The Rule Responder project (responder.ruleml.org) extends the Semantic Web towards a Pragmatic Web infrastructure for collaborative human-computer networks. These allow semi-automated agents - with their individual (semantic and pragmatic) contexts, decisions and actions - to form corporate, not-for-profit, educational, or other virtual teams or virtual organizations. The project develops an effective methodology and an efficient infrastructure to interchange and reuse knowledge (ontologies and rules). Such knowledge plays an important role for (semi-automatically and contextually) transforming data, deriving new conclusions and decisions from existing knowledge, and acting according to changed situations or occurred (complex) events. Ultimately, this might put AI theories on distributed multiagent systems into larger-scale practice and might form the basis for highly flexible and adaptive Web-based serviceoriented/ service-component architectures (SOAs/SCAs).