An intelligence layer reads coastlines from space and compounds on its results

Coastal Assembly’s platform reads satellite imagery, decomposes years of shoreline history into predictive models and picks the placements that will redirect sediment; coordinated swarms of autonomous surface vessels are to carry out deployment at scale, and each new site returns ground-truth data to the models (company site). The structures themselves — porous forms that let calm-weather currents deposit sand while dissipating storm energy — are recorded separately; this item is the learning layer above them, which Coastal Assembly, built on nine years of Skylar Tibbits’s Self-Assembly Lab research at MIT, says gets better at growing land the more land it grows. In the recursive-self-improvement trajectory it is the loop’s first appearance in geography, a data flywheel in the ocean alongside the physical recursion of humanoids assembling their own hands, and it sets up the same issue’s argument that land was the bull case for scarcity. The author discloses a financial interest in the company.