What he grabbed was an XML specification known as FinXML, based on work done in conjunction with the Open Financial Exchange, a specification for the electronic exchange of data among financial-services companies. Astor's not betting that FinXML will become the only way to use XML--but it was one Astor knew could work for his company and his partners.

He's not alone. IT executives are realizing that XML is more accessible, less expensive, and easier to develop than most intercompany communication technologies, in particular EDI. For example, Dun & Bradstreet's Global Access system provides customers with an easy way to download business information, such as credit reports, marketing lists, or purchasing sources, from the company's global network of offices. The use of XML technology makes it easier for these companies to access such information and integrate it into their own applications, Astor says.

Still, EDI is a standard; XML is not. That's the basic dilemma for most IT managers interested in using XML. Though it would be comforting to have one monolithic XML standard to reference, that's not possible, industry experts say. The best IT managers can do is to use a version of XML most relevant to their data exchange needs.

Laura Walker, executive director of Oasis, an international nonprofit consortium in Boston dedicated to the development of product-independent data exchange, says there are thousands of initiatives under way to develop XML specifications, and most industries are starting to develop their own XML vocabularies and formats. Walker says IT managers in most industries shouldn't expect to see a formalized XML specification for a year to 18 months.

Michael Maziarka, a director at consulting firm Cap Ventures, is more blunt. Maziarka says IT managers should never expect a universal XML.

That's OK, Walker says--IT managers shouldn't fear the proliferation of XML specifications. "This is the way it's supposed to be," she says. "It's not a bad thing for groups to develop industry-specific or horizontal application-specific XML versions."

Benoit Lheureux, research director at Gartner Group, divides XML standards into two categories. The "foundation standards" are specifications for the basic nature of the language, such as how it performs document linking, document security, and other fundamental functions of the language. The World Wide Web Consortium, which oversees development for most Internet technology standards, has already developed version 1.0 of the XML specifications.

Lheureux says there's little debate about these specifications. The area of concern for XML is what Lheureux calls "agreement standards," which involve the formatting of XML, including the vocabulary used and the definition of that vocabulary. This information is contained in so-called data-type definitions, or what in the future will likely be part of XML schemas.

Any two companies exchanging XML must agree on a common data-type definition so each application knows how to interpret the XML data it receives. The problem is that these data type definitions need to be specific to any type of data exchange. A car manufacturer, for example, will want to develop an XML vocabulary specific to its business, just as a computer manufacturer will have specific terms for information it exchanges with partners. "It's a rich area of debate," Lheureux says. "There's some controversy at the agreement level."

Fadi Chehade, CEO of RosettaNet, a consortium that has developed an XML specification for the IT supply-chain industry, compares XML to an alphabet. It requires a grammar and a vocabulary to become a viable language. The danger, Chehade says, comes if too many groups make up their own XML specifications, splintering XML into a bewildering array of competing dialects. "If we don't jump on this issue collectively, we will end up with a Tower of Babel," Chehade says.

continued...page 2

Photo of Astor by Edward Santalone