Loading summary
Transcript79 lines
- [00:01]
Michael Stelzner
Are you feeling overwhelmed trying to keep up with all the AI changes? Trust me, I know how it feels. The marketers who are truly thriving today with AI have discovered something really important. They're not doing it alone. At the AI Business Society, we've created a community where smart, curious marketers like you share discoveries, troubleshoot their challenges, and celebrate breakthroughs together. Listen to what Marissa Shadwick had to say. Quote, I found my people. I love chatting about our experiences with AI and supporting each other on our journey towards the future. Unquote. Stop navigating the AI revolution by yourself. Join our community of innovators and let's grow together. Visit socialmediaexaminer.com AI to learn more.
- [00:56]
Sabrina Romanov
Welcome to the AI Explored podcast, helping.
- [01:00]
Michael Stelzner
You put AI to work. And now, here's your host, Michael Stelzner. Hello, hello, hello. Thank you so much for joining me for the AI Explored podcast brought to you by Social Media Examiner. I'm your host, Michael Stelzner, and this is the podcast for marketers, creating creators and business owners who want to know how to put AI to work. Today we've got a really exciting podcast episode. I'm going to be joined by someone who, over the last year has grown her following on social media from Nothing to over 700,000 followers across the social platforms. What she did with AI helped her to launch her very own software company for the second time. And, and it's really exciting because we're going to explore how to use AI inspired social content to grow a really, really loyal audience with Sabrina Romanov. If you're new to this podcast, be sure to follow this show so you don't miss any of our future content. Let's transition over to this week's interview with Sabrina Romanov, helping you simplify your AI journey. Here is this week's expert guide. Today, I'm very excited to be joined by Sabrina Romanov. If you don't know who Sabrina is, she's an AI educator who helps people understand how to apply AI to their life and work. She's also the founder of Blotato, an AI driven social content tool and her podcast can be found by searching. Sabrina Romanov. Sabrina, welcome to the show. How you doing today?
- [02:35]
Sabrina Romanov
Thanks so much for having me. Yeah, doing well.
- [02:37]
Michael Stelzner
I'm super excited that you're here today. Sabrina and I are going to explore how to use AI to inspire and create your social content. Now, before we get into that, I would love to hear your story. How did you get into AI? Start wherever you want to start.
- [02:52]
Sabrina Romanov
Yeah, absolutely. So it Started back in college, actually. I studied computer science and physics at UC Berkeley and took my first AI machine learning course when I was, I think a junior or senior in college. And it blew my mind even then, what you could do with AI. Keep in mind, this was 2012, 2013, and I thought I was late to AI then because I'm being taught by professors who have been doing it for a decade. Plus one of my investments, investors in my first company had been doing AI since the 80s.
- [03:19]
Michael Stelzner
Wow.
- [03:19]
Sabrina Romanov
So you can imagine. Yeah, in 2013 I already felt I was late to the game, but that's how I got into it, you know, And a big part of it was around language. Like, I've always been an avid reader. And so the idea of AI that could like understand what you're saying and reply with something meaningful, thoughtful back, something engaging like that was mind blowing to me. And so my first company specialized in that kind of technology. We did real time speech recognition, natural language processing for sales and customer support, custom conversations. And that's a whole story in and of itself. Happy to dive into it.
- [03:51]
Michael Stelzner
Well, yeah, share a little bit more like, what did that mean back then? Quote, unquote, whenever that is, you know.
- [03:56]
Sabrina Romanov
Yeah, yeah. So keep in mind this was like 2013, 2014. So we had technology that could join a conference call and understand what people were saying in real time and then use what was called natural language processing to detect questions that were asked, objections that were raised in the context of a sales call, just different frustrations customers were having in the context of a support call, and then suggest to the agent what they should say next or do next to help resolve your issue or overcome your objection. So this was like a lot in 2014.
- [04:26]
Michael Stelzner
That's like what we call AI note taking today, really, right?
- [04:28]
Sabrina Romanov
It is, yeah. We even launched a variation specifically for note taking. It was like top two product on product content today. But it was just so early to market. You know, people were not accustomed to the idea of AI joining your call, listening to everything that's being said. That was not an idea that was welcomed in that time frame.
- [04:47]
Michael Stelzner
Very cool. So what happened with that business?
- [04:48]
Sabrina Romanov
Yeah, so we actually ended up focusing on conversations where they already had a culture of recording calls of sales and customer support, for example. And so in a support context, you might call into your bank or health insurance with a set of questions. Our AI helps the agent understand, like, what is it you're asking about? Here's the relevant documentation or procedure to help you resolve your issue. And then we ended up selling that company to a pretty big player in the BPO space called pegasystems. They do over a billion in revenue and have a really big customer service unit.
- [05:19]
Michael Stelzner
Very cool. What came next after you exited?
- [05:22]
Sabrina Romanov
Yeah, that's a good question. I was honestly a little lost for a while. I was definitely burnt out from my first startup. So this was my first startup. My husband and I, we were fresh out of college, did not know we were doing. Made a ton and ton of mistakes. Like, from a burnout point of view, I was physically, mentally, psychologically just super burnt out. It spent probably at least a year just recovering from that and realizing, oh, I should probably do something now. I kind of fell into AI education because I just noticed this mass gap between where we currently live today, which is the Salt Lake City area, and like Silicon Valley, which is like pretty far ahead in terms of AI adoption agents and how companies are deploying it. There's a pretty massive gap between that and just like everyday people and their relationship with AI and how they use AI or how they think they can use AI. And so I started making AI education content about a year ago just to help teach everyday people, like, here's how to use ChatGPT to be a little more productive or to help research things that normally you might have to pay somebody for help and expertise with. And so, yeah, it's been a wild ride. Started creating AI education content a year ago, have grown from zero to I think 750,000 followers now in just a year. AI has been a huge part of that, like helping me scale myself and to be able to educate so many folks.
- [06:37]
Michael Stelzner
What platforms were you using to educate? Was it YouTube? Was it TikTok? Was it all of the above?
- [06:42]
Sabrina Romanov
Yeah. So I actually just started with one post a day on LinkedIn. That was my very first one for about two months. And then I threw in TikTok, started experimenting there. It still took a little bit of time to figure out, like, my voice, my format. I was very uncomfortable on video. Most people don't realize that now, but I could not watch my early videos. It was just too cringe for me. Then I added on Instagram, I'm on YouTube as well, my newsletter, I just moved to substack. I have a podcast as well. So I'm all over the place. Twitter, threads, Facebook as well.
- [07:13]
Michael Stelzner
So about a year ago, so we're recording this in 2025. So it would have been 2024 that you started publishing content on LinkedIn. Did you strategy? Were you just sharing educational tips to just try to help people understand how to leverage things like ChatGPT. And then all of a sudden something happened, like share that journey and then ultimately into blotato. I would love to kind of understand that.
- [07:35]
Sabrina Romanov
Yeah, I was all over the place in terms of like content pillars. I was experimenting a lot. A lot of my earlier content then was, I would say, more technical. And that was kind of the disconnect. Like I was publishing stuff, building stuff with AI agents and agent simulations back then. Those YouTube videos have 200 views. When started doing TikTok and started making beginner level content, that's when things really took off. And so what I learned in that process is like, oh wow. Like when it comes to just like everyday people, the level of education when it comes to AI and stuff, I mean, agents was not even a hot term back then in tech, let alone for everyday folks. And so I realized I kind of had to start at the beginning. What is ChatGPT? Here's how to install it, here's how to like write a prompt, or here's some tips to help you improve your existing prompts. And like when I started batching that level of content with the platform, for example TikTok and Instagram, I realized there's really a massive need for that kind of beginner level stuff. Like a lot of people still don't know stuff that I take for granted cause I use it every single day. Right. But there's so much value in like the beginner level stuff and getting accustomed to using these tools that I think I overestimated when I first started and I started with like pretty technical content. And since then what I've done is I use the social media platforms for pretty beginner level content and then I transition folks through my funnel of sorts. So my new newsletter and YouTube now have more of my technical content.
- [09:00]
Michael Stelzner
Yeah. And tell us about potato.
- [09:01]
Sabrina Romanov
Oh yeah. So it's a tool I've built for myself really to help myself scale without burning out. And so like going from 0 to 750,000 in a year with no paid ads, no budget, no team, basically impossible. Unless you're leveraging like AI and automation in a smart way. For me, I just built out like custom automations for myself. But what happened is people kept asking me like, wait, how are you doing this? Like how are you making content at this level without like a team of five, five people behind you? You know, like researching, ideating, script writing, creating videos, editing the videos, posting, managing comments. And so I started sharing these automations on YouTube. But then I found people you know, didn't want to have to fiddle around with workflow automations and custom builds and things like that. They just wanted a simple SaaS app where they can log in, connect their social accounts, use AI to assist in the research and writing process and then boom, publish their posts. And so I just kept hearing from folks like, I'm overwhelmed with creating content. How are you possibly doing this? And so blotato just came about from solving my own problem and then other people ask me how I solved my problem, so I built it for, for everyone else to use it too.
- [10:10]
Michael Stelzner
Very cool. All right folks, well we're going to talk about how anyone who's listening who has to create content on the socials, whether you be a creator, entrepreneur, entrepreneur, or even a marketer working inside of a small or even mid sized business, could benefit from some of the things that Sabrina has discovered. So first, let me start with the why question. Why should creators and entrepreneurs use AI to help them with their social content? I know that at least on my other show I've got people that have been creators that have been creating custom content on the socials, whether it be written or video, for many, many years and maybe haven't fully embraced what AI could do for them. What's the upside if this is done?
- [10:50]
Sabrina Romanov
Well, yeah, I think the upside is being able to scale your teaching in a way that you don't burn out. At least that's been the upside for me because my goal with all of this is to teach 1 million people AI. And my question is how do I do that at scale but without like burning out or like having to spend a lot of money because then I have to make money through the content. I don't even want to do that. So I don't charge for any of my free AI education. I want to keep it that way. But it's really hard to do that if you don't have like systems in place, automations in place, AI in place to assist you. It doesn't have to automate your entire process. Like even if it assisted you just with research and ideation, creating drafts, like brainstorming different hooks, that's plenty. Or for any of you who's a content creator, you know, it's, it's every single day it's like how do you create the best original content that you can? And it's very hard to do that at scale consistently for many years without burning out or investing a bunch of money in a team.
- [11:47]
Michael Stelzner
Have you found in your personal experience, have you been able to quantify the kind of time savings. How much time do you anticipate it would have taken you to do what you're doing now if you did not have AI versus how much time it's actually taking you now that you do have AI backing you up?
- [12:01]
Sabrina Romanov
Yeah, absolutely. Because when I first started I really didn't use AI for the automation pieces and even for beginner content creators, I don't recommend jumping into automation like use AI, use ChatGPT to have that dialogue back and forth. But I don't recommend people like fiddle around with automations until you figure out what's your voice, what's your audience like, what's your offer like. There are all these insecurities you kind of have to fight your way through when you're starting out as a content creator and you have to figure that stuff out through trial and error and has nothing to do with using AI. So in the beginning, you know, I didn't have a lot of these automations in place. I would create content every single day and it took me a while, I didn't know what to write. You know, I'd have to like post manually to each of these platforms. Every single one is a different app, different interface, and so it take a lot of time. So let's say like maybe 30 hours a week just to make content, distribute it across all the platforms. Now I make content in only one and a half days a week, so about 10 hours a week. And again, that's probably the biggest chunk of that is just my YouTube video. Like all the prep work that goes into building an automation, the templates that are free for people to download and then filming the video and editing the video and then the rest of that time is all of my other content on like eight different platforms.
- [13:14]
Michael Stelzner
What I love about this is you, Sabrina, are saving 20 hours a week and you're not compromising presumably on the quality of the output. That's the other side of it. Right. So with that 20 hours a week, you can go out there and you can do all sorts of other things like develop potato or take on clients if you want to or anything else. And this is the thing that I think a lot of people don't do the math when you think about how if you had 20 hours extra per week or five hours or whatever your number is, what could you do with that? I mean, that could be a really big unlock for a lot of people that wish they had time to pursue something they've always dreamed of doing, but they've never done before. So I Love that. Now what do people need to be thinking about before they start using AI to grow a content following? What are some of the things that you have learned that you might want to share with them before they actually start fully embracing this?
- [14:10]
Sabrina Romanov
Yeah, number one is I do see a lot of beginners kind of treat AI as a silver bullet. Like, oh, I'm going to use AI for this and then I'm going to like magically explode my following and go viral like every single day. And I would just like reset your expectations and also not overthink it. Use AI as honestly like a way 10x better Google search to help you like research ideas, research customer pain points, research hooks. Create like 10 different drafts and see which one like you resonate with. Use it for a lot of that. But like don't overthink it. Don't like over rely on it. Don't think just because you're using AI you're going to go viral or anything like that. There's a lot of you involved in the process. Again, like, what is your voice? Why are you doing this? What is important to you when you communicate your message? Who are you serving and what do they care about? Are you reaching them in an authentic way? Are you being authentic in your content? Like if you're really uncomfortable on video and you don't want to continue, you know, maybe focus more on writing on Substack or your blog or whatever it is. There's all of these like personal things that. But I think people kind of want to avoid shortcut and thinking by using AI. I don't have to like deal with all of that messy personal, philosophical stuff that I have to figure out myself. Tip number two would just be that you are still in control. Like ChatGPT can go down any rabbit hole that you pointed to. Like, but you're the director in charge. You need to like have some kind of viewpoint or perspective that you are trying to share and then use chat GPT to help you like create content around that. So like it could pull up relev research to a perspective that you have if you had a certain experience. Chat, chat GPT or perplexer, you can research other people who've had those experiences or differing experiences and you can talk about the difference. So like at the end of the day like you're very much in control. Don't expect AI to just like write whatever and you go along with it and go viral. It just, it just doesn't work that way. And I just want to make sure people have that like correct expectation when using it.
- [16:07]
Michael Stelzner
I love it. When we were prepping, you mentioned a good place to maybe even consider thinking about is having AI analyze some of your comments. Do you want to talk about that a little bit?
- [16:16]
Sabrina Romanov
Yeah, absolutely. So I think AI for research and prep is a really underserved area. Like, there's a lot of focus on AI for actually writing the content, but just like in data science, it's like garbage in, garbage out. Like, if you don't have any personal experience, you know, backing up the post, you're saying it shows. Like, it shows because you're not talking about specific numbers or like specific challenges you went through. And so I always talk about using AI for the research and prep. Like, what are your customers specific pain points? How have you dealt with those pain points when you served your clients? And then like, use AI to help you maybe structure how that's written so that it's in a format easy to digest. But like always, you know, add your perspective, add your unique insights, add your experiences, your data, your stories, and use AI for that prep work. So like, that is an example of it. Like, you could take your best posts, use AI to analyze the best comments there, the controversial comments, and that can be a seed for new posts that are coming up. You can also do that with other people's videos, for example, if there's a person in your niche that you follow all their content, take one of their videos, have AI help you analyze the comments, look for things that you could talk about that you have personal experiences with. You can literally ask ChatGPT based on everything you know about me, analyze these comments and identify like, the three topics that I am uniquely suited to talk about. So that's an idea for being able to generate those types of content from.
- [17:44]
Michael Stelzner
A practical tactical perspective. Because I like to try to get the practical tactical on this. Sometimes ChatGPT isn't really good, at least historically with you give it a link to some posts and it doesn't really understand what it's reading. At least that's what been my experience. Do you find that these comments need to be like exported into a sheet or is it good enough now on all the platforms? Because I know some of these social platforms are locked down where they. ChatGPT can't actually see some of this stuff. So how would they practically go about analyzing some of their best posts? Is there any prep work they would need to do on that?
- [18:18]
Sabrina Romanov
Yeah, so it depends on the platform. Like if you can get access to your comments, another workaround that's honestly easier is something like Reddit or Quora where it is public so you can have a topic. Let's say AI vibe coding is hotly being debated right now, and you can search Reddit for the top posts of that topic and then that Reddit link, you can have perplexity. For example, you can set the focus specifically on Reddit to search for the top posts in that topic. And so if you can't get access to all of your comments or it's too tedious to just like have a script to copy the comments from your video, then you could use public sources like Reddit, for example.
- [18:58]
Michael Stelzner
Well, if you haven't been tracking There have been some major AI updates over the last month and the changes just keep coming. I know this feels overwhelming, but what if I told you you can actually thrive in this rapidly evolving AI frontier? The marketers succeeding aren't the ones with the most AI tools. They're the ones with the right guidance. That's exactly what you'll find inside the AI Business Society. Our team cuts through all the noise so you can focus on what actually moves the needle in your marketing. As member Shannon Caldwell told us, quote, these trainings are filling the gap between knowing what AI can do and knowing how to do it. Are you ready to thrive with AI? Join me inside the AI Business Society by visiting social mediaexaminer.com AI all right, so far what we've talked about is a couple different things. Number one, don't overthink it and you're the human. You get to choose how you want to direct AI tools like ChatGPT to creatively accomplish whatever you want to accomplish. Number two, well, actually, I got them flipped. The other one is don't be lazy, for lack of better words is what I'm hearing you say. Don't just use AI to create. Create stuff that doesn't have any substance or story. I think we call that AI slop inside of the AI community there's a lot of that garbage, so you still should be actively involved. And then of course, use AI to get insights before you actually begin creating content. Those insights can come from your own posts and comments you've had on those posts, or they could come from someone else's, for example YouTube video and comments, or other sources like Reddit and Quora. So let's say that we are now ready to start coming up with ideas or researching ideas for our next post or video. Talk to me a little bit about some tips and ideas of how we could use AI to help us at this part. Of the ideation and research process that's maybe not outside information, but maybe stuff that actually is inside of our business. Right?
- [21:00]
Sabrina Romanov
Yeah. I think one of the most underrated ways to use AI is to help you document what it is you're doing every day in your business business. Like the meetings you're having with clients, your internal team meetings, your meetings with partners that can be recorded by AI and then spit out like draft social media posts based on all of that content. So I'm a really big fan of when Gary V. Says create social media content just by documenting what you're learning or doing. And I would take that a step further with AI. Like the 20 hours per week that you save. Right. Use that to dive in further to learn, to teach, to engage with clients, record those interactions and then have AI every week send you a summary of like, these were the top five most interesting takeaways from the meetings that you had this past week. And that can be the basis of your social media posts.
- [21:48]
Michael Stelzner
Let's explore this from a tool perspective. Like, for example, anybody who uses the Google ecosystem, right, has an AI note taker built into it. But what's your thoughts on? Because I know some of the people listening might work for slightly larger companies. It might be like, have red flags go up about, you know, using recordings of meetings with AI. What's your thoughts about anonymizing or getting permission to use this stuff? Or said another way, how can we do this in a way that doesn't kind of ring any privacy bell concerns?
- [22:17]
Sabrina Romanov
Yeah, I think staying within the ecosystem that is already allowed for you. So if you're a Microsoft shop and you have Copilot joining the meetings, copilot producing the transcripts, you can also use Copilot then to draft some social media posts and shoot you an email and outlook. Right. So that way it's all contained within your ecosystem. And same goes for Google. If you're using Gemini for meeting notes, then you can use Gemini to create draft social posts from those notes. And so I would just stick to the ecosystem tool, what you're using, especially if privacy is a concern. And obviously, like, it can be touchy to use it with customers and clients, especially if they don't know that. And so just be upfront about it, you know, like they already know that AI is joining the call, recording the meeting. Just ask them like, hey, if there are some key takeaways here from our conversation, I'm happy to anonymize it, but would you let me draft a few social posts? I'd love to tag you in them if you'd love to repost to your network and share some of your insights as well. Like I'm all for transparency. Right. Like everybody now is used to AI joining their calls. There was a time when that wasn't true and people would try to hide it. They'd hide their AI bot joining, they'd change the name of it and say it's someone else. Right. I don't like that approach. I think you should just be transparent. And some, you'd be surprised. Some people would say, oh, that's a great idea. Can I use it for my social media posts too? Because I don't have any time to write content. So just be transparent about it when, especially when it's external facing like clients or partners.
- [23:41]
Michael Stelzner
I like this a lot. If we are recording a meeting, let's say we have a meeting and maybe it's a brainstorming meeting where we're meeting with a couple of our team members and we're ideating on some topics of hypothetically what we want to write our social content on. Right? And it's just me maybe and one other person or two other people and we turn on the note taker once we get that raw transcript or even if we get a summary of the meeting. Do you have any suggestions on how we could prompt. Prompt that transcript into a tool like Gemini or Copilot or chatgpt to kind of take it and look at it through different spins. So it could be fodder for AI?
- [24:18]
Sabrina Romanov
Yeah, I mean I, at this point I usually recommend folks build some kind of custom GPT that's trained a little bit on your voice and business. That way it can like feed that context in, in a very like natural, organic way. So let's say you have that transcript or meeting notes from Gemini and then you can drop it into either gemini or custom GPT for ChatGPT. And that's trained already on your social media posts and your business and your terminology. And then you can have it spit out like three drafts based on analyzing that transcript. So I personally do this every week for my latato office hours. It's basically a time when anyone in the community can kind of join and ask questions. And I have a prompt that's like extract the three most frustrating challenges that people have had and expressed during the call and I'm going to make content about that. So instead of like a brainstorming angle, the angle is like, oh, these are the challenges that people are facing with my product. Here's either how to solve it, or if I didn't solve it yet, it's on my roadmap. Here's what I'm thinking for the roadmap. So there are all kinds of, like, variations for how you can do this. And I encourage more people to, like, adopt AI in ways that don't add additional time and work. Like, you're already having these conversations with customers and your team every single day. It's just about choosing, choosing which context makes the most sense to be the source of your social media ideas.
- [25:35]
Michael Stelzner
Okay, I'm starting to connect the dots here and I really love that example you gave about, like, you're doing an office hour with some of your customers and you're using a tool, whatever tool you're using, and there's a transcript going on in the background, and you're putting that transcript into a custom GPT or whatever system you're using, and you're asking it to kind of pull out what it believes are the most challenging questions that were asked. And then you kind of implied, but I'd love to dig in on this a little bit deeper. I think our next question is probably going to address this. But you kind of hinted at, like, you've probably already trained it on some of your content. So why don't we get into that, that logical next question, which is, how do we take these source materials, whether they be from a live experience or whether they be from support tickets that are coming in. I mean, it could be almost anything, right? How do we do that and actually create original content with that?
- [26:34]
Sabrina Romanov
The way to think of it is like, those are the sources and ideas for your content. It's still up to you ultimately to, you know, finalize the draft, add your unique point of view, your takeaways. And so like, I have a flow that has all of these different sources and then just create drafts for me and I can just, you know, completely remove a draft if I'm not resonating. And like, I don't think this is interesting. So there's an element of, like, taste here. Like, oh, this will do really well with my audience, or this is really valuable for my audience. So that's already one filter. And then the next step is to personalize each post. Like, again, add in your unique perspective. Like when somebody was asking this question or complaining about this issue, here is where I was confused. And like, maybe that's not captured in the meeting notes, but when you reflect back on it, like, that's something that you can add that's unique to your post. Like, I was confused why the customer's asking this, because I thought it was clear, but I realized my perspective is too technical and, like, the button being named this way was super confusing. And I had no idea. And it just taught me that, like, I need to have a more of a beginner's mindset when it comes to testing the product, designing the user experience of the product. Like, I'm just getting an example of where I might talk about a customer frustration and add my own perspective to it to humanize the post a bit more. So it's really about combining, like, AI's documentation of what's happening, what you're learning, what you're teaching, how you're interacting with clients and your team, and then adding your own story on top of that before you hit publish.
- [28:01]
Michael Stelzner
So let's talk about this custom GPT that you kind of alluded to very briefly. First of all, I think I heard you say you've trained it up on your style, so talk a little bit about out how that would work. And then I think you had an example of a viral video copywriter that you had created. Does that, does that sound right? Maybe you could share that story as well.
- [28:20]
Sabrina Romanov
Oh, yeah.
- [28:21]
Michael Stelzner
Yeah.
- [28:21]
Sabrina Romanov
So for training your custom GPT, it's super easy in chat GPT, if you click explore GPTs on the left sidebar, then you can create your own and you can keep it private so nobody else has access to it. And what you do is you take your posts, let's say your company already has a bunch of posts, or if you're a personal brand, a lot of your posts, you can feed them into the custom GPT to train it on your voice, your business, your style of writing, speaking, so that every time you ask it to write a new post, it's going to use all of that custom information you've uploaded to write your new post. And I recommend that. So it's like highly tailored to your business. You don't have to teach ChatGPT every single time. This is my company, this is what we offer. Here's the CTA I wanted the post to end with. Like, it will just have that stored in its memory and instruction. For example, I have one custom GPT that I think has over 20,000 users so far, according to my GPT dashboard. And it's just for brainstorming viral video hooks. So it's been trained on a thousand hooks I've curated from TikTok and Instagram. That first line in your video, if you guys are hormozy fans, like he now says he spends 95% of his effort on the hook. Like, that first sentence scene, visual hook, like, everything, that first five seconds. And so I have a custom GPU just trained on that for TikTok and Instagram. Thousands of people are using it, so that one's public, but that's an example where it's, like, a very specialized use case for content creation.
- [29:46]
Michael Stelzner
Love it. Okay, so up to this point, we have talked about gathering up information in the early parts of this discussion, and we've talked about how you could create a custom GPT, train it up on some of your best examples of writing, and ultimately. And in your case, were you with this custom GPT that you created, did you train it up on. On your writing, or did you train it up on, like, a whole cacophony of really strong copywriting that transcended Sabrina?
- [30:14]
Sabrina Romanov
Yeah, initially it was the latter because, like, I didn't even know what I wanted to sound like. And this is where I keep alluding to that beginner phase as a creator. And it's like, you don't even know who you want to be. It takes time to, like, figure all of that out, figure out your voice, like, your true way of speaking and writing. So in the beginning, my custom GPT was trained on a bunch of people I admired as copywriters. And by the way, I. I still believe copywriting is at the heart of amazing content, whether it's video or writing. That's why I talk a lot about, like, copywriting as the basics and the fundamentals. But, yeah, I mean, initially I just sourced it from people whose writing that I loved. And then as I started to figure out my own voice, then I replaced all of it with my own writing. And I think that's the natural progression. Like, in the beginning, you just. You just don't even know. And content creation is so overwhelming. Like, almost like the last thing you want to think about is, like, having to write, write in a specific way. And so that's okay. Just, like, take some of the people you admire whose communication, like, you love their storytelling. They resonate with you personally. Take that copywriting as an example for your custom GPT. And honestly, as you get going in a couple months or so, you're going to replace all of it with your own writing because you'll get to a point where you're like, this doesn't sound like me, right? And like that. And that's great. That's like, amazing progress and evolution. When you get to that point, you'll swap out everyone else's writing and just use your own as your voice.
- [31:37]
Michael Stelzner
So we're going to get to your software solution in just a second. Kind of explain how it helps. But I just want to unravel a little bit here because up to this point we've talked about the importance of getting some research done and gathering information. It could come from outside sources like comments on your posts, or it could come from transcripts from meetings that you've done or even lives that you've done on any frankly, platform, even YouTube. YouTube. And then we create a custom GPT with some of your best performing work. And I think the part that's missing is like before we get to your product is that, let's say we're now ready to actually create the whatever it is, the transcript, the video. Do you recommend creating a custom GPT? In my case it's Claude projects that I use pretty religiously. But do you recommend or a Google gem or whatever people use? Do you recommend really dialing in the creation of that so that it can create something that's going to ultimately connect on the social platforms? Because you didn't grow to 700,000 by just randomly creating content. There was some sort of strategy employed here. And I just want to talk about that part a little bit.
- [32:46]
Sabrina Romanov
Yeah. And on that last part, as I was listening to you, the question that my initial response was going to be like, don't overthink it, right? Because your, your voice will change. Like what you consider good, what you consider like, oh, that sounds like me, that that will evolve over the months. So like, I don't want folks to get hung up too much on like trying to create the perfect custom GPT that sounds exactly like them. Never expect it to because like you're always growing as a writer. Your bar continues to rise and AI will like always lag behind that in some way. And that's okay. That's why you're going in there and adding your perspective and massaging the post until it like hits truly. Right. But I would say in the beginning, like volume is very important. Important volume and consistency. Most beginners who start out in 100 days, they quit before then. I see it all the time. And like the biggest reason I hear is people are overwhelmed. They don't know which platform to start with, how many times to post, when to post. Do they have to follow up with all the comments? Was their post any good? Like how do they know? And so in the beginning I just recommend consistency, like really posting multiple times per day and it could be different platforms. So it could be one on LinkedIn, in one on Twitter, one on TikTok. Right. You have to experiment. I'll say. For me, I never expected short form video to take off. Like I was experimenting just to get outside of my comfort zone, secretly hoping I wouldn't have to do this any longer because I was so uncomfortable in front of a camera. I love writing, like, I love reading and writing. But like, you have to experiment with it, see how it goes. The most important thing though is just sticking to it for the long term. So when I started creating content, content, my goal was just to stick to it for five years every day, just make one post. Like that was the initial.
- [34:27]
Michael Stelzner
Five years is a big commitment.
- [34:30]
Sabrina Romanov
I figured in five years, like I'll probably be successful doing this.
- [34:35]
Michael Stelzner
How did you know it was working? I mean, like, what were the metrics you were looking for as you were running these experiments?
- [34:39]
Sabrina Romanov
Yeah, I mean, so different platforms give you different metrics. One reason why I do recommend TikTok for beginners is because it gives you so much data and it doesn't punish you for posting multiple times per day. Other platforms like LinkedIn, you don't want to post more than once or twice per day because, like, that's just excessive on that social network. People will just unfollow you. But the way TikTok's algorithm works is like they show your post to 200 people first based on your hashtags and the contents of the video. If those 200 people like it, then they unlock it to 500, then they show it to 1,000. If those 1,000 people like it, they show it to, it's a 5,000. And they don't punish you for posting multiple times a day. So you can post five times a day on TikTok and get that data back really, really quickly. So test five different hooks, five different topics, and you'll see very quickly, within hours which of those hits and which doesn't. So in comparison, if you're doing that on LinkedIn, it would take five days because you would be doing one post per day. So depending on the platform, Twitter is also great. Twitter and threads, they don't punish for posting often. And what I mean by that is like, if your post is bad, like nobody is going to see it, so it's fine. And like Twitter doesn't hurt the reach of your future posts just because you've posted frequently. So yeah, that's just my general advice for beginners. Like, post often, as often as you can honestly stick to the platforms. That don't punish posting often so you can get feedback really, really quickly.
- [36:04]
Michael Stelzner
Where did your time savings come in once you started doing all of this? Was it mostly on the research and ideation side or the creation side or where was that time savings?
- [36:13]
Sabrina Romanov
So it's in a couple different areas. So definitely the research and ideation is a really big one. Another one is repurposing existing content. So in the beginning this isn't a problem. You don't have any existing content, but as you grow, you have a ton of existing. I now have a ton of existing content that I can just repurpose. And this is primarily what I built blotato for. The core use case was just to help me repurpose my YouTube video into social posts. And now I can use it. I can repurpose a YouTube video from six months ago into a social post. I can repurpose a TikTok video video from six months ago into a social post for LinkedIn and Twitter. So repurposing is another one where I it saves me a ton of time and allows me to scale up without, you know, hiring a big team or anything like that. And then the third one I would say is just the publishing piece. Like instead of I used to go to like each platform individually, schedule things out or publish them. And now I have like a dedicated tool where I can sit down for one day, plan out everything, schedule everything for the week, 100 plus posts for the week. And that's taken care of. I don't have to worry about it.
- [37:17]
Michael Stelzner
Let's transition into what blotato is. You kind of hinted at it a little bit and also share how in the world you came up with this.
- [37:24]
Sabrina Romanov
Yeah, so it's essentially an AI powered social media management tool. So think hootsuite and buffer with that social scheduling aspect. But it also has a really strong AI writing assistance. I talk a lot about copywriting and I believe like storytelling, copyright writing is at the heart of great content, whether it's a TikTok video or a LinkedIn post or tweets. And so I really built this tool for myself to help myself scale as a creator. And the first use case for it was just repurposing my own content. So instead of having to sit down every week and stress about writing seven original posts every single week, half of them can be repurposed from previous posts where I can add a new spin. So for example, like AI vibe coding, this wasn't a term when I started using AI coding tools. So I can take a post About AI coding I had from one year ago, Repurpose it with Blotato and with a spin of like, this is what vibe coding is, and here's been my experience with it from the past year. And so doing that, repurposing a post that's already really well written, that I wrote. Right. And that's done well. Repurpose it through blotato, add my 2 cents on top, and then hit publish. That saves me so much time, time alone, especially on weeks where it's like, ah, I like. I don't know if I could come up with seven original ideas. Right now that I'm sitting down, though.
- [38:42]
Michael Stelzner
Let'S expand on that a little bit. So does this mean that Platato has, like, a database, for lack of better words, of everything you've ever created? You can search through it and find anything I've written about coding in general. Right. And then, like, explain how you instruct it and kind of what it does.
- [38:58]
Sabrina Romanov
Yeah, exactly. So it has all of your published posts. You can click add to remix, which will add it up as a source for a new post. Or you can literally clone the post if you just want to make any minor changes. But the heart of Blotato basically uses sources. So it could be YouTube videos, articles, TikTok videos, podcasts, and then runs it through an AI writing assistant trained on your posts, on your voice, and then spits out drafts for all of the platforms. I think I integrate with nine. And then it has the scheduling component so you can schedule everything out. And then recently I've added AI image generation and faceless video generation.
- [39:34]
Michael Stelzner
Ooh, yeah, that's exciting. Talk about that.
- [39:36]
Sabrina Romanov
Yeah, sure. I actually just had a video reached 30 million views, which was a faceless video. It's. It's like a fun pov. You wake up in video. But the reason why I added asset generation is because I looked at the data of, like, thousands and thousands of posts being published, and like, 98% of all social media posts now have an image or video attached. So to me, it's like, okay, cop. I still believe copywriting is the heart of amazing content. But now it's like you need an image or video to even, like, stand out above the noise. And so that's hard to do when you have no background in image or video editing. Like, it's video in particular, I've noticed, kind of scares people off. It seems like a lot of work. But in Blotato, now you can take your post, click, generate AI video. It'll take your post's Content. Generate a video scene by scene based on the content of your post, and then you can post that along with your post.
- [40:28]
Michael Stelzner
Well, what about your voice? Is it an AI version of your voice or is there no voice or what's the deal on the. On the voice side of it?
- [40:34]
Sabrina Romanov
Yeah, you can have a voice. So we integrate with 11 labs, and if you have your own voice in 11 labs, you can use that as well.
- [40:41]
Michael Stelzner
Ah, okay. So for folks that. I mean, we've covered this on other podcasts. 11 Labs allows you to train your voice, but I wasn't aware that it also allows you to kind of lease other voices. So you can get like stock voices, for lack of better words, that will read your script.
- [40:54]
Sabrina Romanov
Oh, yeah, they have like hundreds of voices that are really, really high quality. And then your personal voice, like, if you've tried their professional voice clone, it's amazing quality. So highly recommended if you want to use your own voice.
- [41:07]
Michael Stelzner
Okay. Are you a coder?
- [41:08]
Sabrina Romanov
I am, yeah. So computer science and physics was my background and. Yeah, yeah. So I built the first version of flotato, actually 100% with AI. And I filled myself. So it's my YouTube playlist called Build SaaS with AI. It was really cool. It was really fun documenting it. You know, the MVP was very basic, like even by my own standards, embarrassingly simple by my standards. But I documented it, launched it to my community, got feedback very quickly, and have just been iterating from then. Yeah.
- [41:36]
Michael Stelzner
So this is really intriguing, the idea that you documented the journey of the creation of this piece of software. You had already been writing about AI, right, on LinkedIn and you decided to. When you say document it, does that mean you were creating videos or what were you creating? Like, share a little bit about that documentation journey? Because I think people would find that fascinating.
- [41:57]
Sabrina Romanov
Oh, yeah, absolutely. So I literally started from the very beginning. Like, I hand drew what I wanted my prototype to look like and then I showed here. I'm going to drop it into Claude and have it code the initial first version. Then I'm going to open it up in this tool called Cursor AI, which like, is an AI coding assistance. And every single feature that was in the mvp, I coded it and filmed myself. So it's. It's literally just like a YouTube video of me coding and then talking here and there, explaining what I'm doing. And I just thought it's really cool to document that, just to show people, like, this is what the beginning looks like, the origin. It's really nothing fancy. My biggest fear Is that the MVP was so embarrassingly simple and like, my AI reputation is fancy. It's like, oh, people are going to be like, what? This is just like a simple basic mvp. But I really wanted to show the beginning because I feel like so many people get overwhelmed starting something and finishing it. But like, I just wanted to show here's what it looks like to start from nothing. Like a really terrible hand drawn prototype, convert it into something functioning. I showed people how to deploy it, hook it up to your database, hook it up to your payment system, and then get something live so that you can get real customer feedback and start implementing, improving the product.
- [43:09]
Michael Stelzner
Did that help you? Do you feel like that the act that you actually chose to go out there and show knowing you're a coder. Right. Knowing you could have done this faster and better. What's your take on that? Would you do it again?
- [43:20]
Sabrina Romanov
Yeah, it's funny, I was just telling my husband this, like, I think editing the YouTube videos took longer than the actual filming looking back on it because I was like, pretty new to video editing. But yeah, I would definitely do it again. I've gotten a lot of messages from people who watch that who are just like, so thankful that they could see what it looks like to build something. Because I think it sounds like a scary thing to like build something from nothing. It can sound overwhelming, but it is very doable. Yes. It takes time and effort, like anything. Don't think of it as writing a social media post. Think of it as writing like a mini book. Right. And this is the MVP is just the first chapter in your book and it's going to take many, many hours. And that's totally normal, right? That's okay. In my case, I validated the idea before I started building. Right. Because people were kind of already asking me for what I'm doing to scale myself. And I also made a TikTok video and a newsletter post asking the community for feedback on the idea. And so I had like hundreds of people just validating the idea even before I tried to code something.
- [44:22]
Michael Stelzner
Sabrina, if people want to check out the work that you're doing on the socials, do you have any particular platforms you want to send them to? And then also if they want to check out your company, where do you want to send them?
- [44:32]
Sabrina Romanov
Yeah, absolutely. So to check out blotato, go to blotato.com b l O-T-A-T o.com and then for socials, substack.com is where I do videos and my newsletter posts. Now TikTok or Instagram. You can find me Sabrina Romanov with the mushroom.
- [44:46]
Michael Stelzner
You said substack.com is. I mean, Substack is a huge website. Do they just search for your name when they get there or.
- [44:52]
Sabrina Romanov
Yeah, yeah, you'll find me. Substack.com Sabrina Romanov yeah, Sabrina, thank you.
- [44:56]
Michael Stelzner
So much for sharing your insights with us today.
- [44:58]
Sabrina Romanov
Yeah, thanks for having me.
- [44:59]
Michael Stelzner
Hey, if you missed anything, we took all the notes for you over@socialmediaexaminer.com A57. Be sure to follow this show on your favorite podcasting app. And if you've been a longtime listener, would you give us a review on whatever platform you're listening to and share this podcast with your friends. And don't forget to check out our other shows, the Social Media Marketing Podcast and the Social Media Marketing Talk Show. This brings us to the end of the AI Explored Podcast. I'm your host, Michael Stelzner. I'll be back with you next week. I hope you make the best out of your day and may AI help you become more successful. The AI Explored Podcast is a production.
- [45:37]
Sabrina Romanov
Of Social Media Examiner.
- [45:41]
Michael Stelzner
Just a quick reminder before you go. If you're ready to become indispensable in the age of AI, the AI Business Society is your solution. Join now and secure your discounted membership by visiting social mediaexaminer.com AI I can't wait to see you inside the AI Business Society.