This talk overviews the Aggregate Programming (AP) paradigm for engineering decentralised systems. AP has been actively developed for over a decade. Its execution model (typically implemented within a runtime) assumes a local-first workflow, where each device repeatedly executes asynchronous sense–compute–interact rounds, relying only on local state and neighbourhood communication. The paradigm is formally grounded in AP calculi, such as the Field Calculus (FC) and its recent extension eXchange Calculus (XC). These are functional core languages designed to express computations over distributed data structures, called computational fields. Such fields map sets of devices to values, representing an integrated collection of decentralised inputs or outputs. This formal foundation enables rigorous reasoning about aggregate programs, including the derivation of guarantees and proofs. The main practical support for AP is given by multiple internal DSLs and toolchains: ScaFi (in Scala), FCPP (in C++), and Collektive (in Kotlin). Using these languages and libraries, programmers write a single, global program that specifies the overall behaviour of the collective as a composition of functions. This program can then be compiled down to local actions and deployed over a decentralised network of devices, either simulated or, through middleware support, made of physical systems.