I ditched the algorithm for RSS

I started doing this a couple of years ago. In fact, I found this post when going through my RSS after breakfast.

My biggest problem so far is that the RSS data is often very lacking (body, images, etc) and that there are still lots of content I'd like to filter out by sentiment. Like if I follow a technology outlet I don't want to know that there's an absolute steal on Sonos Arc on Bestbuy because I'm generally not into consuming and I live in northern Europe.

I have high hopes for ML to populate the RSS data and filter content like this. I want to experiment with this.

Seems like your issue is with your RSS reader not RSS itself. You most likely also don't need LLM to filter content a word filter would catch...

If you want to give a word filter a try, you can check out https://lighthouseapp.io/.

Rules are supported in the free plan as well, so it's easy to try out.

Plus it has a new feature which shows the ratio of bookmarked articles for feeds, which makes it easy to know which feeds to unsubscribe if it becomes too much.

I use Feedbin (self-hosted) and it has no difficulty with displaying images.

for desktop i use FOSS RSSGuard which has complex filtering and proper image support. Also a light version without JS or other bloat

https://github.com/martinrotter/rssguard

I've used Feedbin since it started (paid user on a grandfathered plan) and had no idea it was open source and could be self-hosted.

I'm happy to keep paying for the hosted version but I love knowing I have the option to self-host if it goes belly-up. Thanks for pointing this out!