Mobile devices have penetrated our daily lives and currently smartphones and tablets are everywhere. These devices are equipped with different sensors which enable them to capture multimedia information (e.g., several cameras and microphone). Therefore, today is easier than ever to count with a device which could capture information of interest in realtime almost anywhere. We present an approach to enable users to specify the kind of real-time multimedia information they are interested in and to obtain such information (in a continuous manner) from remote devices based on: 1) semantic techniques to handle the knowledge associated with different scenarios and requests, and 2) a network of mobile agents to process these requests continuously over different devices and communication networks.