I'd like to talk about what I did here during the last months: reactive planning and ontologies applied to a kind of environmental domain (a wastewater treatment process). Planning is used, coupled with a case-based diagnosis, to decide about the actuation on the plant (WWTP). The ontology is in a preliminary phase of construction and at the moment is just used to describe formally the domain. In the future, back to Barcelona (December 19, not many days left), I'll try to integrate these things in a more complex supervisory system (implemented in Lisp and G2). Then, the objective of the introduction of the ontology will be the improvement of the diagnosis phase. But will this be realistic and efficient enough?