Confession: we used Google's fanciest model to generate basic text

Transcript

So for last few months, I have been so busy trying to grow the company that I have increased the expenses to the moon. We were literally using the top-end model for doing basic stuff. We created a tool called Instacrosser, and we were literally using Google's latest image model, which can generate faces exactly to create a fucking text images. Yeah, we have done some amazing dumb shit, and it has costed me lakhs, and now I'm finally optimizing, and dude, I just saw this, like I tried Grob, and the Lava model was so fast. It's very good. I'm really trying to move as much as things possible where I don't really need such an advanced model, because we are doing basic stuff, like honestly, like our presentation app or all these things, they just generate JSON. We are in 2006, and the things which we are doing are not, like we don't really need the latest model for them, and paying so much for them doesn't make sense. It really increases our expenses, which ended up being charged to the user. So we'll try to optimize that, and hopefully decrease our expenses, hence increasing the profit margin, hence being able to be more competitive in the market.