How we work

Every claim on a story page can be checked.

This is exactly how a story reaches you: what we read, how we decide what matters, and which parts are written by AI.

What we read

We read the public feeds of 21 sources every two hours, from 05:30 to 23:30 India time. From each report we take what the outlet publishes in its feed — the headline, the time, and the description or text the feed includes. We check each site's robots.txt first, and we never fetch the articles themselves.

Global

India

One event, one story

When several outlets report the same event within three days of each other, we group them into a single story. The story lists every outlet that carried it, links to each report, and shows how many independent sources there are.

A story with one source says so, and is marked to be treated with care. A single-source story can never be labelled breaking.

How significance is scored

Every story gets a significance score from 0 to 100. It is a fixed formula, shown on every story, not a judgement made story by story:

The formula

The levels

Where AI is used

The written parts of a story — what happened, why it matters, what changes, the India angle, the headline and the topic tags — are written by an AI model (Anthropic’s Claude) from the reports the story links to. The impact input to the score is also assessed by that model. The daily brief on the front page is written the same way.

When a story couldn’t be written by the model, it is built from the reports by simple rules instead, and the story page says so.

AI can get details wrong. Every story links to the original reporting, and you should read it before relying on any specific figure, quote or date.

The India angle

Global stories get an India angle only when there genuinely is one. Where there isn’t, the story says “No significant India angle” rather than inventing one.

What we never do