Developed an AI version of ProductHunt with AI
date
Jun 4, 2025
slug
launchgenius
status
Published
tags
web
AI
type
Post
URL
summary
this is a project entirely coded by Cursor+Sonnet4. Thank you, AI, because in the past, I couldn't have completed an online project in a week of my spare time. Finally, if you have a project, why not submit it to LaunchGenius and see how AI rates it.
Author
I believe many people are familiar with ProductHunt! Developers or start-ups around the world will submit their own new projects on it every day. If your project is ranked first in the vote, there is no doubt that this wave of launch promotion of your project is successful.
However, this is also accompanied by one thing. In order to rank at the top, many people begin to invite people to vote. There are often dozens of votes but no comments. Generally, the data with much less comments than votes can basically be determined to be asked to vote.
This is very unfair to those developers who don't have much resources, and it often leads to many good projects that will not be discovered at all!
Based on the above background, I developed the LaunchGenius platform, which is a product ranking list driven by AI. The product logic is very simple. The user submits the detailed information of the project, and several AI models conduct in-depth analysis through the project information, and then calculate the final score by weighting average according to the scoring results, out of 10 points.
Finally, the daily, weekly and monthly lists are generated according to the scores of AI.
In this way, the problem of man-made voting is completely avoided. Not to mention complete fairness, at least it can be relatively fair. Of course, there is also a problem with this, that is, the ability of AI is still limited after all, and some judgments may not be so accurate, which is also the reason why I need to introduce multiple AI models to score.
The above is the relevant background and introduction of the project. Next, let's talk about the story of LaunchGenius development. In fact, it was also an impromptu idea to develop this project, because after my other personal tool TranslateX was released on ProductHunt, there were only 5 votes.
Observing others casually post a link on Twitter and get dozens of votes, there's a bit of a gap in the heart. Then, coincidentally, claude sonnet4 was released that week, so I wanted to try a project to test its power, and so I immediately got to work.
After a week of Vibe Coding after work, it finally went live at 5:30 PM. However, to avoid fixing bugs during the Dragon Boat Festival holiday, no promotion was carried out, and post-holiday, social media promotion and advertising began.
In fact, the entire development process was relatively smooth. The code design and encapsulation capabilities of Sonnet4 are much stronger than 3.7, which also means that the maintainability of the code has been further improved. However, it can also be found that there are areas where abilities are lacking, such as daily, weekly, and monthly rankings that involve scheduled tasks and SQL query functions, which are not performed very well. If you are a programming novice, you might really struggle, or even if the project can run, there may be problems. For example, if you are a programming novice, you may not know about time zone issues, or not know about SQL query optimization issues. Of course, once you discover these problems, it still knows how to fix them.
In summary, this is a project entirely coded by Cursor+Sonnet4. Thank you, AI, because in the past, I couldn't have completed an online project in a week of my spare time. Finally, if you have a project, why not submit it to LaunchGenius and see how AI rates it.