About ten years ago, Harvard Business Review proclaimed Data Science The Sexiest Job of the 21st Century . Ten years later most businesses, especially in manufacturing, did not reap the rewards that were promised by AI. What went wrong? In my experience, AI as a technology did not disappoint. From…

Little’s law is a theorem that equates the Work in Progress or WIP to the exit rate or throughput λ. L = λ x W Where: L is the total number of items in the system λ is the exit rate W is the wait time, i.e. the average time…

In my previous post I mentioned that developers creating standalone solutions is a bad thing because these projects are often abandoned. A standalone solutions is any piece of software that was implemented without a clear maintenance strategy in place. Usually, the developer of that software keeps it running on some…

On the job I interact with lots of different people at the plan that “need data”. Over the years I’ve started to learn that this requirement means different things for different people. Each requirement needs a specific solution and therefore a careful consideration of the technology. …

Today marks one month of daily emails. I started this mailing list on 1 January 2023. Thank you for reading my posts and emailing your questions and opinions! If you think the content is relevant to your company — why not forward this email to a colleague? One of the…

Manufacturing plants have a lot of machines requiring regular maintenance. There are procedures in place to protect both the equipment and the employees during scheduled and unscheduled repairs. A lot of care is put into following the guidelines correctly because worker safety is the absolute top priority. The same can’t…

If your company is just at the beginning of its digital transformation you may ask yourself what technologies are required to get started with data and machine learning. In my opinion not much has changed in 2023: the bread and butter are still SQL and Python. Most databases, especially in…

Manufacturing software systems don’t change often. I’ve worked with command line MES-systems that were older than me. Production is king: as long as the software that runs the process works, there is little desire to modernize it because we don’t want to break things. This stability is good because data…

Data pipelines found at a typical manufacturer continuously extract data from production systems and load them to a centralized place, for example a data lake. When the involves data coming from production, i.e. machine sensors, there is often already a system in place: OSISoft PI or another data historian. These…