One source of truth for product data
Product information has a way of living everywhere except one place: price lists in Excel, descriptions in an old catalog, labels rebuilt by hand in Canva, and the salesperson answering the same product question by phone for the tenth time this month. Every copy-paste between those places is a chance to ship the wrong number on a label or quote the wrong spec to a customer.
What was built
For a Swiss manufacturer, I built a product information management (PIM) database — one place where every product fact lives — and then connected everything that needs those facts:
A chat assistant for customers, prospects, and resellers. A web application where they ask product questions directly and get answers drawn from the database — not from memory, not from an outdated PDF.
The e-commerce shop, fed by integration. Update the product once, and the shop follows.
An MCP server for employees, so the team can query product data from the AI chat they already use — composition, dosage, compatibility, without opening another system.
A label and document creator (in final development): the design template is built once in Canva, imported, and populated automatically from the database fields. The manual copy-paste layer disappears.
What it changed
No number to report yet — honestly, it's too early to count. What changed is the mechanism: product data is typed once, and every label, shop page, and customer answer draws from the same source. The errors that come from retyping stop being possible.
FAQ
Aren't we too small for a "PIM"?
This isn't enterprise software with a six-month rollout. It's a database sized to your catalog, built around the fields you actually use — and it earns its keep the first time a label and the shop disagree.
Can the chat assistant invent answers?
It's built to answer only from the product database. If the answer isn't in the data, it says so rather than improvising — that rule is part of the build, not a promise.
We already have a shop / an ERP. Does this replace them?
No. The PIM feeds them. Your existing systems become consumers of one clean source instead of each holding their own slightly different copy.
Who maintains the product data?
Your team — in one place instead of five. Training them on it is part of delivery, not an option.
If your product data lives in more than two places, one of them is wrong right now.
Book a call