Introduction
Martina Krüger remembers the smell of the last boxes: cardboard, plastic film, a trace of perfume from the shelf beside the till. The boutique was closed, but the inventory lists and supplier emails remained.
Today Martina lives in Leipzig and advises small shops on using AI. She rarely arrives with big promises. Most of the time she begins with an inventory list, a coffee and the question that has been hurting the owner’s stomach for weeks.
Story of the Path into AI
After the insolvency Martina first wanted nothing more to do with numbers. Then, in a support programme for founders, she saw how simple analyses could make slow-moving stock visible. It hit her in an unpleasantly precise way. She would probably have noticed some bad purchases earlier if the data had not just lain around as piles of spreadsheets.
The technical vocabulary was an imposition at first. Classification, forecast, pipeline: it sounded as if she first had to learn a foreign language. She started with no-code tools, old stock lists and rough pattern checks. Once a seasonal product was labelled permanently popular because the test data came from the wrong week. Martina wrote the error in large letters: calendars are not customers.
Current Work
Martina now works with bookshops, toy shops and small concept stores. For one bookshop she set up an ordering system that only makes suggestions: which titles are moving slowly, which product group ties up capital, where unusual returns appear. The final decision stays in the shop, with people who know reading circles, regular customers and local events.
Some clients expect costs to fall immediately. Martina slows them down. AI can highlight patterns, but it cannot repair bad calculations or replace a difficult conversation with suppliers. What matters is that shops see earlier where goods are lying still.
Personal Advice
“Shame is a poor data source,” Martina says. She means her own bankruptcy as much as the situation of her clients. Anyone who only wants to prove that everything was right learns nothing. Anyone who reads spreadsheets as a memory of decisions may find the point where a new process can begin.
Key Facts
Age and place: 56, Leipzig.
Background: retail, insolvency, second professional start.
Entry into AI: warnings for stock levels and seasonal misjudgements.
Focus today: AI for small owner-managed shops.
Typical tools: spreadsheet analysis, no-code automation, simple forecasting models.
Werkstattnotiz
Martina keeps a list of products that the first model misread. One toy on it was supposed to gather dust and sold out after a rainy day. Since then she writes beside every recommendation: weather, shop window, conversation in the store. The machine sees the receipt, not the customer’s wet jacket.