Fish farmers manage assets of considerable value on a daily basis. Many aspects of the daily operation are automated in some way, such as the feeding system. Sensory equipment steadily becomes cheaper and more ubiquitous, yielding data that can be used by automated systems and for post-processing (i.e. data mining) to discover hidden trends in the data. However, a lot of information is only known informally by the fish farmers themselves, through years of experience. Companies that can store this information and reuse it will have an advantage; even more so if high-level human expertise can be linked to low-level sensor data. This paper presents early developments of a system that stores this informal knowledge using case based-reasoning, combined with corresponding sensor data.