Viability theory studies the behavior of dynamical systems, with the aim of keeping them viable, or in other terms, keep their trajectories within desired constraints in the state space. Finding strategies to keep a dynamical system viable is already a challenge for optimization algorithms, but this task becomes even harder for multi-agent systems, where agents’ individual decision can influence the dynamics of the whole system. In this paper, we consider a multi-agent dynamic system and introduce an evolutionary approach for optimizing agents’ interaction behaviour, delimited by some a priori agreements, in the form of a set of commitments, with the objective of keeping the system viable. This approach is tested on the case-study of a collective project grouping several agritouristic activities on a shared place. Experimental results show that it is possible to find a set of agents’ commitments that can maintain system viability, supporting the proposed framework’s effectiveness.