Agent augmented reality is proposed in this paper as a new research area that uses agent technologies for the augmentation of our real world environment by actively integrating information worlds. We introduce a special agent called a real world agent. A real world agent is a software agent that can support its user in performing tasks in real world environments, such as place-to-place location guidance, instruction in physical tasks, and augmentation of human knowledge related to the physical environment. In order for the agent to achieve these tasks, the agent should be aware of the user’s real world situation. The detection of real world situations is performed through the integration of various methods, including location-awareness using a global positioning system, object recognition through machine-recognizable IDs (barcodes, infrared rays, etc. ), and the processing of visual/spoken inputs. IDs of static objects (e. g., doors, ceilings, walls, etc. ) can also give clues to the location-awareness system. When a real world agent correctly deduces the real world situation and the intentions of its user, it can access information worlds (e. g., the internet) in the same manner as other software agents. In this way, the real world and information worlds can be dynamically integrated. Real world agents can detect real world situations, interact with humans, and perform tasks in information space.