Thinking Machines Corp. has been through bankruptcy and other hard knocks in the last 15 years, yet its chief executive officer describes it as "a babe in the woods."
Its past as a supercomputer heavyweight behind it, the Burlington, Mass., company has been reborn as a lean, 40-person operation that since last December has been selling a data mining system called Darwin.
With bankers among those most intent on putting such sophisticated data base programs to work in customer marketing, cross-selling, and profitability analysis, Thinking Machines has the financial services industry centered in its marketing sights.
President and CEO Robert Doretti said that, on a scale of 1 to 10, with 10 signifying maximum importance to Thinking Machines, financial services rates a 9.9.
"It is the appropriate target for our product," he said. "They need it. When you market a product, you find the need and fill it.
"Our goal is to automate the model-building process, but the most difficult part in all of this is to prepare the data," he added. "Once you have that done, the mining portion is relatively fast."
Credit Suisse of Zurich is using the powerful Darwin system for predictive modeling of customer behavior.
Mr. Doretti said he has numerous prospects among banking companies in the United States but the product, which costs $50,000 and up, requires a three- to six-month sales cycle.
"Most people are getting used to this type of technology," said Mr. Doretti. "If they have data miners or data analysts already on board at their own company, they tend to move faster."
"We're a babe in the woods with a great product," he said of Thinking Machines and Darwin.
If his sales prospects are hearing about Thinking Machines for the first time, they may not be aware of the rich but less than triumphal history out of which Darwin sprang.
Founded in 1983 by Massachusetts Institute of Technology computer scientist Daniel Hillis, Thinking Machines spent $120 million in its first 11 years and turned profits only in 1989 and 1990. It sought bankruptcy protection in 1994 when it had $33 million in debts.
Production was shut down but not service and support to the 70 supercomputers it had installed for the federal government, colleges and universities, and oil and gas companies.
The company came up with a new business plan that creditors liked. In exchange for forgiving $7 million of debt they got 15% of a reconstituted Thinking Machines. The rest of the debt went into a separate business called TM Patents, whose intellectual property is expected to pay it off over time.
Mr. Doretti joined the company Feb. 1, 1995, as president and chief executive. A year and one week later, Thinking Machines came out of bankruptcy. "What was fun about it was that we came out with no debt," Mr. Doretti said.
In its previous incarnation, he said, Thinking Machines "imploded. The market dried up in the federal world, primarily because of changing technology and the Cold War ending. What we sold for $5 million or more you could buy for $1 million or less."
The company had also lived high on the hog. The Boston Globe published an article in November 1995 that compared the extravagant lifestyle of previous management under former CEO Sheryl Handler to what had become a spartan outfit. The former Thinking Machines had a gourmet restaurant for employees; the current one has a cafeteria with soda machines.
The company kept its name because it connoted "great technology," Mr. Doretti said. "We have to convince the world that Thinking Machines is now a new company in chapter one, not Chapter 11."
Mr. Doretti, 55, was previously president and CEO of Vision Ten Inc., an X-ray technology company in Los Angeles. He was attracted to Thinking Machines because he "liked the turnaround business" as well as the idea of returning to the Boston area. He had lived there from 1972 to 1986, when he was senior vice president of sales and marketing for Wang Laboratories' $1.8 billion U.S. operations.
Only five of the original Thinking Machines staff of 180 remain. Most of the rest went with the services or parallel technical businesses as they were sold off.
The biggest creditor was a Realtor, so Mr. Doretti moved the company from its building in Cambridge, Mass., to a more modest one in Bedford and then last October to New England Executive Park in Burlington.
Refocusing on software, sales, and support, Thinking Machines has offices in Seattle, Dallas, Chicago, Washington, and New York, and a distributor in Japan. It is about to open a European office in Zurich.
Darwin had its beginnings in the old supercomputer world but has been turned into a Windows-based system. "We've packaged it into a market- oriented application," said Mr. Doretti.
It can sift trillions of bytes of data to make predictions and produce analysis. Mr. Doretti described its data mining capabilities as a next step beyond OLAP, or on-line analytic processing, at which many banks are working to become proficient.
"OLAP is hindsight; data mining is foresight and insight," he said. "We go in and help companies determine sales and find patterns in behavior."
"Darwin is a very respectable product," said Herb Edelstein, president of Two Crows Corp., a data mining consultancy in Potomac, Md. "I'd compare it to Intelligent Miner from IBM and Mind-set from Silicon Graphics."
SAS Institute Inc.'s Enterprise Miner, he said, is "trying to solve the same problems but doesn't address scalability," whereas Darwin affords "the ability to deal with large amounts of data effectively."
Mr. Edelstein said the system is suited for moderate-sized as well as large data bases.
The financial industry is the key to Thinking Machines' success because it is "most quickly adopting this technology" to combat risks associated with loans and fraud. "Data mining is all about giving information to make good decisions about what's going to happen," said Mr. Edelstein.
Thinking Machines has accordingly focused on supplying predictive data mining to the financial services, telecommunications, and data base marketing industries. It believes its supercomputer heritage gives it a leg up.
"We knew the complex algorithms and artificial intelligence techniques needed to manipulate data sets," said Mr. Doretti. "We drew upon the best of the old Thinking Machines to forge our new commercial focus."
"Many data mining companies will work with one algorithm, we work with multiple algorithms," he added. "We can do analysis from different directions. That's what makes us different."
Spearheading the financial services thrust is William Guild, 41, who joined Thinking Machines in June 1997 as director of business development.
He came from Fleet Bank in Boston where he was vice president of the commercial services division; before that, he had worked for Mellon Bank.
"Bill was the first noncomputer person to come here from a bank," Mr. Doretti said. "We needed someone to drive us to where we should take the technology, and Bill has helped us tremendously to do that."
They are looking at financial service applications including customer attrition, cross-selling, retention, and profitability.
"We try to set up a good return on investment for the use of our software," Mr. Doretti said. "We believe it drives up the customer loyalty level."
Tools are valuable, but "we need the applied solutions," said Don MacTavish senior research analyst at Meta Group, Stamford, Conn. Thinking Machines "has the skills to assist organizations build applications. We expect to see the growth of predefined applications, which are then customized to a customer's needs."
Some small companies offer low-end data mining products, but "for all intents and purposes you'll hit the wall with micromining. You can only analyze so much data," Mr. Doretti said. "We do macromining. We work on very large data bases. The power of the product" is ahead of the competition, he added.
Mr. Guild's definition of macromining is that there is no limit to "the number of records, the number of variables we can evaluate. We have a role in the preparation of the data too. One of the functions that our tool performs is to identify which data are important to the business problem."
"Darwin is the heartbeat" of the customer analysis and loyalty management system at Credit Suisse in Zurich, said Mr. Doretti. The bank is looking for a 10% to 20% minimum improvement in customer retention rates.
"This is a new function for a bank, and often it's brand new," said Mr. Guild. "A bank typically would be doing statistical analysis or nothing."
In the data warehouse market, said Mr. Guild, a lot of people build warehouses and then don't realize a major benefit. "Banks are building data warehouses, but they're not sure where to go next," he said. "The next step is to analyze it.
"Data mining is different in that people become comfortable in actually doing it and they really get results. It produces good returns, and that will drive it very, very quickly once it becomes known," he added.
The statistical knowledge market is "hot," said Laurence Jacobs, chief scientist for Europe at NID Research, Boston, and lead project manager for the Credit Suisse implementation. "Everyone is getting into it, and they're all trying to make it easy-to-use and inviting."
"Ultimately these tools can make it in large numbers," Mr. Jacobs said, "if people develop actual applications that speak the language of users. Thinking Machines is definitely doing that."
Since its turnaround, Thinking Machines has formed alliances with vendors and system integrators, including Perot Systems, Electronic Data Systems, Alltel, Sun Microsystems, PriceWaterhouseCoopers, Oracle Corp., Paragren Technologies Inc., Acxiom Corp., and KPMG Peat Marwick.
"Financial services is a nice place for this type of information," said Mr. Doretti. "They've never really analyzed what is happening in their customer base. Now that everything has become deregulated and a bank, insurance company, or mutual fund company can offer different services, this type of product becomes more meaningful."
Mr. Doretti said he hopes that in the next five years the company will have revenues exceeding $50 million and can consider going public.
"Our job is to stay ahead of the curve in technology and to offer easy- to-use applications and support," he said. The ultimate measure of success will come when the company gets Darwin "in the hands of the masses, not the few."