Skip to content
CONTACT

AI in the workshop: three practical uses (and an apprentice to double-check)

AI IN THE SHOP

Artificial intelligence gets talked about as either magic or the apocalypse, depending on the conference. For a small business it's simpler: it's an extremely fast, tireless and occasionally absent-minded collaborator, who needs to be guided and checked. Here are three jobs where it really pays off.

At one event I heard a speaker with rolled-up sleeves and a rally-stage voice proclaim that “AI is not a tool, it's a new way of breathing”. I breathe the same as before. But I do write product pages a lot faster.

1. Product pages and descriptions

Anyone with an online shop knows it: hundreds of products, descriptions copied from the supplier, identical to every competitor's. The same drill, with the same sentence, word for word, in different shops: less a catalogue than a choir. A language model, starting from the real technical data, writes original text, consistent in tone and already designed for search.

The rule is not to make things up: you supply the measurements, materials and dates, and the AI puts them into good shape. If you ask it the length of a table without telling it, it'll give you one anyway. Confident. Elegant. Wrong.

2. Translations

A website in four languages used to be a cost few could afford. Today the AI does the first translation and a person reviews it. On long jobs it's worth adding automatic checks:

  • have the numbers stayed the same?
  • do the links still work?
  • does the text fit in its box, or has a German word as long as a goods train smashed through the layout?

3. Replies and customer service

The same ten questions arrive every day: opening hours, shipping, returns, availability. An AI agent connected to the right information drafts the replies and the owner approves them with one click. After a few days the owner stops typing “yes, we're open on Saturdays too” in the tone of a judge reading a verdict, and gets his good mood back.

The limits, stated plainly

  • It's wrong with confidence: when it doesn't know, it sometimes invents, with the same face it wears when it does know. That's why every public text must be reread by a human.
  • The data is yours: be careful what you let it read, especially customers' personal data. Not everything in the management software should end up in a chat.
  • It doesn't replace craft: it multiplies it. And zero multiplied by anything is still zero: whoever knows the product gets better results.

How to start

Small: one boring process, a one-month trial, a measurement of the time saved. No visions, no “digital transformation of the soul”. A stopwatch and a spreadsheet.

Real revolutions, in a workshop, always start like this: with someone who finishes the job early and gets home in time for dinner.