Hardware constraints are preventing Kiwibank from upgrading to the Windows Vista operating system, according to the bank’s enterprise architecture manager, Bohdan Szymanik.
Speaking at Microsoft’s Tech Ed, Szymanik said that most of Kiwibank’s desktop PCs, which had been purchased rather than leased, were still running Windows 2000. While an upgrade of the oldest PCs would soon allow Kiwibank to standardise on Windows XP, the bank was not yet ready for Vista.
“I would quite like to go Vista as much as possible… but we still have a large number of PCs that wouldn’t run Vista. It’s quite close, but at the moment it’s XP,” said Szymanik.
Describing Kiwibank as “basically a Microsoft shop,” Szymanik said that Microsoft software in general had been a “great enabler” which had helped the bank overcome the obstacles to integrating its systems. Kiwibank was now making widespread use of the Workflow features provided by applications such as Sharepoint, and this was allowing the company to integrate with external providers such as credit card issuers at the services level. Sharepoint was presently being used in the Kiwibank’s IT department but would be deployed across the company in the next few months.
Szymanik said that years of development by Microsoft in areas such as data management, programmable runtime environments, workflow and the desktop interface, were “finally coming together” which was allowing the bank to automate many of its processes.
“Part of me wishes that these workflow functions were around four or five years ago but we needed these other components first,” he said.
Szymanik said he was currently experimenting with the “hidden” data mining capabilities of Microsoft’s SQL Server which could allow automatic decision-making based on statistics rather than rules.
Szymanik explained that rules-based systems, to approve or decline loan applications for example, tended to become very complicated because of the need to manage numerous exceptions. Decision-making based on past statistics, backed up by human confirmation in the more marginal cases, could provide a better route to automation.
“There’s no reason why we can’t make these decisions based on historical decisions, that is data mining. The functionality is all there in SQL Server but a lot of people don’t know it’s there.”