Question Answering (QA) systems promise to enhance both usability and accuracy when searching for knowledge. This thesis presents a prototype QA system built to leverage the extraction capabilities of a modern, context-aware search platform; Fast ESP. Questions in plain English are transformed to queries which target specific entities in the text that correspond with the identified answer types. A small set of unified patterns is demonstrated as adequate to classify a wide variety of syntactic constructs. For the purpose of verifying the answers, a semantic lexicon is compiled using an automated procedure. The whole solution is based on pattern matching and presents this as a viable alternative to deeper linguistic methods.