From the world's news
to your 9-second read.
Four steps. 16 trusted publishers. No hallucinations.
You pick topics
During onboarding you choose from 9 categories — World, Business, AI & Tech, Science, Health, Sports, Entertainment, Crypto, US News. Update anytime from Settings.
We fan out across publishers
Every 10 minutes a Vercel Cron job fetches 16 trusted RSS feeds in parallel — BBC, NPR, The Verge, Ars Technica, TechCrunch, Hacker News, ESPN, CoinDesk, Variety. Each feed has its own 8-second timeout so one slow publisher can’t hold up the rest.
Normalise, dedup, store
Bytes parses each feed (RSS 2.0, Atom, Media RSS), strips HTML, extracts the lead image, and validates every field with Zod. Duplicates are collapsed by source URL hash and the result is upserted into MongoDB.
Delivered to your feed
The curated set lands in your vertical-scroll feed, served from an in-memory cache that falls back to Mongo. Tap any card to expand the summary, follow the link to the original publisher, or save for later.
What we won't do
The principles behind every design decision.
Links to the original source
Every article card includes a direct link to the publisher. We surface journalism — we don't replace it.
Publisher-first, never AI-fabricated
Every headline and summary comes straight from the publisher’s own RSS feed. No language model rewrites, no hallucinated quotes, no synthesised summaries.
Your data stays yours
Bytes only stores your topics, preferences, likes, and bookmarks. No tracking pixels, no third-party ads, no data selling.