The document presents an overview of probability collectives (PC), a distributed optimization approach for solving complex multi-agent systems. PC formulates the problem as agents (variables) that iteratively sample strategies and update their probability distributions to minimize a global objective function in a cooperative manner. The key characteristics of PC include exploiting concepts from game theory, statistical physics, and optimization. PC can handle continuous, discrete, and mixed variable problems in a scalable way and is robust to agent failures. Constraint handling techniques are developed to apply PC to constrained optimization problems.