When Machines Chat

XML is about to spark an efficiency revolution

Life for manufacturers is about to get a lot easier, thanks to an emerging Internet technology dubbed XML. That moniker is short for extensible markup language, but don't confuse XML with cryptic programming languages like C and Pascal. Rather, think of XML as a combination Rosetta stone and Oxford Unabridged Dictionary of the Web--a tool that will smash the language barriers now segregating different breeds of computer, different business-process software, and different database formats. For manufacturing, XML will transform Web sites from online parts catalogs into doorways for collaborating on everything from product design and prototyping to optimized production and intricately coordinated supply chains.

XML's sweeping implications have Bill Gates, Steve Jobs, and many other CEOs hopping with excitement--along with manufacturing engineers and project managers like Maytag Corp.'s Bryan Patty. In the past, shipping the specs of a new dishwasher or other appliance to retailers such as Sears, Roebuck & Co. was a "Herculean process," Patty recalls. A team of programmers would download all the information from a mainframe, then retype the information into an Excel spreadsheet. The manual-translation process was both tedious and ripe for error. "That's not where you want to be in terms of efficiency," Patty says.

DATA DISTINCTIONS. With XML, Maytag now simply labels, or "tags," each chunk of data. Model number, price, and all the other numbers and words are marked to identify what they are. Then Maytag's mainframe zips out the specs to retailers and trading partners. Their XML-savvy computers automatically pick out each spec and insert it in the proper column. The price information, for instance, will wind up in the price column--no matter whether the price column is named B or E, or whether the document is an Excel spreadsheet or an entry form for a database from Oracle Corp. Maytag figures it can squeeze out as much as 80% of its catalog-production costs.

How can a mere coding protocol produce such startling gains? It's all in the eye of the beholder--in this case, a computer. To PCs and mainframes alike, everything is black or white, a one or a zero. Strings of digital bits are recognized as numbers, but a computer has no way of telling whether "10049" is a price, a part number, or a Zip Code. Today, spreadsheets and database forms provide the context needed to, say, sort part numbers. But when the raw data is transferred from engineering to marketing, which uses a different database system, the data will probably have to be reentered manually.

One obvious solution to this incompatible babble of bits would be special translation programs for converting from one format to another. Sounds simple. But despite the untold millions spent on such software, the manufacturing world is still divided into fenced-off camps running different design, engineering, and logistics systems. Electronic data interchange (EDI) networks are perhaps the most successful tool for relaying digital data across supply chains, but EDI is too limited for big producers and too costly for most smaller manufacturers.

"MARKETING DISASTER." So XML looks like the best bet yet for enabling different computers and programs to understand each other. The concept has been around since the mid-1990s. But only in the last three years have companies begun to define the required tagging standards. Manufacturers are at the forefront because, by some estimates, automating business practices and supply-chain systems with XML will save industry a mind-blowing $90 billion a year worldwide.

Cisco Systems Inc. is one leading proponent of XML. And nearly all software companies are scurrying to add the language to their applications. If you don't, says Robert Austrian, an analyst at Banc of America Securities in San Francisco, "it's a marketing disaster waiting to happen." He believes XML will quickly become essential for any program used in manufacturing and business-to-business relationships.

Indeed, some say XML ranks in importance just below the creation of the Net itself. It started five years ago when a group of programmers began working with the World Wide Web Consortium, which coordinates Web standardization projects. Building on earlier efforts by IBM Corp. and others, they sought to enable networked machines to not only read the information on each other's Web pages but also to understand the associations and relationships among the pieces of data.

That could pave the way for business between Web sites--computers dealing with computers automatically, in a fraction of the time humans need, yet exercising far greater rigor. XML by itself may not be a panacea, admits Timothy Bray, one of XML's authors and now CEO of Antarcti.ca Systems Inc. in Vancouver, British Columbia, "but it provides a framework to solve critical business issues."

For example, if a company uses Oracle's financial software to track production costs, the program might be able to uncover new insights to improve efficiency. With XML, it will understand all the Byzantine manufacturing commands spit out by the plant's mainframe as well as the PCs controlling shop-floor operations, work cells, and individual machines. Running routine financial analyses on production data at that level of detail is now just a dream.

Still, reaping XML's benefits in manufacturing may have to wait a bit. While companies like Maytag have done plenty of XML work on the customer-facing side of their business, bringing XML to supply chains has lagged--and that's where the biggest payoffs will probably come. So Richard L. Hunter Jr., Dell Computer Corp.'s vice-president of supply-chain management in Austin, Tex., figures he'll have to twiddle his thumbs for "three to five years before we see any meaningful impact to the business from XML."

The devil holding things up is, as usual, in the details--agreement on precise definitions for I.D. tags. The trouble is, XML has been floating around for so long that some companies have already developed their own variants, and more XML "dialects" continue to spring up.

LACK OF TRUST. In addition, many competitors remain leery of the basic XML concept. They fret that customers may no longer be as motivated to buy the same brand of system to maintain easy data portability. But solutions must be found, says Jan Harris, vice-president of global sales for Motorola Inc.'s semiconductor business, because "standard tags are a must."

As a bare minimum, all the players within a given industry need to use identical tags. Otherwise, "you'll never get the payoff," says Jennifer Hamilton, chief executive at RosettaNet, an industry group that is defining XML standards for the electronics sector. Several more groups, including BizTalk.org, have sprung up to address the dialect problem. So far, their success has been mixed. One problem: Like feuding suppliers of computer-aided design systems, the groups don't quite trust each other.

RosettaNet is perhaps the main success story to date. It has 105 members and at least 87 of them have some sort of XML project up and running. It hasn't been easy "to get an entire industry to sit down and agree on things," says Hamilton. "That can take several years even within the walls of one company." But if XML is to live up to expectations, every other industry is going to have to follow RosettaNet's lead.

By Jim Kerstetter in San Mateo, Calif.

    Before it's here, it's on the Bloomberg Terminal.