About AI Radar

AI Radar is a human-edited publication about AI. It does not automatically scrape, republish, or expose a news stream. We choose questions worth understanding, examine several kinds of evidence, form an independent view, and publish a traceable article.

How an AI brief is made

Start with a useful question, not a trend: did this actually change capability, cost, workflow, or a reader's decision?
Check primary material first: official announcements, documentation, papers, repositories, pricing, and terms.
Look for independent evidence: trusted reporting, third-party evaluations, real-world tests, and critical feedback.
Synthesize before writing: separate confirmed facts, editorial judgement, and what remains unknown.
Make it actionable: explain who it is for, why it matters, whether it is worth trying now, and what to watch.

What we publish

Editorial briefs: multi-source analysis of important AI products, models, and industry changes.
AI basics: concepts and first steps explained for complete beginners.
Practical guides: methods readers can follow for real work.
Help me choose: tool comparisons based on needs, cost, and difficulty—not generic rankings.
Editor's notes: first-hand use, mistakes, and changes of mind.

Editorial principles

Publish less rather than manufacture a daily quota.
A vendor claim proves what the vendor said, not how well a product works.
A benchmark, one test, or community report has limits and is not a universal conclusion.
Important factual claims link to sources so readers can check them.
When evidence changes, update the existing article and explain why the judgement changed.

What happened to the old Radar

An earlier version of this project ran automated ingestion and event archives. That direction has ended. Historical data remains only in the private repository and git history as a project record; it no longer grows and no Raw Radar, event, archive, topic, source-health, or RSS pages are publicly generated.

Honest limitations

Human editing does not guarantee perfect conclusions. Evidence can be incomplete and products change quickly. We aim to state dates, unknowns, and interested parties clearly. When we find an error, we correct the article instead of hiding uncertainty behind confident language.