InfoPulse
An AI news-curation backend paired with a swipe-card iOS reader.
Google Apps Script · Gemini API · Swift · SwiftUI · Discord API
Problem
Around 80 RSS feeds produce far more than one person can read, and ranking purely by relevance skews the feed toward anxiety-inducing news.
Approach
A Google Apps Script backend crawls the feeds and scores each article with the Gemini API, weighing genuine curiosity and delight alongside topical fit. A Swift/SwiftUI iOS app turns the result into a one-article-at-a-time swipe deck.
Outcome
Personal “For You” and objectively-scored “Global” tabs keep two different axes from being confused. Read or swiped cards never resurface. Recent Spotify history feeds back into scoring via OAuth2.
Detail
A two-part system for reading a personal RSS diet without the noise. A Google Apps Script backend crawls around 80 tech, business, science, and culture feeds, scores and categorizes each article with the Gemini API, and pushes the best ones to Discord as rich embeds — with logic to work around Discord's rate limits and Cloudflare's bot-blocking along the way. Scoring also weighs for genuine curiosity and delight, not just topical fit, to keep the feed from skewing toward anxiety-inducing news.
On the reading side, a Swift/SwiftUI iOS app turns the curated feed into a one-article-at-a-time swipe deck for commute and break-time reading, split into a personal "For You" tab and an objectively-scored "Global" news tab so the two aren't judged on the same axis. The design leans into that same growth-oriented philosophy: a strength meter instead of a bare score, a "today's Nth read" counter instead of a read receipt, and a pull-to-refresh gesture on the card itself. Once a card has been read or swiped past, it doesn't resurface. The backend also pulls recent Spotify listening history via OAuth2 to factor current music interests into article scoring.