Graded cueing is a hierarchy of prompts that can be used by assistive technologies to support individuals with cognitive impairments during daily activities. For example, people with traumatic brain injury or dementia may need assistance with cooking, and shopping. Mixed reality, a blend of virtual elements with the real world, is a technology that can offer benefits in this context. To be relevant, graded mixed reality cues need to be tailored to the specific needs of each individual. Behavior trees, a structured approach to transitioning from one task to another in an autonomous agent, might be useful in this context. In this paper, we present a mixed reality model that utilizes behavior trees to deliver graded cues that support the completion of activities. An evaluation was conducted with 11 participants to investigate the model’s ability to guide them autonomously through an activity and assess their perception of the adequacy of the cues provided. The results suggest that the proposed model is able to autonomously guide participants and provide appropriate cues.