This paper proposes an online planning approach to automate complex task execution by autonomous agents. We model the complex task as a Partially Observable Markov Decision Process (POMDP) to capture the inherent uncertainty. Then, using an adaptation of the Monte Carlo Tree Search (MCTS) algorithm, we generate a policy that governs how gradually the complex task should be accomplished. Our approach demonstrates a reliable and scalable procedure for automating a complex task using autonomous entities.