In a quest to breathe new life into public spaces and engage with its today’s increasingly (inter-)connected users, robotic architectural components seem to open up possibilities to give agency to building component. A new field of research has been involving multi-agent robotic systems in the architectural design and/or fabrication process. They require a new understand the built environment as an “adaptive ecology” composed of autonomous agents that interact with each other and their nearby environment (users and context). I am proposing the examination of a multi-agent system composed of unmanned aerial vehicles (UAVs, or drones) that can collaborate together to self-assemble into structures that are user-interactive and responsive to the local climate conditions. These structures are composed of a many of these ‘drone-bricks’ that form shelters that can grow and morph to accommodate users’ changing positions and movements through time. They are able to modulate environmental conditions such as wind and rain shelters, daylight penetration and the acoustic resonance of the space. This architecture is directly informed by the users’ activities, environmental conditions, and other information flows. It relies on principles of emergence, indeterminacy and generative design. This architecture has no blueprint but is self-organising; it is processual, indeterminate and continuously in formation, as opposed to the dominance of static, predetermined forms and functions in the organisation of the environment today. How to translate these architectural intents (performative and qualitative metrics) to a swarm of UAVs in an attempt to create a robotic architectural ‘species’ where autonomous agents collaborate to form temporary shelters that respond to local environmental conditions and human interactions? The research, simulation and prototyping rely on a behavioural and praxiological approach to architectural design that establishes a feedback loop between the algorithmically-encoded robotic capacities (behaviours), the material properties (morphogenetic processes) and sensing capabilities (neurological processes) of the agents and the emergent interactions and their self-organization in the environment in real-time. This project attempts to look at the ways these behaviours are digitally coded and physically expressed, as well as evaluate the overall agency granted to the system through the implementation of these behaviours with the overarching aim of developing new relationships between space and users through multi-agent robotic systems.