Thanks for this question. I understand that goHighLevel is a closed marketing application platform, whilst Openkoda is an open-source enterprise application platform with pre-built application templates for a quick start in specific industries (now: insurance and realestate, more to come).
Maybe something like the old tweetdeck where you could chose different feeds, or maybe a way to browse original inputs (if it's something you really care about), most popular links and so on. I am just so over the scrolling timeline look of social media.
One thing I had thought of implementing (but never got around to) was a UI that would look at a social feed and instead of displaying a vertical list of posts, would use a 2d canvas and let you size posts by engagement/poster reach. So if lots of people were retweeting something, you wouldn't see 30 copies of the same post but one large version with a lot of nodes around it representing the people you follow and their 2-cents comments.
"or maybe a way to browse original inputs" - You can see a timeline of the sources if you tap/click on any of the items in the feed. Guess that isn't obvious enough.
"chose different feeds" - If you click the filter button you can choose multiple categories like tech, business, social media etc. This feature needs to be more obvious too. Or were you thinking of something different?
I like the idea of bigger cards when they have more sources. Don't think it would work well on mobile but certainly a tablet or larger.
1. Summarization - Though short, microblogs still have lots of non essential info (This paragraph could be a sentence). Summarization gets rid of this.
2. Semantic Similarity Grouping - So many headlines and microblogs are essentially saying the same thing. I only need to see it once.
Prior to AI, an algorithm that could identify objects in a picture seemed impossible. It's opened the door to countless use cases and has already redefined how we interact with technology to an extent. But it's early, like the internet was early in 1995.
And yes tons of scams and overhyping by people who clearly don't understand the technology themselves. Those people are always going to be around, they capitalize on excitement, regardless of whether the underlying thing is legitimate.
I've been subscribing to - AI probably won't be as bad or as awesome as everyone thinks.
> Prior to AI, an algorithm that could identify objects in a picture seemed impossible.
... I mean, kind of by definition, surely, in that CV as a whole has always been broadly categorised as AI. Computer vision's been around since the 60s, though outside of narrowly-defined problem spaces like OCR didn't really become useful til the 90s.
The current AI bubble is largely based around generative AI, not CV (CV was the _previous_ AI bubble; remember self-driving cars?)
Fair, needs work. Would you be more interested if you could control the style of all the images? Like realistic, illustration, cyberpunk, clip-art, etc?
Sure. The site on arrival feels like an ugly PHP example app design from 2008. Use some AI-looking metallic gradients for logo, or anything less orangy-brown.
The use of AI-generated images for the news items is totally dystopian, freakish, and dysfunctional: hallucinated faces for obituaries, AI bias all over the place, surreal representations of things... But I like it! It's a strange feature on its own seeing how AI may (mis)represent a headline. Prompt it to use even more different visual styles, and apply filters (ie color filters) so that the UX doesn't become visually repetitive.
The news text underneath is just garbage, I don't get it. Work on putting out a good AI-generated summary instead, plus real links to real news sites that your LLM/ranker/classifier thinks are the best links to get that news, which will also give attribution for your sources. Bonus for summaries+links to social top ranked comments too. Get rid of snips within snips (which do not work).
Add a short sentence-summary under the headline, or make headlines longer when applicable so that they are more descriptive.
