GS1 and UNSPSC: What the 28 February Classification Session Argued

A catalog that outgrew hand-coded GS1 GPC and UNSPSC codes can have a model propose the class. Someone on your side still has to know the standard. That is the argument of the 28 February 2025 session with AICA Data.

If you missed it, the recording is here:

At the microphone:

• Isak Marais, CEO of AICA Data
• Stiaan Maree, CTO of AICA Data
• Stephan Spijkers, Founder of Pimvendors

What the session argued:

• Since ChatGPT, the speakers said, classification can move faster and with less manual work.
• GS1 GPC and UNSPSC stay the codes that make a product comparable across systems and sales channels. The standard is the asset. The model is a way to apply it.
• They described agentic AI as reducing how often a person has to touch a code, with accuracy depending on the data underneath.
• They argued the approach costs less than coding by hand, and that it improves as the data improves.
• AICA's platform was presented as able to hold more than one classification framework.

Attendees were offered 20% off AICA Data API calls. That is a commercial offer from the session, not a measure of coding quality.

If you are deciding how to classify, start from the standard your customers already ask you to publish. Product classification sets GS1, ETIM, ECLASS and BMEcat next to each other so that choice is visible. A proposed code is worth storing when the attributes under it have an owner, which is the subject of product data quality and governance. Agents that answer from your catalog will follow the class they can read. AI-ready product data spells out what that record has to contain.