How AI shopping systems decide what gets recommended.

We study how models behave, where they drift, and what brands need to change to be accurately represented in those machine-generated journeys.

AI shopping is a recommendation surface with its own taste.

Observation 01

AI shopping is becoming a recommendation surface with its own taste, shortcuts, and failure modes.

Observation 02

Product data quality now affects not just feeds and search, but model-generated answers.

Observation 03

Model visibility is measurable.

Research built for the AI recommendation era.

The Genesis studies map how models perceive fashion, which aesthetics they privilege, and how those preferences leak into what they recommend back to consumers.

Genesis 01 Vision benchmark

Testing AI vision on high-fashion nuance.

A foundational benchmark comparing CLIP, SigLIP, and DINOv2 across runway images, uncertainty detection, and collection coherence.

Genesis 02 Model taste map

Tracing what LLMs repeatedly choose in fashion.

A multi-model look at aesthetic consensus, canonical fashion objects, and how preference clusters emerge across current AI systems.

Research becomes action when your products need to show up.

The second half of Caeliai is applied: audits, feed correction, AI visibility strategy, and instrumentation for understanding which systems are naming you and how.

Audit

Check how major AI systems represent your products, pricing, and links.

Repair

Fix weak product data before it compounds into agent mistakes.

Measure

Tie AI-originated discovery back to commercial outcomes instead of treating recommendation surfaces like black boxes.