Modern manufacturing and production systems have growing demands in energy and cost savings, which can be ensured using more advanced control algorithms at the regulatory-level industrial control loops. However, developing such algorithms requires case-dependent approach that involves complex mathematics and expertise in various fields of technology and engineering. Gathering all the needed experts to conduct the control system engineering cycle is nearly impossible for organizational and economic reasons. Thus, in this work, it is proposed to employ an agent-based approach, which is substantially different than the conventional engineering cycles for developing the static control system. The idea is to split the entire control design procedure into smaller tasks of developing the modules (i.e. agents), which encapsulate the expert knowledge (e.g. on sensor failure detection, input signal modelling and estimation). It enables clear division of the competences between the experts and allows for dynamic inclusion of the expert knowledge into the newly designed or already existing system. Therefore, the expertise may be distributed both in location and in time, which stands in contradiction to the static approach based on sequential engineering cycles.