← All cases
Industrial parts · e-commerce

A thousand listings prepared without writing a thousand descriptions

Uploading products to Conrad or Mercateo means descriptions, categories and eCl@ss codes for every item. A dozen is manageable by hand; a thousand is not.

Client
Spare parts & industrial equipment · Germany · $15M revenue · 100 staff
Scope
CSV in → clean → enrich → marketplace export
Targets
Conrad, Mercateo and similar
Role
Solo — design, build, hand-over
Marketplace listing preparation platform
Parsed CSVs cleaned, enriched and turned into a marketplace-ready export.
Batch size
1,000+
items per upload
Generated
3
fields: description, category, eCl@ss
Manual step
Export
the manager only triggers it
§ 01 / Problem

Enrichment that does not scale past a dozen products

Marketplaces will not accept raw parsed data — every item needs a description and a classification.

To upload products to a marketplace the data has to be prepared first: generate a description, assign a category, add the eCl@ss code. For a dozen products a person can simply do it.

At volumes above a thousand items the same work takes an enormous amount of time, and the parsed source files are rarely clean to begin with — duplicates, formatting noise and missing fields all have to be dealt with first.

§ 02 / Solution

A platform that cleans first, then enriches

Parsed CSVs go in; a marketplace-ready file comes out.

01

Upload the CSVs

input

The manager uploads the parsed CSV files into the platform.

02

Clean the data

normalise

The agent validates and cleans the data — formatting is normalised and duplicates removed before anything is generated.

03

Generate and classify

enrich

Pipelines generate descriptions and determine category, eCl@ss and the other required fields from the file contents.

04

Export

output

The manager generates the final export, ready for the marketplace listing.

§ Built with
PythonLLM generationCSV pipelinesData cleaningWeb UI