I've made a very simple website that tells relevant categories (and their codes) from really any input text description. Categories are of some of the most widely-used industry classification standards like NAICS and UNSPSC How does it work? On the backend, it creates multi-hundred dimension vectors (essentially numbers) and compares those vectors with a vector of each category in said standard. Closes are the most relevant semantically - closest to each other in meaning.

As a result, you get instantly what could have taken hours for a single product - searching a relevant category manually. Of course, it can make mistakes i.e. the top result is quite often not the best, but because it shows Top 5, Top 10 or even Top 100 categories if you wish, it is still much easier to find the correct category. And, perhaps more importantly, you can be certain that you didn't miss anything in a classification of 1000s of classes and subclasses.