Root cause analysis on high tech equipment can be cumbersome. Especially when it requires to manually go through thousands and thousands of PowerPoints with previous errors and solutions – to find a reoccurrence of the same issue in order to prevent the problems in future.
One of our customers is selling their equipment with a service contract. They monitor the products, fix them when they break, and do analysis after an error to find the root cause and define if an error occurs more than once.
This process of root cause analysis could take up to weeks, because of the above manual way of working and the complexity of the products. A time-consuming way of working that required broad domain knowledge to understand the reoccurring problems.
This unscalable method of root cause analysis was simply not an option anymore. It was asking for a data-driven solution.
To automate this process of root cause analysis, we first uploaded the entire database of PowerPoints with problems and solutions to an Azure cloud storage. Then, by means of automated pipelines, we ensured that all future files will be automatically uploaded to this same storage.
After bringing the database to the cloud, we were able to connect it to a search engine. This enables engineers to search the entire database by entering keywords. And to further speed up the process, we’ve built in a machine learning model. This model is trained on our customer’s historical data and generates smart suggestions based on the given keywords, so that the engineers can further specify their search without requiring broad domain knowledge.
Finally, we’ve used low code platform Mendix to build an intuitive user interface for the search engine, fully integrated with the systems at place.
With our solution, the time spent on root-cause analysis to identify reoccurring issues went from multiple days or sometimes even weeks, to seconds. This means that the engineers can now easily cope with the growing number of serviced products, and our customer saves 40 hours of work (on average) per problem that occurs; time that can now be spent on improving the product.