With the rise of chatbots and voice interaction, conversations with machines become more human-like. But when people communicate, in addition to their voice they use everything they have, including gestures, emotions and facial expressions. The support of such multimodal communication will be an important requirement for human-computer interaction in the future. In this talk we outline concepts and a basic architecture of multimodal systems. We illustrate advantages of multimodal interfaces through several technological demonstrators, which have been developed in research and industrial projects of the DFKI.