← Use cases

E-commerce

Product descriptions from master data

From dry article data come appealing, consistent descriptions — in minutes.

A small language model (SLM) is enough

The problem

Hundreds or thousands of articles need descriptions. Manually it’s endless; bought-in copy sounds arbitrary and inconsistent.

The solution

From attributes (material, dimensions, properties) an AI generates descriptions in the brand voice — consistent, in several languages, with room for a final check.

How it works

  1. Article attributes are read in a structured way.
  2. The AI writes in the stored brand voice.
  3. Multilingual variants are produced in one go.
  4. Spot-checking instead of typing each one.

What changes

Time-to-shop drops drastically, the storefront feels more consistent — and SEO benefits from unique instead of copied text.

Honest take

A smaller model is usually enough for good descriptions — what matters is the quality of the master data, not the size of the AI. And: invented properties are off-limits, better concise than wrong.