This project has been deleted.

Breaking News

Breaking News Used AI

11 devlogs
23h 48m
•  Ship certified

A desktop app that uses LLMs (HackClub's endpoint) to rank and present the most important news stories. It is built using React and Electron. Basically a fancy RSS reader. You can download a compiled binary from the releases section on github.

Timeline

Ship 1

1 payout of shell 413.0 shells

Konstantinos Fragkoulis

over 1 year ago

Konstantinos Fragkoulis • Covers 11 devlogs and 23h 48m

Minor improvements, ready to ship

Update attachment

I added prebuilt binaries on github.

Update attachment

Today, I fixed article layouts, implemented a system that deletes old cached articles, and made some minor changes to the app. My app is definitely usable as is, and with only some minor bug fixes and usability improvements it will be ready to ship.

Update attachment

Today, I made some changes in the UI. I worked on some article layouts, and generally improved the way my app looks.

Update attachment

Today, I finally added variations for the articles placed in the middle column. Most of them don't look great, but the underlying system is there, and I could fix their appearance relatively easily. I also made an icon for the app, fixed a few bugs, made some parts of the code more consistent.

Update attachment

Today I created a settings page and gave my app true support for multiple RSS feeds. Also, users now can declare their interests based on which news stories are given an importance score. Finally, I fixed some major bugs and improved the way articles are cached and user preferences are stored.

Update attachment

I fixed some bugs in caching the importance score after changing the topics you are interested in.

Update attachment

Today, I improved the way caching works, so that alongside the importance score for each article, my app also saves the user preferences for which the score was generated. Also, I added a progress bar while loading articles.

Update attachment

Today, I used HackClub's chat completions endpoint to rank articles based on the user's interests. I then cache the article's importance to reduce API calls and load times. I also made reloading the website fetch and parse the rss feed again.

Update attachment

Today, I improved the way I parse RSS feeds and managed to correctly extract the largest image available for each article. I also temporarily made images take up the full width of their columns. In the coming days, I will design article layout variations and build a more robust system for parsing RSS feeds.

Update attachment

Today was the first day of development! I created a quick UI to be able to test my code, and got started working on parsing rss feeds and dynamically creating articles. I had never used React or Electron before (or any other JavaScript framework for that matter), so I still need to get used to 30+ second build times 😔. I didn't get to use AI to sort articles based on importance; that's coming in a day or two.

Update attachment