We argue there is need of starting from existing works on computational fields to develop a sound engineering approach for developing self-organising coordination systems, from formal models to specification/programming languages, from patterns to platforms and applications. We presents early results in that direction.