This paper presents SOMA, a multi-agent digital twin designed to support collaborative decision-making in helicopter engine maintenance. The system addresses challenges related to information loss and coordination among experts by integrating multi-agent modeling and stigmergic mechanisms. SOMA enables structured interactions between human and artificial agents, improving situational awareness and organizational learning. A prototype was developed and tested with maintenance professionals from Safran Helicopter Engine company. Results show that the system is relevant and that stigmergic indicators, i.e., visual signals based on past experiences, can influence decision-making. These findings suggest that SOMA can enhance maintenance processes by supporting effective collaboration and decision support.