This study, based on a Master Thesis discussed at the Politecnico di Milano, takes a closer look at the Unreal Engine (UE4) as a tool for creating a real-time design environment using Artificial Intelligence (AI) technologies to represent users’ flows in the space, which can be adopted to carry out design strategies and to evaluate design options. For this purpose, various simulations have been developed, either considering the pedestrian planar flows interlinked with the form of space, either parametrically generating the spaces on the bases of the number of users. The first series of tests is then based on pre-assigned spatial contexts. In order to test the AI programmed sets, different situations were figured out and modeled in advance. Given the assigned space, a series of points has been subsequently assigned, working as ‘attractors’ according to possible users’ interest locations, and a virtual robot (NPC) has been placed to explore the various possible paths, based on a random sequences of movements towards the assigned attractor-points. In order to be able to graphically represent the visual simulations, the NPC has been equipped with a tracing system, allowing to reproduce in real-time the flows as graphic diagrams in the space. The AI system was also tested in a 3D spatial context characterized by differences in heights, such as inclined corridors, vertical lifts, and so on. Other series of tests has been carried out considering the inverse process, implementing a generative system able to create new spaces, such as rooms and paths, according to the needs emerged from the users. Therefore, a generative algorithm was set, able to update the geometry of space according to the number of people supposed to ‘need space’: in other words, space expanded according to the number of users. The thesis, focused on the use of UE4 to realize an AI system helpful to represent and control – visually and parametrically – pedestrian flows in a three-dimensional environment, either pre-existing or generated according to specific inputs. More generally, linking analysis and project, especially in more complex scenarios, it can serve as a tool for mapping and analyzing architectural contexts, as well as for implementing, verifying, and comparing design choices, that is, efficiently sustaining the whole chain of the architectural design process. Finally a simulation was made by superimposing on a video shot in Piazza Leonardo – Politecnico di Milano, the representation of the movement of the silhouette obtained through specific software of video tracking and video editing.