Developing AI planning-based applications is a challenging task, which requires expertise both in planning and software engineering. It demands proficiency in architecture design, integration of planning functionalities, utilization of service-oriented technologies, deployment solutions, etc. To uncover the challenges of building systems and applications involving AI planning, three separate software projects with varying scenarios are implemented and analyzed. Each of the projects is built with a different architecture: monolithic, Service-Oriented Architecture and microservices architecture, and integrates several planning tools. Subsequently, such findings of this practical research as: the importance of patterns in architecture design, the difficulties of planner selection, the utilization of service and deployment technologies are discussed and summed up. Additionally, this thesis contains an attempt to formulate guidelines for planning technology selection. They cover determining whether a planner is up to date, its availability, compatibility, and usability.