I've been working for a while on an intelligent RSS news reader. At its core, it works just like any other RSS news reader: you subscribe to the sites that you want to follow and you get delivered everything published there. The difference is that the news aren't sorted chronologically, they're sorted by how close they are to your interests. Zuperlist also remembers when you open the web app and separates the news by these moments. When you open it, you see what's new and noteworthy since your last visit. Another nice feature: related news are grouped together.
There are some ways to bypass the recommendations engine:
- you can mark specific sites as highlighted and see everything published by them, before anything else
- you can mark specific topics as highlighted and see everything that matches them. For example, mark Bitcoin as highlighted and see everything about Bitcoin in a separate block, regardless of recommendations
- everything that doesn't match your interests is still available at the bottom of the news feed, where it's categorized by topic and by site
There are also some measures taken against creating a perfect filter bubble:
- you get a quick glimpse at the most shared news, even if they don't match your interests
- you get a few recommendations from sites you didn't subscribe to
- from time to time you're shown a few items from a topic that's outside your interests
Sure, it's not perfect yet, but it works ok.
I've been using it during the development and I love it, although I may be a bit biased :)
If you're intrigued by the idea, give it a try and let me know what you think.
The zuperlist.com site works on both desktop and mobile, so there's no need to install an app.
- the landing page is well done: it is very clear what this app does and who it is for!
- I liked the onboarding flow
- one thing I didn't like: Your app is very well designed, and I can tell that you spend a lot of effort into every last detail of the design... but for reading, and especially skimming over article headlines, the tile design isn't ideal in my opinion. It looks good, definitely! But I think for readability, a simple text-only list, like on HN would be more suitable.
> Another nice feature: related news are grouped together.
one reason I signed up, was to check out how you would do that. maybe I didn't select enough feeds to see it... but do you only group articles by category tags (e.g. Startup, Europe, Games, etc.), or do you group articles within the same category, if they write about the same topic. E.g. within the Startup category, there are two articles about the latest Uber scandal. Are they grouped together? If so, do you do that manually, or do you have some smart ML implemented?
Also, for your press kit: you might be interested in my app PressKitHero (https://presskithero.com), to present a bit more information about you and your company.
I decided to go with the tile design for the MVP precisely to improve skimming. Just by looking at the image you could get an idea about what the news/article is about. But I understand your point. One potential improvement would be to allow users to select how to display the news in each block: thumbnails or titles.
News/articles are grouped both by category and by specific subject. When multiple items are about the same thing, you'll see, for example, something like "TechCrunch + 3 others" above the item's title. If you hover over it, you'll see the other 3 items. It's all done via ML. It's not perfect yet, but it can only get better from here :)
I'm also working out a smart RSS reader with feedparser. Some of the features I plan to implement are similar to features you listed. I have a few goals I'm trying to fulfill with my own project, for example a powerful query engine and tagging / sentiment analysis system.
I'd love to chat for while about where you plan to take this project, maybe we can figure out a way to avoid duplicating work or even combining our ideas and effort into one project if it seems that we have similar trajectory in mind.
Well, first I'm trying to validate this idea and see if enough people find it interesting. Any future direction depends quite a lot on the feedback that I receive.
Kudos, but my bias is against feed content selected for me, and towards me deciding what is of interest - which is somewhat at odds with your unique selling point.
What led you to decide on a trial followed by paid subscription model?
I started this because I wanted a single place for me to deal with both high-volume and low-volume sites/blogs and give me control in both scenarios.
Regarding the subscription model: I wanted to validate the idea and see if it could sustain itself. I didn't add a free version (until now, at least) because the whole recommendation engine part requires a little bit more computing power, hence it costs me to keep it up-and-running.
There's few enough quality RSS readers that if a free (or just cheaper) version didn't have the key recommendation feature it wouldn't just be adding noise.
I didn't include this feature in the MVP. However, if you want me to manually delete your account, log in, go to Menu > Send Feedback and just type "Delete my account."
Very cool. A little while ago we created Winds, an open source RSS reader that uses machine learning to figure out what you're interested in. It's leveraging React/Redux, Stream and Algolia. An updated version is in the works: http://github.com/getstream/winds
Looks neat, one thing I love about my RSS reader still is it doesn't try to guess what I like. I have all my sites in there, many of them are blogs, and while most of them focus on a topic, they also post random stuff that's just interesting. I'm afraid of losing that. It's like what Facebook and Twitter do "for" me when they pick and choose what I see there.
It's like browsing the shelves of a library vs. using Amazon's recommendations. Sometimes a recommendation from some algorithm is fine, sometimes it's cool to just browse the shelves.
My feed reader is still one of my favorite things because it shows me random stuff I had no idea I'd be interested in because the sites I've picked are already doing a good job. I'm afraid if something else filters out the things it thinks I don't like then I'm going to lose out on some really interesting things. This is exactly how I found Hacker News, I had no idea what this site was before it just turned up in my feed reader a few years ago.
(I'm way old school when it comes to my feeds probably)
I understand. That's why I added the feature to bypass the recommendations engine for some feeds, so that you could still see everything published by favorite blogs. That way the recommendations engine can only watch high-volume sites.
Thanks for the feedback! Well, I hate it when some sites show you a large monthly price and in a corner it says "billed anually." So I wanted to be as clear as possible.
I've been working for a while on an intelligent RSS news reader. At its core, it works just like any other RSS news reader: you subscribe to the sites that you want to follow and you get delivered everything published there. The difference is that the news aren't sorted chronologically, they're sorted by how close they are to your interests. Zuperlist also remembers when you open the web app and separates the news by these moments. When you open it, you see what's new and noteworthy since your last visit. Another nice feature: related news are grouped together.
There are some ways to bypass the recommendations engine:
- you can mark specific sites as highlighted and see everything published by them, before anything else
- you can mark specific topics as highlighted and see everything that matches them. For example, mark Bitcoin as highlighted and see everything about Bitcoin in a separate block, regardless of recommendations
- everything that doesn't match your interests is still available at the bottom of the news feed, where it's categorized by topic and by site
There are also some measures taken against creating a perfect filter bubble:
- you get a quick glimpse at the most shared news, even if they don't match your interests
- you get a few recommendations from sites you didn't subscribe to
- from time to time you're shown a few items from a topic that's outside your interests
Sure, it's not perfect yet, but it works ok. I've been using it during the development and I love it, although I may be a bit biased :)
If you're intrigued by the idea, give it a try and let me know what you think. The zuperlist.com site works on both desktop and mobile, so there's no need to install an app.
Very nice! Here is my first impression:
- the landing page is well done: it is very clear what this app does and who it is for!
- I liked the onboarding flow
- one thing I didn't like: Your app is very well designed, and I can tell that you spend a lot of effort into every last detail of the design... but for reading, and especially skimming over article headlines, the tile design isn't ideal in my opinion. It looks good, definitely! But I think for readability, a simple text-only list, like on HN would be more suitable.
> Another nice feature: related news are grouped together.
one reason I signed up, was to check out how you would do that. maybe I didn't select enough feeds to see it... but do you only group articles by category tags (e.g. Startup, Europe, Games, etc.), or do you group articles within the same category, if they write about the same topic. E.g. within the Startup category, there are two articles about the latest Uber scandal. Are they grouped together? If so, do you do that manually, or do you have some smart ML implemented?
Also, for your press kit: you might be interested in my app PressKitHero (https://presskithero.com), to present a bit more information about you and your company.
I decided to go with the tile design for the MVP precisely to improve skimming. Just by looking at the image you could get an idea about what the news/article is about. But I understand your point. One potential improvement would be to allow users to select how to display the news in each block: thumbnails or titles.
News/articles are grouped both by category and by specific subject. When multiple items are about the same thing, you'll see, for example, something like "TechCrunch + 3 others" above the item's title. If you hover over it, you'll see the other 3 items. It's all done via ML. It's not perfect yet, but it can only get better from here :)
I'd love to chat for while about where you plan to take this project, maybe we can figure out a way to avoid duplicating work or even combining our ideas and effort into one project if it seems that we have similar trajectory in mind.
What led you to decide on a trial followed by paid subscription model?
Regarding the subscription model: I wanted to validate the idea and see if it could sustain itself. I didn't add a free version (until now, at least) because the whole recommendation engine part requires a little bit more computing power, hence it costs me to keep it up-and-running.
And the Hnews discussion: https://news.ycombinator.com/item?id=12932429
It's like browsing the shelves of a library vs. using Amazon's recommendations. Sometimes a recommendation from some algorithm is fine, sometimes it's cool to just browse the shelves.
My feed reader is still one of my favorite things because it shows me random stuff I had no idea I'd be interested in because the sites I've picked are already doing a good job. I'm afraid if something else filters out the things it thinks I don't like then I'm going to lose out on some really interesting things. This is exactly how I found Hacker News, I had no idea what this site was before it just turned up in my feed reader a few years ago.
(I'm way old school when it comes to my feeds probably)
[0]: https://www.bloomberg.com/news/articles/2017-09-24/in-battle...