Existing translator’s workstations still provide very little support for the central and defining aspects of the translator’s task. The concept of bi-text is shown to open altogether new possibilities for the workstation. A bi-text is a pair of texts ( a source and its translation) linked by a set of explicit translational correspondences. If translations are organized bi-textually, the production of a translation service can become a veritable corporate memory, whose riche scan be tapped by tools like bilingual concordancing systems. Bi-textual representations also provide the foundation for translation critiquing tools, capable for example of detecting the presence of deceptive cognates in a translation. Algorithms have already been developed that can automatically produce a (partial) bi-textual representation for a text and its translation. We are currently in the process of developing prototype concordancing and translation critiquing tools.