a methodology is presented that integrates diverse artificial intelligence techniques in order to build a forecast system that determines the convenience of applying a solution with tensoactive properties to an oil well, so as to increase oil production. the methodology begins by processing the data obtained from an experiment consisting in the injection of tensoactive products into a group of wells in an oil field. different exploratory techniques were used, such as pattern recognition, selection of variables, and methods for the automatic generation of hypotheses. the information obtained through such processing was modeled in a knowledgebase which, together with the inference machinery of a language named haries, permitted the construction of a system capable of decision-making in relation to the injection of tensoactive substances and of suggesting the most appropriate technology to be used in each instance. the system was applied to different wells, obtaining in satisfactory results in every case.