After 24 years of IT work in the insurance industry, Rick Laabs can clearly hear the advancing footsteps of a technological revolution.

That revolution is XML and Web services, says Laabs, vice president of product management American International Group Technologies (AIGT) of Livingston, N.J., the IT arm of $67.5 billion insurance company AIG. He says these technologies finally might erode the data and systems integration headaches that have historically plagued the process-driven insurance industry and its myriad disparate systems.

Insurers have such distinct information silos, experts say, that they can't easily share data up and down the supply chain. It's so bad that a single company often can't aggregate information on all the services it provides to a single customer.

"If there is going to be anything close to a silver bullet for the industry, it is XML," Laabs says. "XML is the first true opportunity that the insurance industry has to address the integration issue."

He says XML and Web services has the potential to deliver two Holy Grails: straight-through processing, which eliminates the need for manual intervention to manipulate data by integrating systems; and Single Entry Multiple Company Interface (SEMCI), which lets data be keyed in once and sent to multiple recipients regardless of platform.

IT strategy and research firm Celent estimates that 23% of the $6.3 billion the insurance industry spent last year on new projects went to fund integration work because of the lack of common data models and communications protocols. The firm says the insurance sector could save $250 million on integration costs per year by adopting an industry-specific brand of XML, called ACORD XML, for integrating legacy systems, internal applications, and data and transaction streams among agents, partners and carriers.

"XML represents a significant opportunity to streamline the integration process," says Matthew Josefowicz, manager of the insurance group for Celent. "It won't change the business of selling insurance, but it should relieve some of the IT burden."

ACORD XML is a set of nearly 600 standard messaging formats for executing transactions and exchanging policy information in three areas of insurance: property/casualty, life and re-insurance. The standards are being developed by the Association for Cooperative Operations Research and Development (ACORD), which has spent 30 years setting insurance industry standards, first with paper forms, then with electronic data interchange and now XML.

"Most transactions today are still EDI, but we are seeing the transition over to XML," says Lloyd Chumbley, director of standards for ACORD, which held its annual conference last month. EDI transactions, typically exchanged through an insurance hub called IVANS' Transformation Station, adhere to ACORD's AL3 standard. The AL3 standard is a batch model, whereas XML delivers real-time integration.

AIGT's Laabs still uses AL3, but it is being phased out and replaced by XML, which is now the strategic technology.

As Laabs helps oversee the adoption of XML, the focus today is on the company's WINS Digital system for property/casualty insurance processing. The software has been upgraded with a GUI, built with Java, Java 2 Platform Enterprise Edition, IBM's WebSphere and XML Web services. Agents use the thin-client WINS to access back-end mainframes and databases regardless of data formats.

"We're using XML to extend the system to extranets and portals," Laabs says. "We can do quotes, rates, renewals - the whole gamut of transactions." The previous interface was a green screen that enforced a rigid data entry and workflow process. XML lets AIGT be more flexible and change business rules and back-end systems without having to alter the GUI.

There are challenges for the industry, however, in that support for ACORD XML in vendor products and end-user deployments has yet to hit critical mass. This limits integration internally and externally. Also, highly customized insurance carrier systems will be difficult if not impossible to fully convert.