Emerging computational multimedia tools and techniques promise powerful ways to organise, search and browse our ever-increasing multimedia contents by automating annotation and indexing, augmenting meta-data, understanding media contents, linking related pieces of information amongst them, and providing intriguing visualisation and exploration front-ends. Identifying real-world scenarios and designing interactive applications that leverage these developing multimedia technology is certainly an important research topic in itself but poses a number of challenges: the currently practiced methodologies and tools in the field of Human-Computer Interaction and Interaction Design seem to work better when the target users and usage requirements have been clearly identified and understood in advance whereas much of what emerging multimedia technology could offer is expected to create completely new user activities and usage that we are not aware of; immature multimedia tools currently being researched are not good enough to be the core engines of real-world applications today, making realistic user studies through deployment difficult; our future interaction platforms will be more than just desktop PC, Web, or mobile devices but many other forms of tangible, embedded, physical appliances which we expect the currently developing multimedia technology would be coupled with. In this paper, these challenges and the insights into how we could get over them are explored based on the author’s decade-long experience in designing novel interactive applications for multimedia technology.