Can computation amplify our ability to achieve complex collective goals? Today’s techniques in areas such as crowdsourcing often fall short of this vision, in large part because their architecture is based on workflows, which are so inflexible that they can only be used for simple and modular goals. In this talk, I offer an alternative architecture based on computational organizational structures, and demonstrate that this approach enables groups to flexibly collaborate toward complex and open-ended goals such as product design, software engineering, and top-tier research. I will introduce techniques that fluidly assemble flash organizations and continuously adapt their efforts, evolve team structures and membership over time, and coordinate volunteers around the world in pursuing open-ended research. This research argues for a shift away from crowdsourcing as simple microtasks, wiki edits, or competitions, and toward computational systems that proactively aid groups in working together nimbly, reactively, and effectively toward complex goals.