How we summarize

InfoReader is a free news-briefing service. We don't employ reporters and we don't rewrite the news behind a paywall — we organize it, and we show our work.

1. We read the sources

Our pipeline follows a maintained list of reputable publisher RSS feeds (plus the GDELT news index) across five desks: World, Business, AI & Tech, Science and Health. Feeds that stop working are replaced, and the list is public in spirit: every story page names its sources.

2. We group the same story together

Articles that report the same event are clustered into one story. That's why stories show a source count — "5 sources" means five outlets' articles were merged to build that summary, and all of them are linked on the story page.

3. We summarize — and label how

Each story gets a short summary and key points, generated once and shared by every reader. Summaries come from stacked free AI tiers (Cloudflare Workers AI, Gemini, Groq); when those quotas are used up, the pipeline falls back to an extractive summary composed from the sources' own text. Every story page labels which method produced its summary. Nothing is published from a single unverified claim without its sources attached.

4. We link you to the originals

Summaries are a starting point, not a replacement. Every story lists its original articles so you can read the full reporting from the publishers who did it.

The model

InfoReader is free for readers — no accounts, no paid tier. The site is supported by labeled advertising via Ad.Plus / Google Ad Manager; the email digest carries no ads. Running costs and the full stack are published on the Free Model & Costs page. Machines are welcome too: our agent & API guide documents the public JSON feeds.