Having now patented what it calls the breakthrough discovery of an algorithm conceived four years ago to solve major operational problems, AT&T has this month unveiled an $8.5m decision support system based on the work (CI No 994). The algorithm was concocted in 1984 by Indian born Bell Laboratories mathematician Narendra Karmarker, and caused a great deal of excitement within AT&T at the time. The new system, dubbed Korbx, will be marketed through a new AT&T Advanced Decision Support Systems division, which will be looking for custom from Fortune 100 companies and government customers with extremely difficult or previously unsolvable resource allocation problems. Included within the massive price tag is the hardware: an eight-processor parallel+vector FX80 minisuper bought in from Alliant Computer Systems, which costs below two million dollars according to Alliant’s vice president of marketing, Dave Rolme; and software technology, terminals and services from AT&T, which according to Aristides Fronistas, president of the new division, will help design customer systems in areas such as personnel planning, vendor selection and equipment scheduling. This includes expediting solutions to time-consuming linear programming problems with thousands of variables and constraints, said Fronistas. AT&T is already using a system internally at Kansas, where it is used in conjunction with an Amdahl mainframe running UTS Unix to optimise AT&T’s worldwide telecommunications network, and Fronistas says there are already has between one and 10 outside customers signed for Korbx systems.