Coalition operations are very likely based on a number of different groups of people, non-governmental organizations, institutions providing humani¬tarian aid and also army troops and official governmental initiatives. As a result, to manage any coalition operation an efficient knowledge sharing between multiple participating parties is required. The paper proposes an approach that combines knowledge logistics and information fusion at level two of situation assessment. The approach is based on such advanced information technologies as intelligent agents, ontology management, and constraint satisfaction/propagation. The main aim of the paper is to present the developed multi-agent architecture for intelligent support of coalition operations that would enable efficient real-time coalition operation management on-the-fly. A particular attention is given to the following three novel tasks: (i) ontology-driven knowledge representation via object-oriented constraint networks, (ii) processing of free text requests, and (iii) design of adaptive agents for on-the-fly problem solving. As an example of a coalition operation, a fictitious but illustrative case study of mobile hospital configuration from the area of health service logistics is described.