Supply-chain-management programs are structured sort of like flow charts, following a make-and-sell model of supply and demand. Software that IBM Labs is building works more like bees in a hive, with lots of autonomous agents going out into the world collecting data. The result, says Grace Lin, a senior manager at IBM's T.J. Watson Research Center, is a system that can more easily consider new sources of information. These "sense and respond" systems make their own decisions based upon the variables at hand and aren't strictly confined to a set of rules.
IBM's effort essentially gives the program artificial intelligence, so it compares current business conditions to historical ones and forecasts what's likely to happen next. "It doesn't just react, but anticipates," Lin says. And based on what actually does occur, the program can compare its forecast against reality, learning if it made a mistake. Lin says she expects a finished prototype later this year and a commercial version within five years.