Caeliai
Independent research on AI search, taste, and why.
-
ArchiveAgent
A small fine-tuned model beat much larger ones at reading runway photos.
Brand, season and year all right. 287 new photos, Oct 2026. “Before training” is Kimi K2.6, the model it started from. ResultsLimitsarchiveagent.si -
Search audits
When AI names a brand, almost half the time there’s no working link to it.
brand’s page 36.8% · someone else 17.6% · no link 45.5% 380 brand links from ChatGPT and Gemini, May 2026. PaperField studyMethod -
Home GPU test
On one home graphics card, the best free image model found the right brand 3 times in 4.
86,874 runway photos, 6,480 checks, one RTX 5060. Results -
Shopping Answers
ChatGPT showed a product carousel in 6 of every 10 shopping answers.
carousel 59.7% · text with links 37.3% · other 3.0% 274 of 459 saved answers, 463 brands named. Saved Mar to Aug 2026. ExploreUse with an agent
AI search
How does AI rank what it shows?
Shopping
- DatasetShopping Answers v1: 459 questions and ChatGPT’s answers
- PaperField Study 01: Where ChatGPT routes jewelry recommendations
- BriefDoes ChatGPT show the official Shopify product card?
- BriefWhich AI shopping surfaces to measure first
- PaperGenesis 03: AI shopping visibility and the buy path problem
Taste
What does AI think looks good?
Why
Why does AI do what it does?