How AI Pulse works
Sources, checks, limits and corrections for ThinkSuite's AI news.
Where stories come from
AI Pulse monitors 100+ public sources: company blogs and newsrooms, technology press, research feeds such as arXiv, GitHub releases and Hugging Face. Every article links to the original source it was based on.
How stories are chosen
Items from the last 48 hours are scored for importance by keyword rules, near-duplicates are merged, and URLs we have already covered are skipped. The highest-scoring items are written up.
Who writes and reviews
Articles are drafted with AI assistance from the linked source and published by an automated pipeline under the byline ThinkSuite AI Desk. They are not individually reviewed by a named human editor before publication. Translations (Hindi, Spanish, French and others) are machine-generated and are listed only when a complete translated body exists.
What the confidence label means
Before publishing, a language model assesses whether a story is plausible given its source, and we search Google News for headlines that corroborate it. "AI cross-checked" means that assessment ran and scored 75% or higher. It is a plausibility check on the story, not proof that every sentence, number or implication is correct. When the assessment is unavailable we fall back to the source's reputation alone and label the story "Source reviewed · not independently verified".
Static reference data
Company profiles, funding rounds, leader and heatmap scores in the Intelligence section are hand-compiled editorial snapshots. They are not live measurements and can lag behind company announcements. Always confirm figures against the company's own site or filings.
Corrections
If you spot an error, tell us through the contact page with the article URL and what is wrong. Corrected articles show an "Updated" date, and the sitemap date changes with them.