GPTube, my first "serious" side project

Hi everyone 👋 this is another story from another guy trying to build some profitable product on internet so here goes the story.
GPTube is my first "serious" side project, I was working on it during the last 4 months and I'm continuously adding to it more features (while writing now I am working on the implementation of payments system). The story of the idea is not very fantastic, in fact is a simple one. I was scrolling on my twitter and started viewing a lot of new AI products since the launch of chatGPT. So I thought "why not do something with this?".

In those days I was also introducing myself to the world of side projects. Looking the stories of all those guys like Peter Levels, Jon Yongfook or Danny Postma building profitable stuff with its skills and some early capital. Also joining to communities like @IndieCreatorsHQ a LATAM indie devs community. So I told myself: "Saul you have a part time job that can give you some little money to maintain a project, if those guys can do it, why you don't?".
Then I put my hands on and started thinking...

It has to be related with AI.

Since the advent of generative AI and LLMs nowadays everyone should only think in an innovative way to mix this and create something cool. So I knew that the heart of my project was to be AI, specifically sentiment analysis.

It has to be a "Big Thing"!

A disclaimer here: "A side project doesn't have to be a disruptive idea". If your product is a simple one or even a copy of others but solves efficiently a problem. You have your chicken of the golden eggs.
In my case I though about an "scalable" product, by "scalable" I mean something that solves a problem but it has the potential to be used by thousand of customers.

GPTube

With this in mind one day while "Youtubeing" I noticed that various Youtubers have this problem: I'd like to know what my audience thinks about my content (you have likes and dislikes but that doesn't give you the hole picture) and why they like or why not. So I identified the problem, and then...

I though about feedback on internet and identified three types of comments on internet in respect your content: those that give you support or simple ones (positives or neutrals), those that throws you shit (negative comments) and those that give you true feedback (a negative comment in companion with a good advice). So I decided to focus my attention on the second and third type since only negative comments can give you the key to improve.

So I decided to build GPTube a tool that could identify negative comments on your posts for various social media platforms (Youtube, TikTok, Instagram, etc...) and give you feedback from that comments.
For a real use case @Serudda one of my favorite youtubers could identify that in some videos people were telling that his audio was not good enough or that in some videos he should moderate his language because of too many vulgar words to put a few examples.

This use case could be scaled to other social media platforms (TikTok, IG, Twitter, etc), but I'm starting with Youtube since is the social media which has a free ready to use API.

So how GPTube works? You just paste the URL of your video, then you just wait for the results (on the fly if the video has few comments or receive an email later otherwise):

• Sentiment scores from different AI sentiment analysis models on Hugging Face there are plenty of them, I am using this ones: BERT multilingual and RoBERTa.
• Feedback is provided by ChatGPT that could summarize and understand why negative comments are precisely negatives.

But does not all maybe in future GPTube could be used to delete comments from haters (I know this is a controversial feature even more for the cancellation culture but it would have to be final decision of the content creator), this is still an idea.

So what do you think about GPTube?, Hope your comments indie hackers. Please try the early version for free! up to 20 videos until I finish the payments system 💥.