Two barriers that obstruct the implementation of AI in manufacturing are:

  1. People’s resistance to change
  2. Difficulty of connecting the many different data sources

I have spent the last two weeks interviewing several companies that address the second problem. There are really good solutions on the market that allow to integrate systems in weeks rather than months. In addition, terms like Unified Namespace (UNS) are trending on Google Search. New players are offering historians and MES-systems built on open-source technologies with open APIs - the future looks bright!

I’m hoping that in 2023 or 2024 we can leave the swamp of painful and labor-intensive point-to-point integrations between databases. With data integration solved, data engineers and scientists can focus their attention on building models ( whether statistical or simulation).