We propose SEASALT, an architecture based on the CoMES approach on developing collaborative multi-expert-systems using case-based reasoning and software agents technology. SEASALT is built on a modular structure and will allow implementing intelligent information systems for different kinds of application scenarios based on the architecture we are presenting. Introducing SEASALT furthers knowledge intensive services based on distributed knowledge sources. In our approach we integrate “knowledge work” in a community eliciting new information. Hence, in SEASALT agents act alongside with human beings on a community platform proving and receiving information.