Transcription
I just built an AI system that completely automates Instagram research and basically predicts viral content with scary accuracy. And this thing monitors your competitors 24/7, analyzes their best performing reels, and literally tells you exactly what to post next. And um the crazy thing is that it costs literally $2573 per month.
Okay, it costs $2573 per month to replace a full-time content researcher. A full-time content researcher that will give you a uh full competitor list. Okay? It'll give you you all of your best performing reels transcribed with the hook and an analysis of the hook. It'll give you the caption, the views, the likes, comments, and it will also go out, find your five best competitors, and scrape their best videos, their best performing videos, and put them in a spreadsheet with transcript, a hook analysis, caption views, likes, every all the data you need.
And then what it does after that is it actually goes through all the data for you and uses AI. It uses Claude 4 to go through all of that data, process all the data, process all the hooks, process all the winning components, and then deliver you this weekly breakdown with the links to your top competitor videos, right? So, you can go copy all these videos, like this is a video that I would probably copy, right? Um, and then winning patterns, so like strategy insights into the core underlying themes of what's going on here and why this content is performing well. And then it gives you 10 hook ideas. Okay, these are basically like templates or hooks that based on data, based on the data is pretty much like proven to perform. And guys, this spreadsheet automatically updates every single week with new data from your competitors. And you don't need to do a single thing. It literally over time updates every week to get new content and new poppingin stuff, right? So, you literally don't have to do anything. Okay?
And this research agent basically runs every single week or if you want it to run every day, then you can have it run every day. It doesn't matter. Anyways, let me show you how this whole thing works. Okay, so your sheets's going to be blank to start off. Right now, here is what you do. You just click this little button right here. Okay? And that is going to search and find your top competitors. And then it's going to get their profile data. And then it's going to fill in that competitor list for you. Okay. So, looks like it just finished up there. Boom. All right. So, there we go. Boom. Our competitors are now filled in. Um, basically what it did there is it just searched my account, right? So, this is where you fill in, you know, your own username, but it basically just searched my account and found similar accounts. Um, this will work really well if you have a big account like me. But if you have a smaller or like super niche account, then you might need to enter someone bigger in here to get good competitors, right? But anyways, that just gives you a starting point, right? So now I'm going to reorder this list. I'll probably move Ben Kuza up. I will move. So we got Nick, Ben, Callaway, RPN, and Sue. Okay, those are my top five. Cool. Now that my competitor list is set up, I don't really ever need to touch it ever again, unless I just want to rearrange the top five so that it scrapes different people, right? So I don't know, maybe later on I might move other people into the list so that they get put in the mix, right? Whatever. But uh this is the only manual part is just curating the competitor list.
So now let's imagine it's Sunday. Okay, this is set to auto run every single Sunday. So let's test out the trigger. So it's going to clear any previous data, right, from the previous week. Um I like having it do that because you know every right, like I said, I wanted to update um over time, right? So, um, at the end, it's kind of going to look over all of the data and give you a breakdown. And if you don't clear the the previous week data, there's just going to be more and more data. It's going to fill up, right? And so, like, you don't want to give too much data to your agent, right? You kind of want a small selection of really good performing reels to base your data off of because if there's too much stuff in the spreadsheet, it's just going to mix up your results. It's not going to be as as good, right? So, that's why I like to clear the sheet and auto update it every every single week, right? So, um yeah, that's just going to happen every single week.
So, we can see here that it just um scraped all of my reels and now it's downloading and transcribing them. Okay, so now it's like doing the hook analysis. It's kind of breaking down each hook and it's using four it's using um OpenAI 4.1 Mini. That's the chat model that's using for this step. Uh it's a relatively fast and light chat model. It's not super smart, but it's good enough just to like because we're having to do very basic analysis, right? That this agent later on is going to be the smart one that does like the crazy analysis. Um but we can see okay, it just it just put that. Okay, perfect. Wow. Just look at that data. Okay, full transcript hook analysis. It's got the thumbnail and it's got the URL, bro. Like, this data is um amazing. Anyways, onwards.
So, now it's doing my niche outlers. So, now it just um ran over my top five people in here. 1 2 3 4 5. And now it's going to research all of them. And it's basically going to find here's what it does. So, it gets their 20 best reels and then it sorts them by views and then it limits it to the top three. So, it basically for each person in your competitor list, it's pulling their 20 latest reels and then it's getting the top three by views and that's what it's using, right? So, as we can see that just ran. Um, so boom, we just scraped um Nick Sarv and these are his top three posts, right? His top three like most recent posts. Um, so it's just going to do that for all five of the top competitors and then this sheet will be filled out. Oh, okay. So, it just got done with that phase two. Now, it's doing the last thing. Okay, the last step here is it's just it's this um AI. We're basically using Sonnet 4. Okay, Sonnet 4 is like a clawed model. It's way way smarter than like these little guys, right? This is the guy that does like the ideation that basically processes all the data. So what he's doing here is he's basically getting all the rows. He's getting all the rows of our own content and he's getting all the rows of our niche outliers content. And then he has some like very specific instructions in here on how to come up with the hooks. Right? I'm showing you the system prompt right now. So if you want to copy it, go ahead. Um, but yeah, cool. So, he's basically the smart guy that goes through all the data, comes up with the hooks, and as you can see, it just sent me a Slack message.
So every single week, I get a breakdown that gives me top competitor reels, so I can basically copy all these videos, and it gives me winning patterns and some hook ideas. People used to pay me $2,000 a month to do this, and like now you can run it literally whenever, and it takes 10 minutes, right? I could run this again for as many accounts as I want. Like all it takes is I just got to enter your username in here and like I could give you an entire spreadsheet full of all your data and like this Slack weekly breakdown in literally like 20 minutes.
So anyways guys, if you want the template for this um automation, you can join my school community down below and click this Instagram research agent. I have an entire overview and demo and then I have the setup guide with the JSON file. Okay, so this is the secret sauce file that if you download this, you can literally go into NADN and create a new workflow and you can just import from file and then you can just click that file and boom, now you have my entire workflow that I invented with my own brain. Okay. So, um yeah, if you want the oneclick copy and paste this thing, then you know what to do. Hit that first link in the description. Um I hope you guys found this video valuable and informative. And I hope this shows you what's possible with NA and automation. Yeah, that's pretty much going to end the video. Oh, right. Drop me in the comments if you have any questions about how this whole thing works. Um and I will answer them. Okay.