Natural architectures such as bone, bamboo, and nacre have long been a source of inspiration as engineers seek to design architected materials with robust failure characteristics. Much of the literature on the failure of cellular solids has focused on regular, periodic lattices. However, natural structures often exhibit a significant degree of spatial variation and/or stochasticity of structure. These qualitative differences arise inevitably from the differences of the manufacturing processes used in the two cases: Synthetic architected materials are typically made using deterministic processes, whereas natural materials are generated by decentralized processes involving many simple agents (e.g., many bees generating a honeycomb, or many cells generating a tissue). In this work, we explore the idea of using simulated ‘swarms’ of agents to generate new designs of architected materials. The agents themselves are unaware of any global plan or design for the desired structure. Instead, similar to ants, they are only aware of their immediate surroundings, making decisions based purely on their sensory input and simple rules (e.g., ‘extrude material when another agent is encountered’ or ‘move away from higher temperatures’). The generated structures are thereby the result of the collective action of a swarm of simulated agents rather than a predetermined design. To accomplish this, we build a simulation platform that can realize : 1) intensive virtual design, 2) high fidelity finite element modeling, 3) fast-paced prototyping and testing, and 4) optimization via local sensory input. We use this platform to discover ‘rule-structure-property’ relationships: i.e., correlation between the rules that the agents obey and the structures that they generate, and the relationship between the generated structures and their mechanical properties. In this work, we will particular focus on the energy dissipation of these architected materials relative to traditional regular lattices.