Nowadays, oil companies are trying to extract more and more information from already existing data. However, it is still done by operational or engineering personnel, and drilling lessons learned are not an exception. It may lead to the loss of some small lessons or even to the mistakes in data interpretation. That is why companies are attempting to digitize everything to eliminate human influence on most of the processes. The purpose of this work is to develop a concept for automatic lessons learned extraction from gathered data and giving recommendations to the drilling engineer based on them during a well design process. To achieve this goal, the concept of using machine learning algorithms to digitize the well design process was developed. Status quo of lessons learned capturing and analysis at OMV was investigated, and its downsides were found. Discovered problems may be solved by the implementation of a recently developed knowledge graph database because it has many benefits in comparison with standard databases. For instance, a comprehensive information search which returns not only documents that match sent query but also important information related to that document, therefore, a user does not have to look through the whole document to retrieve required data. Additionally, the knowledge graph database is capable of returning information from other documents which have a specific value of similarity with the uploaded one. Additionally, this work provides some prototypes of machine learning models for wells clustering based on its trajectories, lithologies and activities. Also, a simple OCR algorithm was coded to digitize PDF documents. The developed approach showed the applicability of machine learning algorithms to automatically capture lessons learned from already existing data and smartly apply them during the design of a new well.