NewsDiary

News reading is mostly throwaway. You read an article, maybe nod, and the thing evaporates by Friday. NewsDiary is a personal news diary that does the opposite — capture what you actually read, annotate it as you go, look back at it day by day, and let optional AI surface bias, fact-check, and historical context only when you ask. The cardinal principle: layers off by default. Your reader stays clean. Toggle the layers pill at the top of the article when you want an AI second-read; everything else gets out of your way. Capture is anywhere. On iOS, paste a URL or use Sign in with Apple and ingest from inside the app. On desktop, the companion **NewsDiary Chrome extension** lives in the side panel — capture the page you're reading, annotate in the same way, and the article + your notes show up on your iPhone next time you open the app. One account, one diary surface, both devices. What's in the app today: • Highlight, written notes, and voice notes on any passage. • Five AI tools — summary, bias, fact-check, historical context, glossary — each runnable per article. Default models picked per task; power users can override per tool. • BYO Anthropic API key (encrypted at rest, AES-GCM). When set, your calls go through your own key and don't touch our platform budget at all. • Monthly cost projection so you can see what AI tools cost you, broken out by tool and by who's paying (us or your key). • On-device summary via Apple Foundation Models on iOS 26+ — when it's available, the article doesn't leave your phone. • Diary feed grouped by day with full-text search across every article, annotation, and entry you've written. Filter by source, month, or content type. • Single-file export (JSON or Markdown) of everything you've saved. • Account deletion with a 30-day window. Mutations stop the moment you trigger it; reads + export stay open so you can leave with your data. Privacy is the default, not a feature. We don't sell your data, we don't share it with advertisers, we don't train AI models on your reading. LLM usage logs hold token counts and cost — never the content of what you read or what you wrote. Full policy at https://newsdiary.app/privacy (also served directly from the app). This is the first internal TestFlight build (v26.5.0, build 2). We're looking for honest feedback from people who read news daily — journalists, researchers, policy folks, students, anyone who treats reading as work. Use it for a week. Tell us what's clunky, what's missing, and what surprised you. Every edge you surface here is one we fix before we open this to the next cohort. Send feedback via the TestFlight app (shake → Send Beta Feedback) or email services@capewesley.com directly. Bug reports with "what you tapped / what you saw" are gold.