Today’s challenge is to design complex systems that operate within evolving environments. Multi-agent systems and self-organisation are an answer to implement such systems. However few tools are provided to designers for helping them to build or improve the behaviour of agents. The aim of the work presented here is to reduce the workload of a MAS designer with a simulation tool that would give agents the ability to self-design their behaviour considering that (i) a behaviour can be expressed as a set of rules and (ii) agents interact in the dynamic environment they are situated in. This paper shows how simulation has improved the behaviour of simulated foraging ants.