Our editorial process

Step 1 — Discovery

We use the YouTube Data API v3 to search for trending content across categories including side hustles, online courses, viral products, investment apps, giveaways, and more. Searches are run on a scheduled basis. We track videos from a curated list of channels known to cover these categories — ranging from personal finance creators and consumer-protection accounts to independent product reviewers.

Each video is scored on recency, view velocity, engagement, channel authority, and topical relevance. Only videos above a minimum threshold are processed further.

Step 2 — AI summarisation

A large language model (Anthropic Claude) reads the video metadata, title, and description to produce:

• A short summary (2–3 sentences)
• A long summary (200–400 words)
• Key takeaways
• Warnings and caveats
• Extracted claims, each rated for evidence and risk level

Step 3 — Human review (for high-risk topics)

Videos covering investment apps, side hustles, online courses, giveaways, and remote job offers do not publish automatically. They enter a review queue visible to our editorial team. A human reviewer checks the AI output for accuracy, flags overreaching claims, and either approves or rejects the summary before it goes live.

All other topics are reviewed on a sampling basis.

Step 4 — Publication

Approved summaries are published with the official YouTube embed, evidence ratings, takeaways, warnings, and the mandatory disclaimer. We never download, modify, or rehost YouTube videos.

Corrections policy

If you find a factual error in a summary or an incorrect evidence rating, please contact us. We will review and correct the content as quickly as possible. We do not silently edit summaries — significant corrections are noted on the page.