In this paper, we seek to define an experimental framework for the application of a new industrialization method for penetration testing. This work- in-progress research is placed in a particular business context: that of a company with an extensive and decentralized information system. The objective of this research is to give companies the tools to develop a penetration test task force capable of testing any system in a fully automated way and to form proper communication channel and support for risk assessment reporting. It is based on the use of artificial intelligence to make the penetration test autonomous. This research considers the conduct of penetration tests both through their technical issues and through the managerial issues specific to a decentralized information system.