it might not work to people just to go through the hassle. Second is like sometimes like just like a different audience, you know, like developers is like one audience. They do a lot of self-hosting. So if you actually look at Post-its, we have around 6 million Docker dial-ons. So like there's more than the actual paying customers. But some of them are maybe, you know, and they do just like UGC for me. So, you know, they post about Post-its because it's easier when it's self-hosting. So a lot of people write about this. And then some of the other people, they just, you know, automation people. It's like technical people that are not developers. It's more like the audience that can actually want to pay you and not self-host it and go through the entire thing. That is pretty hard. But I was like an open source person for the last, I think, five years. I also worked for a different open source company before. I think this is a hack that not a lot of people know of. A lot of people think like if you go open source, everybody will just copy you. They'll just like self-host whatever you do and compete with you. And the reality is that some do this and after a week or two, they just like get discouraged and stop. Because we all know that like to actually push a product, it requires more than just building it, especially today in the vibe-coded area that everybody can vibe-code something. So marketing is a lot. It's also kind of like a frowned upon. If you just like steal something, nobody will actually take you seriously if they understand this is actually like Posties or something like that. They'll just better go to Posties. And it does help a lot of people that would never pay. So, you know, there's like some audiences that they are like in this state of mind, I will never pay. But they will use the self-host. So it's just like more ways for you to expand to new markets that will never pay you anyway. So all in all, I don't think I could ever do this if I was not open source. This is so interesting to me because it is so counterintuitive. Like you were saying, the fact that open source is open source. You're basically providing the thing out there for free. Anybody can clone it. Obviously, that serves very well. In your case, you have built a $118,000 a month business off the back of it. It's growing very rapidly. Why do you think, you just mentioned a minute ago that there was Buffer, there's Hootsuite. There are a bunch of other social media scheduling tools out there that have been around for decades. Why has posters been able or rather how has posters been able to crack through this red ocean of a market and establish itself as now a dominant player? How has posters been able to do that in your own words? So I think it's definitely there is a few parts here. The first one, be fast. It's always like the first thing. So I was extremely fast. When OpenClaw decided, like went out and stuff and all these other ones, I was one of the first to have a skill, to be in this whole ecosystem, listing myself everywhere. And this is what we said before about the power of already having a product and just waiting for the right moment with different things that are coming along your way. So Buffer, I think just like a week ago, released their API. Very long time. A week ago. When did you release your API? How many months ago? No, I mean, Buffer in general, I think they have a different problem. They didn't have an API almost at all. Because I think my assumption, maybe I'm wrong, is that because they have so many users, because they also have a free tier in their product. And you can see that they have like millions of users, giving all of them like an API. I think the usage will go to the roof. It will be like insane. Because with API, usually people that do automation, they create a lot more. So they didn't even have an API for a very long time. They had, they removed, they put it back. Now they have it again. Of course, with a lot of rate limits for people. That's first. Second, I think because I'm building this for agents, so I'm very much, I did really good like a marketing positioning stuff. So if you actually go to the main Posties website, before it was like a normal, you know, scheduling post with Posties, like a normal scheduler. And I changed the entire website to be like your agentic one. So if you're looking for agentic. Can you show the website for us? Yeah, for sure. To see what it looks like? Yeah, for sure. It's still working. You're growing like crazy. It's not like you had, I think a lot of people watching. I thought myself when Open Claw was like, Open Claw blew up, right? You got on top of that. And then obviously there's been Hermes agent and a bunch of other stuff since. But Open Claw, I think, at least in the ex Twitter space, has started to quieten down. People are realizing the constraints with it, etc, etc. It's not as hyped up as it used to be. But still, bro, you're flying. It's not like you're not tied to Open Claw's virality anymore. You are on your own. I'm very lucky. Obviously, I'm very lucky because you can build for one agent and you basically are building to all the agents. But the positioning was really good. So you see, run your social media on autopilot with AI agents. And the main video here, basically, let's refresh it here, is also like it's starting with Open Claw, with a prompt. So I really changed the entire marketing positioning. I want people to understand they're coming for AI agents. Like I'm dropping the old stuff thing of manual scheduling, basically. Yeah, this is really nice. Yeah, and actually I was afraid to do this a little bit. So I even put here a button, I need normal scheduling. So it's showing you the old video of normal scheduling. Tentering to both. Yeah. But generally speaking, I wanted to do this. And I think it helped me in a few ways. First of all, yes, the marketing positioning was better because I'm focusing only on agents. Like I'm not doing anything else. So people are expecting more agentic stuff and so on. Second thing, I'm open source. Open source is for technical people. People are using agent are more technical. They might lean towards more technical product, basically. So that's like the second thing. Third, I'm supporting so many social media platforms. So they can with their agent 3D schedule it to all of them together. Some of them are like super niche. I have also Web3 platforms and so on that nobody else supports. And then in the end, I think I have, like compared to Buffer, again, I have competitors that they are like younger and they have all these kind of things. But with Buffer, I've seen also that they are missing some, when you schedule stuff to different social media platforms, you don't have all the options like schedule to an X community, for example, or something like that. Like something like specific or making a LinkedIn poll or whatever. Something like more specific per this. It was like normal scheduling. So I think that's what's pushed me a lot. And also the fact that I know a lot of people now in the old vibe coded area, they are like building really fast. They vibe code stuff. But a lot of the things are also like very clunky sometimes that people build. With posties, I always use designers. I know it's like a little bit old fashioned, but I always use designers. I never use, I never vibe code myself. I want to, I want things to be like a little bit high end. I know that right now today, maybe AI can do high end, but let's say vibe coding a year ago or something didn't. So it's like much, much, much better. I think agents, I think AI is terrible at design right now. It's great at backend. It's great at doing the technical, like the planning, the infrastructure kind of stuff. But bro, the design, I think you've done the right thing by hiring designers. It's not only that. Have you tried claw design? Have you tried claw design? I tried, yes. It's so generic. Like it is so, it just looks like a template. If you, anything that is generated with claw design looks the same. And I think in general, this is what, like what you said, like who cares? Okay, look like a template. But what you're trying to do always is like try to do, to look different from everybody else. So, you know, everybody look vibe coded. You will not look vibe coded, even though the functionality can be like very similar. So that works also really well. And, you know, I go to, I refactor, I think my scheduling stuff, just to schedule post maybe three times in the product. Every time I go in, I check the other companies. I make a change. Okay, again with a designer, totally different UX and stuff. Like really making the product like crisp, you know, like good, good product makes people stay basically. Makes sense. Clearly it's working. What I find interesting about that is your moat, by the sounds of it, is basically that you're small. For a lot of people, 120K a month is a lot. And it is. But obviously if we're comparing to Buffer, I don't know what they're doing, two, three million something a month? Two million per month. Which is humongous. Obviously they've been in it way longer than you have. But the moat is you're just one guy. Obviously you have a designer, but you're one guy. Would you say that's right? Is that a fair thing to say? I work with freelancers, but I'm the only one in the company. So the designers, they're like freelancers, obviously. I had SEO people before also as freelancers. Developers also here and there. Like I don't hire developers for the core product, but many times I outsource stuff for developers for, you know, like the marketing website, landing pages and stuff like that. Things that, you know, don't need like crazy programming around it. That makes sense. But you can move fast, right? That's I think the key thing. Definitely was moving fast. Yes. How would you say you're not moving fast anymore? You know, when your usage becoming like really high, you start to encounter like a lot of stuff that you haven't when it was low. So, you know, if you used to send before to, I don't know, Pinterest one post a day, and now you're sending like a thousand posts per day. And then you start to look at stuff. Oh, wait, Pinterest has some rate limiting per the amount of posts that you can schedule. You know, even Pinterest have like a mechanism of finding, like I would never experience that with any other schedule, but with Pinterest, they send me a strike that I'm spamming them. And if I have three strikes, they will ban me. And there's only one that I've seen that do something like this. So you need to find mechanism and stuff. Your usage for the servers like is much higher, basically. So really making the product better now is like the most like top priority, fixing bugs, improving the UX and so on. So it's a little bit slower. And I understand this is, I guess, what happens with every big product. It's starting to become slower. You're solving a different problem by the sounds of it scaling instead of just getting off the ground. Exactly. There's this quote. I shared it with you just before the call or like last week that Y Combinator said, the next trillion users of the internet aren't going to be humans. They're going to be agents. So build things for agents. How much, given the fact you've built posters, your startup for agents, how much do you agree or disagree with that quote? I'm just thinking this is definitely the future. The question is how further ahead are we? I think it's going to take some time. It is good for now to build for agents. Like, absolutely agree. I do a lot of stuff with agents myself, not for posties or with posties together, using multiple services at the same time. Definitely yes. I find it less effective in B2C stuff. So, you know, it's like very, very effective in business stuff, but less in like commercials, like, you know, Netflix or things like that. Things like B2C and things like that. I don't know what. Building for agents? Sorry? Building for agents? Yeah, building for agents. Like building consumer stuff for agent. I haven't seen like a very big use for this myself. It's the thing that I'm doing day to day. What do you say with posties? Is it B2B or B2C? B2B. Because it strikes me as it's B2B. I'm selling to businesses. I'm not selling to consumers. I'm selling to people that have a business or trying to build a business. I mean, I'm like the old style. I know a lot of people look at this as B2C. But if you're looking at, because I'm doing like this entrepreneurship for like seven years with many failures, we always look at B2C as things like Airbnb, Netflix, Spotify, things like that. This is like a consumer for me. Yeah. Okay. That makes sense. I never really thought about it that way, but it does make sense. So if you're trying to make your product agent usable, I think a lot of people listening to this podcast, listening to it and thinking, all right, I want to build my startup so that it can be used by agents so that I can be ready for this next truly in user wave on the internet of agents. How, as someone who's built something for posties, for agents, how do you go about doing that? How do I go about making my product agent usable? Do you have a breakdown? Something you can share on screen? Show us how you did it for posties. Yeah, for sure. It's actually very, very easy. If you already have an API, basically, for your website, for things that you can give people with postmen or something like that to do, you're already ready for agents. You can just take all this content of your documentation, make one MD file describing everything, call it skill MD, put it on a different GitHub, that's it. You're agent ready. You can install it as a skill on any local LLM. You can just do NPX, space, skill, space, add, and the GitHub repository. That's it. You're basically finished. This is like the most, like, not something too complicated. It works. This was the first version of posties before I figured more things out. But it's like the easiest one. And then on the other side, so skills, and it's connected more to like personal assistants, like Cloud Code, OpenClaw, and stuff like that. If you want to make it for like chat GPT or Cloud, the remote ones, not the one on your computer, then you want to build an MCP, basically. Because this is like the way that they work, basically. And you can use different frameworks. I use Mastra. Some people use Langchain. Some people just like use Entropic. That doesn't matter, really. And use it with an MCP. That's it. Can you show this to us? Can you show us what you got for posties? Yeah, for sure. I can actually show you how I give it to people when people are seeing this. And of course, I can definitely make it better at some point. Was posties originally like agent built at all? Or was it just a UI built for humans? No, it was UI. First UI, then API, then automations in general. And then, you know, agents, basically. So if someone has built something, if someone's built their startup for, like they've got product already and they built it for humans, it's okay. They can shift it to agents later. It's not like you have to build something. It's just a marketing message. You can shift it to agent in a minute. You just need to create one MD file. If you have a public API, it takes you a minute. If not, you need to basically create like a public API. And for me, once you have this, basically, you go here to posties to developers and you have here the API key if you want to use the API directly. Here is if you want to use a CLI and skill. And here you need to install a CLI. And I'll give a little bit further explanation about the CLI, but you don't have to use a CLI. You can just install the skill. It's basically a GitHub repository. So if we actually go to the URL here, basically, I have this repository. It's called posties agent. And I have here the skill MD file. And it just like explain how to use posties, basically all the CLI commands or API commands if you have. So it's like super, super simple to make this. And then you have the MCP, which is a URL. Basically, you have a few ways to do it. So I put here for all the users already what they need to, like what is the command. So, you know, if you want to install for Cloud Code cursor and something like that, basically, you can just write this command and it will add the MCP server to Cloud Code. And if you use remote servers like chatGPT or Cloud, there is a different URL that is like this slash MCP slash your API key. And I want to maybe explain a little bit more on these two things, which is this specific MCP and this specific CLI stuff, basically. What's that CI, like the remote servers thing under the CLI versus locally? Yeah, sorry. So this one, I will also explain about this one. So this is if you want to, basically, I have a CLI. So if you want to authenticate your CLI, people that use Stripe CLI and stuff will already understand this. So same as Stripe. You can do posties, auth, login, and it will get you to this login authentication that after you've done that, it's basically your CLI is authenticated and you don't need to copy-paste your API key. So it's very common in many CLI tools, just easier people need to copy and stuff. And this CI remote service, so for example, if you already have a service that you are deploying, let's say you have like an auto-deploy of posties there. Not posties, your own service and stuff. And you can do like the authenticate because maybe it's like a Docker that's going on app or something like that. Then you can just like add an environment variable of the API key that you have and it will also serve as your authentication, basically. Now, why we have here a CLI? It's because it's just much easier for the LLM to work with a CLI than to work with an API. So because if we are actually thinking about this, imagine that you have an API. So every time it's like a very long request that you're going to do. So let's say you want to create a post in posties. So you have like a long URL. You have a big JSON file with all the parameters that you need to send, basically. And it's trying to send it. And what happened is that your context window has so much context inside, basically, because you write like the entire API URL and JSON and stuff. But when you use a CLI, it becomes like much, much, much shorter, basically. So this prevents a lot of the context rot. So that's like the first thing. The second thing is that if you don't have a CLI and you send an HTTP request to your server, then it means that the LLM needs your API key. So it means that it's going to write the API key inside the context window, basically. Most of the time, it's not a big deal for people. But it can, if you have moldbook or something, it can expose it somewhere and you don't want to have something like that. So a CLI for me was super easy. I already had the docs for Posty. So I just said to Claude, go to my docs and create a CLI from this. That's it. It's like super simple. Not a lot of work. So you built the, if you listed out the order in which you built these things, API number one, is that right? API number one. I had it before agents even. Okay. And then from there you built the, I mean, do you even use MCP or CLI? Do people even use MCP much anymore? Or can you just look at me? People actually, let's call it the boomers, they use MCP. Because they have not a lot of options because they are using ChetGPT directly. So let's say they use ChetGPT. They can't, like ChetGPT and Claude.ai and stuff, they can't accept skills. Because skills is like mostly locally, you know, it's like command lines that you're installing on a computer and stuff like that. So they usually work with MCPs still. Even though people said MCPs are dead, they are working only with that. So there is nothing that you can do. But there is something here that is very important. Because you can also work with Cloud Code with MCP. By the way, with the OpenCloud and stuff, usually you can't work with MCP. Because MCP is like a heavier protocol that requires requests that lives in the background for a long time. It's like a get request that gets streams all the time. And then from the other side you send post requests and you get information back. So you must be like, it's kind of like WebSockets. You need to be connected like all the time to the server. It's heavier load on the server. And because all this like OpenCloud and stuff that lives in Telegram and stuff like that, and sometimes they are off, sometimes they are on, they usually don't support MCP. Usually. Some of them do. But the interesting part here, and this is something that not a lot of people know, and this is something that I did, is the authentication for MCP. So if you want to authenticate MCP, usually you have three options. The first option is that you, and something that not every server support that, you can pass an authorization header to the MCP, like in Cloud Code, for example, support that, with your API key. This works with Cloud Code. It's great. ChatGPT, for example, doesn't support this. You can't do this. So with ChatGPT, do you have another option? It's that you can create like your own auth authentication, basically, and edit there. But I found it really complicated for people, and also not really always working. So I did something that, I don't know, it's between secured, not secured, but you can always rotate your key. Is that I did that the MCP URL already contains the API key inside. So basically, you know, it's like no authentication. So you can just like edit to ChatGPT with no authentication, just like that, basically. So I found this working like much easier for people when they want to use it with other servers. But you know, just any person that like watching this, it's so, it's only the positioning that is hard. Creating a skill just requires you to have a public API, a skill or an MCP. That's it. Okay. So API, and then from an API, you can create a CLI and an MCP, and then you're covering all bases. You're covering Cloud Code, ChatGPT, Cloud, boomers, non-boomers. Anything. And the nice things, actually, is that there's even a little bit of boomers, not too many, that use like OpenClone stuff. They can just go to OpenClone and say, go and install me my Post-it skill, basically. And that's it. It will go and install the Post-it skill. It doesn't need to go to your website and stuff. But you do need to have like OpenDocs, at least that it can search for it. So for example, if you go to my actual docs website, I also have here CLI that explain also like publicly how to use the CLI. So, you know, the CLI or the agents and so on. So, you know, if an LLM search for this itself and trying to install it, it can also find it basically. Okay. Do you have the skills listed anywhere like Claw Hub? Is that still a thing? Or do you, where do you need to host skills nowadays? Do you have it in your GitHub repo? Where is it? So, it started with Claw Hub. Claw Hub was the first because it started with OpenClaw. It's like the most buggyest website possible. But you also need to look at this as a marketing platform. Because basically, if you have a lot of downloads, then it recommends you. It puts you higher on top and so on. So for me, I started with Claw Hub. But now it becomes like industry standard to use NPX skills like I have here, basically. Like this one. So, NPX skills, it's basically from Vercel. You just write NPX skills add. And then you add your GitHub repository. In this GitHub repository, there should be a root file called skill MD, basically. And then once you run this, you have like a whole interface, basically, that's saying, which assistant do you want to install it to? And you can install it. It supports almost any possible assistant, basically. So, this is like makes things much, much, much easier compared to Claw Hub that was initially was supposed to be only for OpenClaw. Okay. So, if someone wants to do this with Vercel, there's a Vercel page for this? Is it skills? If they just look up online skills. If they want. So, NPX skills. So, I think it's called agent skills, basically. I think it's this one. Yeah. This one. Okay. So, agent skills. And they also have like documentation on how to use this. But it's basically, yeah, they can just look it up. But it's like, it's become like really kind of like a standard for these things. Okay. So, if you follow those instructions, you'll be able to get yourself set up on NPX skills. And then you can say, hey, it's all the skills to post this. You don't even need to do this. All you need to do, like in today's world, is like if you have like a public API already, go to your repository on public API, please build me a skill. And that's it. And create a skill.md file and just like upload it to GitHub. And that's it. You're finished, basically. Okay. So, if we're recapping everything, what's the order in which you're trying to make your product agent usable? At the moment, it's just for humans. What are we doing and what order? First of all, make sure that you have some kind of like a public API. You don't have to have a public API. It's just like much, much easier than you need to find some other protocols. Because even my CLI basically sent API requests in the end. So, you want to have an API, public API. It's also good in general. I believe most of the products today give API for people already a long time ago. So, you want to have an API. Then you want to build yourself CLI from this. So, you just ask, go to my API, create a whole CLI. Once the CLI is ready, you want to publish it on npm.js, basically. Usually, I mean, some people also use Python. I just find npm much better because it's easier to install, at least, for me. So, you publish it on npm.js. Then you create a new repository on GitHub. And you put there the skill.md file on how to use your CLI, basically. And also tell it to install the CLI. So, if somebody's just installing the skill and they don't have the CLI, the LLM will install it automatically for them because it's inside, basically. Of course, you can say to Codex or ChGPT or CloudCode just to write the skill and everything for you. You can even tell them, if you have the GitHub CLI, publish it to GitHub. You can do everything with a skill. With the CLI, you don't even need to go to any other website for that. This is the easiest, usually. MCP are a little bit more harder to create, basically. But, you know, anybody can Vibe Codex. It's still very easy. So, you also make the MCP. And then we can talk about this also, about the other side of marketing. How can I... Why we need to create this? And what is the marketing opportunity in all this kind of stuff, basically? I wanted to ask you exactly that. I think that's the next step. It's like, okay, you make your product agent usable. But at the end of the day, the agents aren't the ones with the money paying for the startup, paying for Post-Its. It's the humans who are managing the agents who are paying for it. Is that right? So, you already mentioned Claw Hub. Lists your tool everywhere. Claw Hub is one. Maybe that's... You said a super buggy is dying. If we had to go through how to market it, what's the first thing you would do if you were starting all over again selling Post-Its? So, I learned it because somebody did the work for me and make me super trending, which is Oliver Henry. And we can talk about that. I think the way that is easier, the easiest way today to push agents is on specifically on X. And it's not because, you know, building in public or something like that. There is an actual growth hack on X that you can do that you can sell to agents. So, what happened to me, basically, is that after I had the skill and everything, there was a guy that has 200 followers on X. Could you show us the tweet? Can we see it while you're talking about this? Could you share screens and pull up the Oliver Henry tweet? Because I think the way that this is structured is really, really important. Yes, definitely. And he made, like, more articles, not only one. So, I met this guy. I never met this guy before. I'll just, like, give a bit of a context before I show the tweet. I've never met this guy before. He, what I do, and I think many people do this if they have a little bit of brand, I go to different social media platforms sometimes. And I just, like, you know, I write posties in the search. And I search for the latest tweets or latest YouTube videos just to see if somebody wrote something about you. You can also use a social media listing tool, whatever. So, I saw a guy named Oliver Henry writing a few posts about posties, how he automates his marketing. The guy has, like, 200 followers on X. And he wrote that he has an app. It's called Snuggly. And he brings, like, customers automated inside of TikTok. And I was, like, okay, this is actually very interesting that he already uses OpenClo with posties. But he had, like, 200 followers, basically, on X. So, he said he will write an article on X. And I was, like, okay, cool article. It's article. X pushing articles. Maybe if he posted a post, you know, maybe, like, 5,000, 10,000 people will see it with 200 followers. Maybe. I'll get, like, five, 10 subscriptions. Sounds, like, really cool and stuff. I never talk to him because I have a lot of people doing, like, QGCs like that. But that's not what happened. He actually did it when he had 200 followers. He published it on X. And he got 7.2 million views. It's insane. It's insane. And if you actually see how many times posties is written in this article, it's, like, seven times, basically. So, it's, like, all around the article. And everyone is actually an affiliate link to him in each one of them. He has, like, a nice MRR in my sister, in my affiliate. So, this is the spike. This is the spike when you were sharing your screen earlier on the revenue. This is the February-ish spike, right? It's what caused the spike. Suddenly, you go in and you see, like, you have 700 trials. And you're, like, what the fuck? And you were at how much? I never in my life do something like this. Usually, even if I do marketing, and maybe you've seen this also. So, you do, like, tons of marketing. And for you, it's, like, branding, you know? And then you see people coming directly to you. Like, not specifically from one tweet or something like that. Like, over time, like, a lot of word of mouth and something like that. But I was, like, a direct from this marketing. You were at how much recurring revenue at that point before that post was? $21,000. And that's, like, one and a half years of grinding, basically. Incredible. And this is what people talk about, overnight success, right? Overnight success. Yes. You went from $21,000, I know, in three months to over $100,000. But immediately after that post, how much were you at? So, I started to grow by 2 to 3K MRR per day. I was shocked by this. I can't even explain to you how much. Because I felt like these kind of things cannot happen anymore. Like, it's something of the past, you know? Maybe during COVID-19, you know, when SaaS was booming and stuff. I was, like, too many Vibe-coded apps. You need to grind for a long time. It's not possible to have something like this. It's a red ocean. You've got Buffer and Hootsuite and PostBridge. All these other tools out there. So many tools. They all, you know, I talk to Jack sometime. He's crazy in marketing also. So, you know, a lot of people bidding so many. And people, like, you know, flooding the market. Also, you can see, I remember in 2020, you launch your product on Product Hunt. You win first of the day. You get, like, a lot of MRR from that. Today is mostly branding. You don't get that much amount of money from launching on Product Hunt. Things have really shifted. Yeah. So, things really shifted in the way SaaS is being sold. People are more, like, suspecting, not trusting. And actually, we can see it on X. Like, people, you go to somebody X, and they have, like, 10 Vibe-coded tools on their profile. It means that they probably made one through it, next one through it, and so on. Like, so much. Like, we are in the sloppiest time of SaaS, basically. I agree with that. That's why I think a lot of people are saying SaaS is dead. I think it's not SaaS is dead. It's just that the idea that you can just build something, and then people come just because of what you built. I think that is dead. It's the marks, and we're in a new distribution era. And you've nailed that. You said there's a hack. Yeah. You've cracked. What is it? I'm, like, on the edge of my seat. Actually, after he did that article, he made another article. Not 7.3 million views, but 1.5 million views. And I was like, how he does that? Like, I even started, of course, after I DMed him, we're good friends, we're talking now, and so on. But, like, how did you do that? Like, how can it be that you got so many, like, I've never, I don't know. Have you ever seen somebody with, like, 200 followers get 10 million views almost for an article? Like, for a post? It was an anomaly, for sure. I still remember seeing it. I still remember being in my apartment in Chiang Mai and seeing it on my phone and being like, what the is this? Yeah. And they understood something. And this is where all the social media platforms, I think, going today, is that all those social media platforms, instead of becoming follower-based, they're becoming interest-based. So, if before you had, like, a certain amount of followers, and social media will expose you to a certain percentage of your followers, and then if it's really good and people liking and stuff, and the algorithm likes you, it expands you more and more into your network. Instead of that, it's going into this way of there is a certain amount of people that like specific stuff. You post about a specific thing that people like. Like, we will match you together with this content, basically. And this is exactly what happened. He posted about OpenClaw and how to automate, like, the social media and stuff. OpenClaw was, like, at the peak. At the peak, at the peak. So, just like X started to match him with so many people, basically. And I think this is the hack specifically for agents. It's not because it's agents. It's because it's the most trending thing right now on X. So, the way I do it now, now what I do mostly, and it's like my main marketing, and I'll even dig in more into that, is that I scale articles on X. But the way that I do it is that I'm looking for the trends. Because I know that if there is a trend right now, a lot of people is going to be exposed into this trend if I write about this. And I also did it myself. I also had 200 followers. I wrote another article on X and also got half a million views. Okay, not seven million, but it's still a lot. I can also share my article. I had a few like that, actually, but especially in the biggest open cloud time, basically. So, I had something very similar. I had this article here that got 579K views when I had 200 followers. And then I had also this one that I wrote that has 500, like half a million or something views. Crazy. And I understood. How do you scale an X article like Nebo? So, I have a few things that I do. First of all, what is the latest trends? So, it's really good to follow. So, you have the X trending feed on the right side when you enter X. And basically, what it does, it's trying to understand where there's like a lot of posts on a specific topic. So, it shows it on the side based on your followers. So, if you follow like a lot of AI people, for example, you will get something in the AI space. If you follow a lot of building and public people, you can get this. And if you have both, you will get like both of the stuff. So, first of all, I want myself to like and comment on stuff on every AI stuff. So, I match the X algorithm to me. So, it will show me the most trending stuff basically. Then I try to see, okay, what is the biggest trend? For example, Claude 4.8 was released. Okay, great. Let's write an article on Claude 4.8. Okay, Hermes Desktop here now. Let's write about this. So, this is the first thing. And I think if you really want, you can do. Because basically all agents are the same. They are just smarter and just like have a different interface. But you can like make an article already, put it in your stock. And just like wait for the next big release of something. So, for example, we are already prepared for next week with people rumoring that Claude Mythos will be released. So, we'll just like change the article on top to be Claude Mythos basically. Because we understand that this is something that's going to happen. And the thing about this is that it's not specifically agents. But it's just like agents are the most trending stuff on X right now. There is nothing more trending than this. This is why this is an opportunity. Maybe tomorrow there is something that is not agents. Totally different. Some new phrasing word like it was vibe coding on staff. And this will be the most popular one. But right now this agent is like the biggest opportunity. So, this is the first thing. The second thing, and this is for people that have a little bit more money. Of course, everything is like super legit. I book a lot of creators to repost my post. Of course, all the time with a paid promotion tag. So, what we are doing is trying to find people in the same niche basically of AI. So, this is why it's so fun also. Because it's like the most trending topic. It's the easiest people to find basically. Okay. If I had, I don't know, something super niche about, I don't know, SEO maybe today. That is, it's popular but not super popular. People don't talk about this all the time. And I find only creator for SEO would be much, much harder. The way I do it is that I go to other super trending article or videos that I see from other people. I just click on the view engagement of this post basically. And I go to quote post, who quoting this post basically. And then I just like see all the people right there. And I start to send them a message if they are up for like paid promotion basically. So, I have like my own system and I manage this there because I have like a lot of creators. And every time, and again, it's not cheap. So, I pay between like $100 to $400 per repost basically from a person. You want to have a little bit more money but you can also do it if you exactly right now, write about specific topic that is super popular. You can get like a lot of views anyway without all this. We just like make sure it always happens. Like there is no such thing as an article would not get at least 200,000 views or something like that. And then we have these creators and they have like a vibe coded system. I can't show you because I don't want to reveal the creators even though you can go to my post and just like see who quote posted it with a paid promotion tag. But it's like a system that basically after I put the article in, it sends every five minutes to the next creator that they need to repost the post basically with a paid promotion tag. Also, what I suggest people to do is follow Nikita, beer, beer, I guess? Bill. And see what he writes on X. Everything that he says is true. He said videos are going to be very strong on X. It's becoming super strong on X. Building in public doesn't work. Building in public doesn't work. Everything I've seen so far from what he says, even though many people don't agree with what he says, works. So we understood already that videos are going very strong on X. Because, you know, in the end, every social media platform want to keep people on the platform as long as possible. And videos is one of the best thing for that. So we also, we had before, we had two things. Before it was really popular, somebody even posted an article about that, that you take a meme of like 30 seconds, basically. And when you do the repost, you put this meme with some text, basically. And it gets like a lot more views and attention. But then Nikita also uploaded on X. He said if the content is like not original and being used a lot, it will get less views. And it's really happened. Suddenly we've seen that we're not getting everything he says. It's really true. So what people are doing, actually, is they're going to YouTube right now. They clip, they cut like very nice video about something in the same topic. And then when you repost, they put like this 30-second clip with a quote repost of the article. I saw that. I think on one of your articles was Kevin O'Leary, right? A clip from one of this podcast that he was on. And it was him talking about how, is this on your article? Or is this someone else's article? No, I don't think it's mine. Okay, well, there's a massive article that went viral that was about clipping. And I think it was on some clipping tool. Anyway, it was another example because what you're explaining works. And it's all over my feed. You're absolutely right. I see Nikita Beer, the head of products of X, talk about something. And then all of a sudden it's all over my feed. And it's just people constantly quote tweeting these articles with videos and with clips from different videos. It's insane. But I want to show you even more. I'm surprised more people aren't doing it. There is one creator that I'm working with a lot. I don't pay him. We do collaboration because we are pushing both of our tools together. His name is Vadim. He has some amount of followers. So he's not like emptied basically. It was this guy. This is the guy that I was talking about in the article. Clipping agent. Yeah, because it was a Kevin O'Leary clipping thing. Yeah. So this guy, really like him. We're very good friends. We're doing a collaboration on articles together. I'm not paying him because we have like one article. We have both of our tools inside. But what he does is that he quotes his article like a million times, basically. You see? Quote and put a video. All right? So you can see a quote and a video and it's like 3.5K views. But then he quotes again with a video and this one gets 142K views, basically. And it's all a clip from YouTube, basically. And then, again, this like with Tannot. Okay, this is not this one. But again, with this woman and it's 35K views. And he's like, you know, pumping himself the article. He doesn't even need creators. He can do it himself without any creators for the things that he does. So for people who don't want to pay hundreds of dollars to creators, there's no excuse. You can just do what Vadim's doing. How many views does the article have? Out of curiosity, the article... The last one didn't fly because the problem is that we did clothe 4.8, basically. And it's not as popular anymore. That's my suspicion. So with his reposts and stuff and what X is calculating, because we can also talk about what X is calculating and whatnot. It got only 381K views. So it's not that much compared to the previous ones that we had that got 1.2 million views and 2.5 million views. But still very good, regardless. That's still very good, basically. Exactly. But I can say that this is the biggest growth hack right now on X. People that do this, it's actually... They need to think really good about how to do this. Because they... So there is in the latest X algorithm that they released on X. Maybe it's changed since then. It says not to have repeated content, basically. Because if you have a repeated content, X is going to demonetize you. Then you will get less views and stuff. But you can still repost it many times with the different angles. I think X, as you can see, it's not punishing you from this article. Yeah, it's original content per se, because every quote to it is unique. Fascinating. I see this all over my feed, so it's very interesting. I've got a question for you, Nevo. So if you were starting tomorrow with what you know now, what are three agent-ready SaaS ideas that you think nobody's built yet? So there's so many. And you can be in a niche market, or you can do a mass market, or you can start niche and go mass market. I specifically am in a mass market. But I think there's so many in the niche market that you can just do that I really needed, for example, for myself. So you can start with one of the main things that maybe you've also experienced at, is that when you have a SaaS and you're growing it at a certain point, you're starting to get a lot of emails from SEO outreach from people. They want to exchange backlinks and do different stuff. All the time. All the time. You get it all the time, people. Yeah, yeah, yeah. The problem is that you know that it's good many times, but you just don't have time to deal with this or energy to do this and stuff. So you just ignore the emails or stuff like that. And they're all AI, so I'm like, ah, it seems lazy. So I'm just like, delete. I see the end dashes, I'm like, delete. Yeah, but they might be good. They might have a good website to exchange link with. So it would be nice to have, like, I wanted to build this, but I was like, okay, focus on the one SaaS. It would be nice to have, like, an agent that really go talk to these people, check all the website with Ahrefs, SEMrush and stuff, see how good is the DIR of this website. Are they toxic or not? Find the blog post that you want the backlinks from. Ask this person, basically, where they want their backlinks from. If you have a third website, like an ABC exchange, so you send a backlink and they send a backlink to a different website or something like that. It's even better. Where they want it from, the agent do everything, put it on the website, and put a monitoring also that, you know, they don't remove the backlinks or something like that. Yeah. So this is like, I think it is like the kind of things that you start to see when you need business stuff that you can't find. So that's... Like a backlink assistant, your own personal backlink agent. And it's a crazy marketing machine. You don't understand how much, because really, I used to exchange some stuff with other people. There is really good backlinks that you can find, basically. So this is the first thing. Second that I had is maybe to have... And it's not a $1 billion SaaS, okay? It easily can get to like 10K. Is to have a more outcome system, not like a tool like... So post is a tool to schedule posts. But you can have different strategies on how to create content, for example. So one of the really nice things I've done lately is that I used... I wanted to do something like this. Like I will focus all my energy on X, X article, X creators and stuff. But I want, because they write like good articles and they invest time and so on, maybe I can repurpose this to like many other stuff. So I built like a whole skill that go to agent, that generate avatar. There's like an avatar to talk about this. And then there's like a split screen of the avatar and the top. And then on the top, it can take different screenshot or it can take some stuff from the actual article. And build basically a very short form, nice video from this. That can be scheduled to any other social media. You can use post these after that to schedule everywhere. Something like that. So this is something I'm also using. I think it's like super, super good. Not only that, I have this and I use it on my website. This is easier to make. Also, you can do it with an agent that takes the X article, go through an SEO research. So you have three types of SEO research that everybody uses, which is Ahrefs, SEMrush, but they are expensive. And data for SEO, it's the most popular one. So we do like a research on that and then go to my WordPress and create a post there, like telling the story of that specific person. So this also works nicely. Because I feel like SEO become very sloppy lately. So people generate blog posts that basically could be found if you just do a chat GPT search. But if you actually look at all the snitches, like maybe it's a very good X article that somebody wrote. You can repurpose this, tell it in third person about this person and write like all the details inside and basically create a good SEO article with real value, not like just like slop inside. So this works good. And also I did the same thing with YouTube videos. That it goes to a YouTube video, transcribe the entire YouTube video, cut it to clip it for different things and then also write an article. And the nice thing with YouTube videos that I've done is that you can also make the agent to be like, take a video that is not about posties. So maybe it's just like about like social media posting like right schedule, but from some really good creator. And then do the SEO research and make it a posties article in the end. We post this stuff inside with the agent. I need this for the My First Dollar podcast to clip it and turn it into an article. Because it does so well on X. It's just, yeah. Yeah. Maybe I start building these out myself. If anybody listening doesn't have a go, I might steal them myself. Because these ideas are really, really good. Is there another one you said that you had a third one or is that the third one? I think this is like the two main ones that I created for myself basically. But there are so many business operations today that people are doing to move stuff from here to here and so on. Creating an agent is really easy around this part. But what people should do, and I think this is like the top thing, go to a website even. Like TrustMRR basically. See like what is the top website there. And see if the top one that's making the most amount of money, if they have, there's like a lot. If there is one too, it's not a problem. But if it's like a lot of alternative that offering only for agents and just build it for agents basically. Do go through the playbook that you just broke down earlier. API, CLI, blah, blah, blah. Yeah. That's cool. That's a nice idea. Take something that's already proven, that's already working and just make it agent ready. I, in general, support that. In general, because I've been working with many startups, both in the, you know, VCs and not VC startups and stuff like that. And people all the time think like if something doesn't exist, you should not create it basically. Like you want to create only things that don't exist basically. Totally agree. And it's not a wrong thing because like most of the most expensive, like the richest startup in the world, they created something that didn't exist. But you nailing that, finding this exact, exactly the idea that doesn't exist is obviously super hard. Like most of the people won't make it. And second, if it doesn't exist, then it means nobody's searching for it. It means that you need to educate people around this thing. So this becomes like much, much more. You need to create a demand and it becomes like much harder. So, yeah. Proven idea always works. Don't reinvent the wheel. I completely agree. Never David, do you have anything else to share? Is there any final words you want to share with people listening? Anything you missed out or didn't cover? I mean, in the end, I feel like I'm very lucky, especially with Oliver Henry article that he created for me. I think people should really, I know it's like going against the notion. And I totally understand that because if you look two years ago, I would say the exact same thing. Build many product until something sticks. I don't think this is the same situation today because the market changed. I think it was true before. Now it's not because people are creating too many SaaS. Nobody actually investing time on the right one. I think people should not create a totally different startup that never existed before. They should find something and put their own spin on it. And then keep grinding on it until they see an opportunity. It's what I did. You find an opportunity and then you go all in on that. I don't think people should create so many startups. It's rarely ever work. The people that are, you can see a few people that it's working for them, but it's not the most of the population. Most of them like don't make it, basically. And many started before the AI slope area. So it was more valid back then. Yeah. Thank you so much, brother. This is gold. You shared so much good stuff. I appreciate you coming on and sharing this stuff from how to build startups for agents to getting ideas to marketing it. Dude, I'm expecting a lot more articles. If people actually listen to this podcast and take away from it what you actually told, I'm expecting to see a lot more articles on my feed and that's what I want to see. Yeah, definitely. Maybe even you can find some Super X feature to help people with all these articles. Well, you've inspired me for sure. Super X CLI is coming. All right. Sounds good. Thank you, brother. I appreciate you taking the time. Thank you for having me. Thank you.