Audiovisual concatenative synthesis is an analysis-driven granular technique using a corpus of multimedia data to sequence audio and video streams on a microtemporal level. This text outlines my development of this technique as a tool for multimedia composition, using my work, Replica, as a case study. The paper illustrates how the concepts of granular structure, gesture capture, and replication are integral not only to the software but to the architecture of the composition. In doing so, machine learning approaches to music and visual art are reviewed and related to my personal compositional practice. Additionally, I attempt to show how audiovisual concatenative synthesis provides a composer with strategies for shaping one's sense of time through disorienting audiovisual cues and tightly organized counterpoint between sound and image, stage and screen, and the real and virtual.