We love the future and the technological possibilities of artificial intelligence. Therefore, we were particularly excited when we came across a company that had started an AI project with the aim of streamlining their production environment. AI and machine learning is one of our focus areas. Imagine our curiosity when we heard about a manufacturing company that wanted to use AI to create predictive maintenance. In other words, they wanted to reduce production downtime by being able to predict the need for technical maintenance at an early stage. It was a highly interesting approach and we were eager to learn more about their progress - but were told that they had already put the project on hold.
The trial backfired on its own staff
The project had resulted in an unnecessary number of alarms within the company and thus drastically increased the workload of the staff. Nor had the company seen any reduction in downtime in its production - which was the main purpose of the exercise. On the contrary, the project had instead created frustration among the employees, which in turn meant that the production staff now felt reluctant to digitize. It was clear that they now wanted to focus on other areas than managing another IT system. A real backfire.
So the initial energy around the AI project was gone. The need to reduce production downtime and increase production rates remained like a wet blanket over the business. This company is not alone in its experience. Many organizations face this kind of problem, which is why we have reflected on how a project with AI can be successful for you. A flying start to your tomorrow, plain and simple.
Smarter future by doing the right thing from the start
Before starting an AI or machine learning project, you need to think about how your data has been collected and over what period of time. The amount of data collected determines the expectations you can have for the outcome of a project. The more historical data, the more effective the AI solution. So, you need to do things in the right order, avoid tempting shortcuts and adjust your expectations according to the situation.
The company we met with had collected their data over two months, which means they needed to have different expectations of the results than if they had done it continuously over two years. So if you have access to relevant data that extends far back in time, it is a clear advantage. Note that we use the term relevant data. It is important to consider what type of data you really need to collect. You may also need to add more sensors or increase the frequency of your collection in order to obtain good material. If you are not sure how to do the right things in the right order, we can be your external support.
Also ensure that the project becomes something common for staff to rally around. Inform them early on that there may be initial problems and plan carefully when it is appropriate to start the project. For these pieces of the puzzle to fall into place Location , it is important to set aside plenty of time for careful planning.
Go mainstream and be future-proof in the process.
A good guideline for this type of project is to always use established software. This way, you will avoid dependencies on single individuals while future-proofing your established service.
Also, don't forget to keep the business close to the project during the journey. Even if it is internally run as an IT project, their joint knowledge and input is important for your success.
Go ahead - here's a smart summary:
- Think clearly before starting your project
- Make sure to collect data for a long time
- Skip all the tempting shortcuts
- Choose established software
Let your heart and brain decide when the timing is right for your AI project. Today. Tomorrow. And on and on.
And you, we are only a click or a call away if you need help moving forward.
