← Startup Stories - Mixergy
Loading summary
Transcript26 lines
- [00:00]
A
My revenue just exploded once I started sending to agents. Andrew, you've got to see this. Postease is huge.
- [00:07]
B
With OpenClaw presented by Zapier, the AI automation company Nevo, I see the revenue was low and then it took off. Before you tell me why it took off, just walk me through where the original idea was.
- [00:20]
A
So we've started as a social media scheduling tool like everybody else. There's so many people doing it in the business and I always see so many new people on X and I'm like hahaha, it's going to be very hard for you because it's a very flooded market. There's like so many people that do it. And I understood like I have to find like my own blue ocean inside like of this Red Ocean, 20 years old market, so many people inside. So started really as a social media scheduling tool in open source. So that was already my blue ocean. So every time I posted about open source for posti is the first social media scheduling tool that actually like fully open source got tons of views. So that, that was really nice. So you can see here on my metric, like it started like I started here in September, basically $350 growing a little bit, it started to grow and so on. And then you see it like plateaued here a few months.
- [01:19]
B
I didn't know what marketing this as a tool for posting on platforms like X and LinkedIn that human beings can go and use. But because it was open source, anyone can see the code and use it if they don't want to work directly with you and pay you.
- [01:33]
A
I also started to write some hypes. Like what kind of hypes? Every time some hype coming on I was like I'm the first one to put it might be a hype, might be not people are going to use it. They're this is like a very easy way to attract people in. So the first one was the MCP hype. So everybody talked about mcp. It was really cool. You can connect it to your ChatGPT or you can connect it to your cloud and so on and you can just like talk to it and schedule post for you. I mean it's so easy to build it. So I already like introduced MCP back there. I see.
- [02:09]
B
You know what? And I remember getting excited about mcp. The idea was that I would go into Zapier and I would connect it to a bunch of apps that I like, Google Calendar, post it, et cetera. Then I get one code and I give it to ChatGPT. I give it to Claude, I give it to other tools and suddenly all those tools have access to all the tools that I connected a hundred percent.
- [02:27]
A
Many people were excited back then.
- [02:29]
B
So you see that we're all getting excited about it. You add it, people get excited about using post its and the numbers start to increase.
- [02:36]
A
Yeah, we see a little bit of an increase still. This is like when I started to understand this and then I started to also use automation platforms because I understood that a lot of people like to automate their social, their scheduling because they can just create many cool stuff like maybe they generate a video of each VO3, put subtitles on it and then schedule it into Zapier. And that was why I really wanted to go into this area. Mostly because I knew it will decrease my churn. Because when people need to post manually, if they get discouraged, if they don't post, they'll cancel the subscription. But when people are posting with automation, that's a whole different story. Some people, you know, they run their automation and every 15 minutes they push a post to post it. Like they really spam the system Even
- [03:36]
B
the insight that I've gotten from you is that when you allow users to connect their automations into your software, they're going to use it more often because automations naturally are consistent where human beings are undependable that way.
- [03:51]
A
And then we got here, you can see we got into 17k mrr. I think it's if you don't see it in the graph, we even stretch it a little bit, 20k and so on. And then boom, Open Claw came out. I can't say I was the first one to use OpenCloud but I understood there is something big here. Might be a hype, a trend, but I think, but I understood really fast that it's not because it's not about openclaw itself, it's about the way that people are going to behave in the future. So Today you accept OpenClaw, you have also Claude Cowork, right? And you have Perplexity Computer and so on. Like we see there is like a shift. People are going to move to these like chat apps and they're just gonna write what they want and they're gonna get it and it's going to use all their tools together basically to do it. So I understood that and I said okay, what can I do? So what can I do to be like on top of that? So of course it's like to create my own page on the website for Open Cloak. But I think that's I Thought like that's not enough. Let's create a whole CLI for it. So basically if we are looking here, I have a full CLI on how you can use posti. So you install it globally and you have the CLI and you could create different posts. And we also have a skill on clawhub that you can install it after that. And the goal here is basically to reduce allot the amount of tokens that you send to your openclaw LLM. So instead of writing like a full API request with all this JSON and so on, you have like a very short Command. And then OpenClaw can also iterate if there is a problem or it needs to get different stuff and so on. And it really blew up. Somebody I've never met in my life wrote an article about his openclaw Larry and he just like said how he get like a lot of TikTok views. He did something, he is very good in storytelling. So he created his OpenCloud and he'd call it Larry. And I think that's like the main thing why the article was also like super popular. Because it's not Larry, it's not a computer, it's not Open Claw, it's a person. Yes, his name is Larry and they also co wrote this article. So it's like Larry. I saw that some experience that you feel when you read the article, which is really, really cool as a summary
- [06:26]
B
for people who hadn't read it. Essentially what he did was he got Larry, his Open Claw agent to post on TikTok on a regular basis these images with text on them. And he needed a tool to post onto TikTok. He couldn't make a direct connection between Openclaw Larry and TikTok so he used post its. And you're saying and this post by the way took off and other people then discovered it. But you're saying the reason he used you versus other tools is what. What is it that the CLI enabled? Why do people like him pick you?
- [06:57]
A
He just used it. He might as well maybe use some other tool, I don't know. But post it just worked for him. And then he wrote this article. And the nice thing about this article is not like how to use postits because this article will not like post. It is a small thing inside the article. The main core value of the article is like how to get a lot of views on TikTok. This is exactly what he did. He just used Postis. We know that in TikTok. In TikTok right now if you post a slideshow, there is a higher chance the algorithm will like you and will give you a lot of views. He already saw that and this is what is the article all about. And then he just created something really cool to generate all the slides and everything. And he needed it just away quickly without doing a lot of stuff to just schedule it to TikTok. So he just hook up Postis really fast and he could like schedule everything
- [07:46]
B
in the end, I want to understand what it is about, about post its that allows you to court agents and be more useful to them.
- [07:55]
A
So I can't say that I did it like Claude did it for me. Good. So we, because I'm using the command line, I can describe a lot better how to use the command line because you know API requests like so long. But command line can say use this command line and if it doesn't work, use this command line. If it doesn't work, use this command line. You can like iterate with this. And we created like a very cool skill in openclaw. It's also on the postis agent, which is this one. And we really have like, you know, this is like the readme but let's open the skill. So it's really like a description of all the things that you can do with the skill really fast. Like you want to see all your integration, right? Post is integration list. You want to upload a picture, do post these upload. Like there is like really short things to do different stuff basically. And then in the end it's telling you like for example, Important missing release ID handling if analytics false or 10 missing true. Instead of getting the analytics like in order to get the analytics array, you need to write this kind of command and then reconnect your post. And you see it's like simple, simple lines. You don't like need long API request
- [09:13]
B
or let me just make sure that I'm zooming out and understanding this clearly. What you're saying is agents really need good documentation. You didn't write it yourself, you didn't leave it to them to figure it out for themselves. You used Claude to say, you know my software document what other agents are going to need to know in order to use it. And then you packaged it up as a skill. Am I understanding that right?
- [09:38]
A
Yes, but it's not only that. So I could create two types of skills. I could just like say to Claude, listen, you see that my docs just create a skill around it. And actually docs are already kind of a skill because docs are MD file already. Apparently so. But I wanted to do something better because if you actually like let's, let's take for example the actual docs and in the actual docs, let's say I'm going here to the public API in order to create a post, you need to send a post request to this URL here. Like tons of different things you can send and so on. That's a lot of content. So I could definitely just take it like that. But we simplified it a lot. I just told Claude, because let's face the truth, who is writing code these days? I just told Claude, take all my public API and create and turn it into a cli. So this is the first thing he did. And then I pushed it to NPM js. So if you can see, like, if you see in the top of the skill here, you first need to install the post this command line before you, even before you even start to write, like to do whatever is inside the skill. Okay. And then because now everything is so short because imagine this post, this integration list, gives you all the list of integrations that you have if you don't do this, if you do it with the API, you need to do list integrations. So you need like all this long API request and I need to give you all the parameters that you need to pass and then you need to send it in your request. So everything becomes like so much shorter when you turn it into a clique. And then it's like you have a so short skill file compared to like a very long one with so many API requests inside a lot less context rot, a lot less tokens when you use it. Easy to iterate. Like it's so easy for Claude to just like write again this command, get the list of providers and so on. Then to write again all this long API request with the curl in order to get the information basically. Okay, so this is I think the future. The future. How much. It's weird, the futures of all this open claw and agents and stuff is the CLI people that are going to build startups with a CLI are going to win big. Actually, my friend and I working now on a new startup, he's working on it. I'm the marketing guy, but I'm telling him what to do. It's called agent media. It's if you know like, like a UGC content creator so you can like create UGC content, like this woman talking and so on. But the nice thing about this is just it's a CLI tool, same as postings you install the CLI tool, you even do a login from the CLI and then you generate video, Edu media, generate video minus B and so on. Like you, you pass the picture and so on. Again, no API. Because we understand this is like where it's going with all this. Like agents and so on. Yeah.
- [12:46]
B
All right, let me ask you a dumb question here. CLI means going to like something like terminal and typing in what you're looking for and getting it out. How does that enable agents to interact with your software versus an API? Why is that more effective?
- [13:02]
A
Basically, the CLI is a proxy to API. So when you write, when the actual agent write post integration list in the background, it's sending an API request to HTTP API.postis.com integration list, filters and so on. It's just a proxy. It just makes you to write a lot less content when you do different stuff when the agent's trying to talk
- [13:32]
B
to you because you don't have to send all the first information to get to the next bit to add the next bit of it.
- [13:39]
A
Yeah, and I think it becoming even bigger when let's say for example, you want to schedule a post for posties. Okay. Like you want a very long post, a lot of social media and so on. So let's say I'm coming here and you have like create post. This can be a very long JSON that you need to send if you want to schedule your post. Let's hear an example. Um, not general, maybe let's say, okay, Instagram. So you start to send the integration id, the content array of images. There can be a lot of images settings that you want to send posts, Ethereal, the collaboration, so on with the cli. All this become like a fe, like a few words because you don't need all this like lot like the date become minus D and the post array is just like minus P. And you put like question marks and you put the content inside and then another one and another one like quotation. Sorry, no question. Like quotation, put the content, another quotation put the content. It saved like a lot of these things that you need to pass. And I think when the agent does that and it doesn't need to like build the whole JSON, it actually has, it sends better content because it doesn't have this content trot. It doesn't need to reproduce the entire JSON with all these like little parentheses and quotation marks and so on. So it becomes like much more efficient and it works much, much, much better. So I think everybody will Going to move into the CLI in the end every startup. And by the way, why I go into the CLI actually where I even thought about this. I'm like a genius that thought about this. I really like this project. I think it's amazing. From Vercel, it's called Agent Browser. It's basically running a headless browser for you that you can go to different pages and so on. And the way they did it is the same thing with a cli. So you do agent browser open and go here, Agent browser click and so on instead. Just like running you a playwright in the background and do different stuff for you. And I understood, okay, they know what they are doing. Maybe I should also copy that.
- [16:07]
B
I see. If this is the way that they're going to enable agents and others to connect to web pages that are really complicated sets of back and forth information, I think I could do the same thing.
- [16:18]
A
Do I have that right? Exactly. Because you know, this is like really saving so much context when you just use it instead of writing the command for playwright every time.
- [16:27]
B
How hard was it to do all this? Is it really as easy as just telling Claude, get it done for me. Take this API and turn it into cli.
- [16:35]
A
Yeah, like Claude is really, really good. Like all the llew ones. I mean I'm. I just like the only problem I have with clothes sometimes is that I forget to pass the dangerously skip my commands because I, you know, just like tell it to do something. I go, I go back. Shit, it didn't do anything. It was stuck on this. Like I need to approve permission. That's the only thing that a hundred percent we've reached now. $45,000 per month. I assume next week I'll get to 50k per month. I think it's a very life changing money. Especially when you're not like a funded companies. You just take all your money, like most of the money for yourself. You can spend it on whatever you want, marketing and so on, but you take most of the money.
- [17:25]
B
All right, Eva, thanks so much for doing this.