The digital transition of companies has made the use of software for the management of their commercial, logistical and financial activities part of their daily life, and the list goes on and on. Each sector of the company has a module that centralises the resources it uses within this software. Ensuring the homogeneity of the information, guaranteeing that it is updated in real time, allowing better management of processes and facilitated coordination between departments, is the raison d'être of the ERP, and it does this successfully. Nevertheless, in this ocean of information that occupies more and more space every day, the user may find himself faced with a difficult choice as to how to allocate his working time. Data storage and availability is provided by the software, but the exploitation of the data is a task for the user and his subjectivity. Thus, misinterpretation is possible, leading the user to allocate working time to a task that did not deserve it to the detriment of another task that would have been more beneficial. However, time has this very unpleasant property: when it has passed, it is definitely passed. The recommendation system we are working on would make it possible, after defining the criteria according to which you want it to guide decisions, to propose to the user an allocation of his working time that satisfies the conditions initially specified, thus limiting the risks of inappropriate interpretation and generating a dead loss for the company.