Ruud,
Often, a "one-size-fits-all" approach leaves much to be desired. Usually, it will fit, but poorly. However, under the notion that a poorly fitting coat is better than no coat at all, I would encourage you to look for an application that will fit easily into your budget. Trying it on, so to speak, may convince you that you need a better one. Or, you may find that it is an entirely adequate application. Or, you may find that you really don't need the coat in the first place.
I am of the opinion - and this is only my opinion - that just because we CAN do something is not always enough to say that we SHOULD do something.
What does your maufacturing business need to know that you don't already have the means to determine?
What do others need to know about you that is not already available?
If there are answers to these questions; answers that make good economic and business sense, then perhaps a foray into AI might be beneficial. But if there are no answers, or the answers are not profitable, then I would avoid the step. But that's just me.
Regards,
Horace

Hi Ruud,
One the best known AI efforts is IBM's Watson (https://www.ibm.com/watson/whitepaper/solutions-guide/). It wasn't created with one specific industry problem to solve so it that sense it is generic but definitely not a startup in the traditional sense. Watson is now being implemented across several unrelated industries albeit with some customization.
I'm of the opinion that if a well executed AI product is broad enough but more importantly user friendly and reliable, people will find ways to apply it to their particular problem.
I agree with Horace (above) that at first blush it may be just a trial to determine feasibility that justifies further development/budget.
Thanks,
Phil