pull down to refresh

HOLY COW BROO! I HAVE BEEN TRYING TO START THE SAME KIND OF PROJECT! But I might not be able to start it anyway cuz of my big exam this year. Let me share the idea with you. First let me tell you existing projects that are similar to your idea:
  • Damus (iOS): You'll find it on nostrapps.com. It is a twitter-like client with a nice UX, but no deep filtering or prioritization. Built more like a broadcast social tool than a reader.
  • Amethyst (Android): Similar to Damus, social-heavy, discovery-light.
  • Iris.to: Browser client. Has tagging, threading, and social graph filters but still misses semantic aggregation. Typical X copy prototype.
Iris.to: Browser client. Has tagging, threading, and social graph filters but still misses semantic aggregation and long-form curation.
NOW LET'S COME TO THE BIG SHOW.
So. I named the project: Noēsis (Greek: "direct knowing") 😤 (Btw, this will be bit hard to build since it is a neural interface) The basic framework is as follows:
User Tells the App What They Want Via:
  • Mood or Intent Setting: Instead of filters, you get a slider — “I want to learn, laugh, rage, wonder.” Adjusts tone, length, and density.
  • Semantic Bookmarking: “I want more content like this” - on press, trains a local vector model. Something like Instagram likes or youtube like on reels and videos.
  • Time-Attention Budget: “I have 7 minutes - give me the best 3 things I should consume right now.”
  • Whitelisted Minds: Add people whose signal you trust, regardless of follower count. Mix “muted minds” to hide all posts from sources you consider untrustworthy.
  • Content DNA filters: Instead of “categories,” the system uses natural language filters: “Crypto but no NFTs,” “AI but not VC hype,” “History of empires in 1-minute infographics.”
Interface Design:
  • Tiled: Modular dashboard of themes. "Black Hole Week" → Shows 5 posts, 3 threads, 1 podcast. Think of it like content islands, not a timeline. Like @Cappy_SC said: A trello dashboard, something like that?
  • Inline Summaries by AI: TLDRs, not headlines. Each tile has an AI-based key point digest. (RSS + ChatGPT?)
  • Highlight/Remix Feature: Highlight sentences → Create new posts, visuals, or memes.
  • Temporal Anchors: “Show me posts similar to what I liked last March when I was in a different headspace.”
  • Explain-Why Button: Why did the app show this? Transparent vector matching + user behavior loop. Like Brainly.in Additionally, you can attach tiny Nostr zaps to content you like. Curators can emerge naturally.
If you want relay health, popularity, and content type support:
  • nostr.watch – Relay performance metrics, uptime, and reach.
  • nostr.band – Search + analytics. You can see top users, event types, content formats, trending topics, etc.
  • highlighter.com – Attempts to create longer-form publishing with Nostr.
  • plebstr.com/relays – Another directory with focus on active, public relays. Most used relays: wss://nos.lol, wss://relay.damus.io, wss://relay.nostr.band, wss://nostr.wine. Hope this helps!
This is incredible. And yes you basically called out most of my private notes on this so far. Yes, we are thinking about this in a very similar way. I will keep you posted!!!
reply
This is incredible
Thanks!
you basically called out most of my private notes on this so far
lol
I will keep you posted
Sure! I'd like to help too if possible.
Btw, as for help, I think this might come in handy for your project. I am building a behaviour prediction engine through python that logs a user's actions, their moods and time of action etc. to decide the percentage of them doing a certain action in future as per the prompt (sort of small scale LBM). I haven't commited it to github yet but the basic skeleton is ready along with the files.
I think it will help you in identifying and predicting content based on your users' interests. If you require I can inform you of its progress.
reply