Methodology

How this site works

US Immigration Info is a small, automated newsroom for immigration policy. It is designed to keep visitors informed without the noise of a traditional news feed. Here is what happens under the hood.

1. Gathering public sources

A scheduled job watches a wide range of public sources — USCIS alerts, White House announcements, State Department notices, Federal Register entries, and reputable mainstream news outlets like NPR and the Associated Press. It checks for new content incrementally, so it does not reprocess the same announcement twice.

2. Deciding what is relevant

Each new item is run through a classification step that asks whether it is a genuine US immigration policy shift, court ruling, or enforcement change. If it is not relevant, it is set aside. This keeps the feed focused on things that could affect visa holders, applicants, and their families.

3. Summarizing in plain language

Relevant items are grouped into topics — such as “H-1B processing changes” or “Birthright citizenship proposals” — and summarized by AI. The summaries are written to be calm, self-contained, and easy to read. Jargon is explained or avoided.

4. Mapping possible visa impacts

For each topic, the system tries to answer: “If you are on this visa type, what might change?” The result is a short, per-visa impact grid. Because policies can be uncertain, impacts are described with words like “may,” “might,” or “could” rather than definitive promises.

5. Quality checks and scoring

Every summary is graded on dimensions like accuracy, clarity, completeness, tone, and usefulness. Low-scoring drafts are rewritten automatically. A severity score also helps visitors quickly see how much a topic might matter to visa holders.

6. Surfacing updates for visitors

The final output is a feed of topic cards and dedicated topic pages. Each topic page shows a chronological timeline of events, a shared summary, and a visa impact grid. Visitors can save a filter for their visa type to see the most relevant updates first.

Help me improve

This is a solo project, and I want to keep making it better. If you see a confusing summary, a missing source, or something that feels off, I would love to hear from you.

Submit an improvement suggestion