• 1 Post
  • 663 Comments
Joined 2 years ago
cake
Cake day: June 12th, 2023

help-circle
  • The original article and headline don’t mention AI, only the the headline of the English translation does. Shame to see them trying to clickbait international audiences. It’s our national broadcast, they don’t even make money off clicks. They usually have fairly good reporting too.

    I’m guessing the editor Andrew Whyte had something to do with this. He wasn’t credited in the original article.

    Also if they’re using AI for anything in the proposed system, it’s probably to detect which photos have drivers staring at phones. Simple image categorization. Not AI but a machine learning algorithm at least.


  • Poor translation of poor wording.

    In Estonian it’s common to say “today we do x” to refer to status quo. If you say “today we’ve decided to do x” like in the article, it can sadly be taken in two ways.

    Also that part refers to the lack of clarity on who would install average speed cameras IF we started using those. Nobody wants to pay for them basically.

    But the core of the article is an autonomous system that would be installed on top of police cruisers and send out tickets without officer intervention. Check speed, check insurance validity Check if driver is looking at phone or if seatbelt is undone. Same system in stationary cameras would work as average speed camera. The insurance thing is just an API call, the other stuff is ML (image categorization) so I guess you could call it AI, but AI is not mentioned anywhere in the Estonian article.








  • Yeah, the requirements should also be clear - or at least clear before any sort of implementation starts. Defining the requirements is a large part of what our consultants do and the more experienced ones know how to ask questions to get perspectives of people other than the “stars”. Takes months usually to get things to where us developers can get started on anything. We’ve built some hella cool shit for some customers but then you look at the git history and realize that it took the customer over a YEAR to go live. They must’ve easily invested six figures getting this ERP just right for their needs. Automatic imports from other software they use, lots of customizations, including some brand new in-erp apps. They’re loving it so far. But you don’t get this without considering a bunch of people’s needs.




  • It’s because you’re supposed to customize them, not use as-is. We’ve had a lot of happy customers. Some send us gifts! But for the first year or maybe even couple of years, you probably pay more to your partner for implementation, customizations and advice than to the ERP developer for licensing.

    ERPs aren’t for every company, different ERPs work best for different companies and different partners themselves have their own specializations. The one I work through (used to work for, but now I have my own company and just contract for them), does small to medium sized production companies. Think 5-200 employees usually. The ERP we work with is meant to cover every imaginable use case - which is why it doesn’t have enough depth. We add a bunch of stuff that isn’t there OOTB, sometimes remove things in default modules, etc.

    But first you NEED an ERP partner to make the most of it. At ours the CEO is also the biggest salesman. He’s not afraid to tell you if he doesn’t think it’s a good fit. A bad partner will still try to sell you and that’s going to end up in disappointment for everyone.