Cloud applications in research and in the industry can be modelled with application deployment models to increase maintainability and to deploy them automatically. The modelling process requires domain specific knowledge about each component used inside the application, such as a Java application or an operating system, for example, Ubuntu. Furthermore, frequent update schedules of these components force developers to regularly update their deployment models to satisfy compliance rules. The process is error prone, as updating a single component in a cloud application deployment model can break the whole model and its deployment process because updated components may not be able to work properly with other components already in use. This thesis presents an approach that enables developers to automatically update deployment models by formulating constraints that ensure those models are deployable and define which component versions can work together. The approach described in the thesis uses artificial intelligence planning to automatically detect such problems and update components as needed. The updating process is realized by transforming deployment models into a Planning Domain Definition Language problem which is solved by a planner. The resulting plan is generated and describes how the model can be updated in multiple ways. The plan is then applied to the model and the updated model is presented to the cloud application developer. Constraints are defined which assert that the model is deployable and that each desired update of a component takes place. A prototype is presented which implements the approach shown and can be included in the modelling workflow of cloud application developers.