How DailyRite Works
Last updated: October 8, 2026
DailyRite assembles a personalized daily briefing on our servers from public data sources and AI summarization. Your browser extension or the website stores only your sign-in token and a cached copy of today’s briefing. It never stores API keys or your browsing history.
For what we collect and share with third parties, see our Privacy Policy. For limitations on AI-generated content, see our Terms of Service.
Overview
Each morning (and optionally in the afternoon for Pro), our backend:
- Fetches public data for weather, markets, sports scores, and news.
- Uses AI to summarize and organize that data. Google Gemini writes the news summaries, and Anthropic Claude takes over if Gemini is unavailable. Claude also writes the daily lesson and the weather summary.
- Caches the result so your new tab or the web briefing loads quickly.
All AI and search calls happen on our servers. The extension talks only to
dailyrite.org.
Section by section
| Section | Data sources | How we get it | What you see |
|---|---|---|---|
| News (your chosen topics) | Serper, Tavily and Brave Search (news search) → Google Gemini, with Anthropic Claude as backup | We run topic-based news searches with three search providers and merge and deduplicate the results. The AI then writes a short summary of each story in its own words from the search results’ headlines and short excerpts, with links back to the original publications. | Every briefing is labelled at the top as AI-written summaries, and each story links to the outlets it drew on (publication name + URL when available). |
| Weather | OpenWeatherMap; U.S. National Weather Service (NWS) | We geocode the city and state you provide, fetch current conditions and a short forecast, and pull active U.S. government weather alerts when applicable. | Current conditions, forecast, and any active alert boxes (with alert source). |
| Markets & crypto | News search (same pipeline as above) | Market and crypto coverage is editorial: headlines and summaries about stocks, the economy, and cryptocurrency come from our news search, the same way as other news sections. We do not display live price quotes. | Market and crypto headlines with source links. |
| Sports | ESPN; news search (same pipeline as above) | Today’s scores and schedules from ESPN’s public API, plus sports headlines from our news search for your followed leagues. | Scores/headlines with source links where applicable. |
| Daily lesson | Anthropic Claude; Wikipedia; web search | Claude generates a short lesson from your curriculum track. A separate automated accuracy reviewer checks it; recommended resources are confirmed via web search; further-reading links may point to Wikipedia. | An expandable “Fact-checked for accuracy” panel (optional; you can start it expanded in dashboard preferences). “Explore on Wikipedia” links when available. |
| Calendar agenda (Pro, optional) | Google Calendar API | Read-only access to your primary calendar for today’s events. We strip locations, meeting links, and descriptions before summarizing. | Event times and titles only. We do not store your event details after the briefing is built. |
What we verify and what we don’t
Daily lessons
- Every lesson goes through an automated accuracy reviewer (Anthropic Claude) before delivery.
- Lessons that fail review are not shown. We regenerate or hold them back rather than publish unreviewed content.
- Recommended books, videos, and podcasts are checked against web search where possible.
- This is automated review, not human editing. If something looks wrong, use the “Report an error” link on the lesson.
News
- Summaries, not copies. The AI never sees full articles. For each search result it gets only the headline, the outlet, the link, the date, and a short excerpt from the search provider, cut to about 300 characters (a sentence or two). It is instructed to write each summary in its own words. DailyRite is a starting point that points you to the original reporting, not a replacement for it.
- Labelled and linked. Every briefing and briefing email says at the top that its news stories are AI-written summaries, and every story links to the original reporting it was based on.
- Automated checks. A check compares each summary with the excerpts it was written from and flags wording copied from them word for word. Today the copying check records what it finds without blocking it; we are moving it to block flagged text before it reaches you. Another check flags a story that says someone was charged, sued, arrested, fired, or died when none of the search results it cites uses such a word; it also records for now rather than blocking. Checks for figures and source links that don’t appear in the search results are built but switched off for now; we will turn them on the same way, recording first and then blocking. These checks are automated, not human editing, and they can miss things.
- Our search providers. News search uses Serper, Tavily and Brave Search. Brave Search was switched off in October 2026 until Brave confirmed in writing that our use fits its terms; it has, and it is back on.
- We filter out known low-quality or banned sources.
- We run automated freshness checks on source dates and monitor quality with regular evals.
- We do not fact-check every news story the way we gate lessons. Summaries are AI-generated from third-party articles we do not control or endorse.
- We are not a newsroom. For anything important, read the linked source.
- Every story on the web briefing and in the browser extension has a “Report an error” link. Reports go straight to us; a story several readers flag the same day is pulled from the rest of that day's briefings. What we got wrong and what we did about it is logged publicly on the corrections page. Email readers can reply to the briefing email, and anyone can write to us at [email protected].
Weather, markets, and sports
Weather and sports scores use live or recently cached data from the providers above, so numbers can lag briefly during cache windows. Markets and crypto coverage is news from our search, not live prices, so it reflects when the stories were published.
Free vs Pro
- Pro users trigger fresh briefing generation on a fixed schedule (morning, and optionally afternoon). News searches run when your briefing is built.
- Free users read from a shared daily cache that Pro generation fills. Your briefing uses the same news pipeline output, but it may appear later in the morning if the cache for your topic mix has not warmed yet.
- Pro adds more topics, more stories per section, an optional Google Calendar agenda, state-level regional news, and other features listed on the homepage.
What we never do
- Give the AI full news articles, or republish articles. Summaries are written from short search excerpts and link to the original reporting.
- Collect or sell your browsing history.
- Put API keys or provider secrets in the browser extension.
- Use your data for advertising or cross-site tracking.
- Store raw Google Calendar event details after a briefing is generated.
- Claim to be a substitute for professional financial, legal, or medical advice (see Terms).
For publishers
If you publish news and want to tell us about a story we summarized, such as a copyright concern, a correction, or a request that we stop using your site, email [email protected] with the story’s headline, the date it appeared, and what you would like us to do. We can remove a story from briefings and exclude a site from future searches. For a formal copyright notice, follow Copyright and publisher notices in our Terms of Service.
Questions
Email [email protected] or read our Privacy Policy for collection, retention, and your rights.