Every AIBUY report opens with a verification card — test set, window, what was automated and what a human checked — and links to its prompt log. Reports are free to read and quote with attribution.
40 purchase-intent prompts × 8 assistants. Which brands the assistants actually recommend, how they describe them, and where they get the specs wrong.
Test window: Sep 5–12, 2026 · Panel: ChatGPT, Claude, Gemini, Perplexity + CN panel for the cross-market section · Status: pilot dry run, final tables with full log publishing soon
What moved this week in AI shopping answers: new models, notable shifts, one dataset. Fully automated pipeline, human-flagged items marked. Ships in the newsletter.
One category per month, full method. Merchants can request the next category — requests with the most votes go first.
Quarterly state of AI recommendations in one market, plus the cross-market comparison run with both panels simultaneously. Human-written conclusions.
One email a week. The brief is generated from the same pipeline as our reports — and the verification card rides along in every issue.
Weekly · English · unsubscribe anytime. 中文版请订阅 aibuy.org.cn。