I built a TT prototype in 2 days.

@codingai

50.0K followers · usually 1.4K views a post

Views

26.0K

Likes

1.2K

Comments

117

Shares

57

Selling · SaaS

codingaiseen in 4 videos →https://links.yimcodes.com/?utm_source=tt ↗

Why it worked

The video combines a high-stakes challenge (manipulating the algorithm) with a transparent, 'build-in-public' demonstration of a custom tool, making the product feel like a natural extension of the creator's expertise.

Summary

The creator demonstrates how he built a prototype app to manipulate the TikTok algorithm by selectively liking his own content. He uses his own tool, codingai, to showcase the data and results of this experiment.

Structure

1Creator introduces his 2-day app prototype
2Explains the experiment to manipulate the TikTok algorithm
3Shows the data collection process using his app
4Demonstrates the results of the algorithm training
5Concludes that the experiment successfully prioritized his content

Product placement

The creator uses his own app, codingai, to visualize and manage the data for his TikTok algorithm experiment.

Original caption

@Mark Cuban get me on this!

More from @codingai

Their other posts in the index, biggest breakout first.

On-screen text

I created the mvp app In 2 days Day 3 We have an Algo!!

Transcript

I built a TT prototype in 2 days. Mark Cuban asked who has the skills? I have the skills. Help me get this video out. Okay, everyone wants to know, is my algo doing anything? And so here's my little experiment. Things that I like to talk about are tech, software engineering, and I also have videos from this other creator who mostly talks about sales. So my videos, I think I have about 60 of them. You can see 59, and her videos, there's more than twice as many. The experiment is, I'm going to start a new account and then first I'm going to get a random assortment of videos. So remember she has twice as many as I do. I'm going to like only my stuff and hoping that the algorithm is going to start to show me only more of my stuff. In machine learning, you need yeses and you need nos. A yes will be if I like the video, a no will be if I swipe it without liking the video. I'm speeding this up, but I don't cut away so you can see that it worked. For remember, she has two-thirds of the videos in here. I think it's working. How long did this take? It took me about a couple minutes to get onto this where it's almost all of my videos.