According to Richard Cross, principal knowledge management officer at Xerox UK, Eureka reflects a "triumph of practice over process". This, he explains, means that any KM system that is effective has to honour work practices - not the formalised processes that are put in place but the way people actually work.
In the case of Eureka, Xerox wanted a KM system for engineers charged with maintaining printers and copiers to provide them with online access to manuals, best practice and useful ideas.
Cross says, "It was vital that the system incorporated the less structured ways that our engineers work. Much of the information sharing comes via breakfast meetings and brief chats by the coffee machine - their culture is about fixing problems among peers. Incorporating this way of working into the system is as important as including the more formal type of training programmes."
To ensure that this was the case, the systems design team even brought in the expertise of a sociologist to study the work practices of Xerox's staff.
"We asked managers what processes they had put in place," says Cross. "We looked at how work was prioritised, and analysed what kind of contingencies were not covered by the processes."
Cross explains the need to study how people work rather than simply mapping electronic processes into a KM system. "The print industry has had a lot of computerisation installed in the past few decades but we found the MIS system for the printing machines switched off on some sites because each print job has to be given a job number which follows the job through from start to finish. Sometimes workers had to wait for a job number to be allocated which meant that the computer system was preventing them from getting on with the job," he says.
"To get around this they simply switched off the machine and filled in the job number later on when the paperwork caught up.
"Our analysis identified this as a hindrance - so the Eureka system ensures that workflow systems are flexible so that it is possible to incorporate local ways of working into a general package."
Cross says that by the same token it will be important to monitor newly implemented KM systems so that improvements can be made in view of how they are used.
But as a KM system grows and takes in more data will users not continue to suffer from information overload, wasting valuable work time sifting through this information in search of something relevant and useful? Cross is convinced that "digi-glut", as he calls it, can be avoided by using a "knowledge pump" which enables users to rate the value of the information as it is used and accessed.
He explains, "The higher that information is rated - the more that information is pushed to the top and the higher it will be positioned in future searches about the same inquiry. We are also working on an 'ask once' meta-search engine. Incorporating these kind of rating models is one of next big challenges."