Software Gang Puts ‘Experts’ in the Office
They are known, in industry circles, as “The Gang of Four”: IntelliCorp., Teknowledge, Inference Corp. and Carnegie Group.
Their mission, proclaims Alexander Jacobson, president of Los Angeles-based Inference, is nothing less than to usher in “the dawn of the second generation of the computing industry.”
They are makers of programs that allow customers to build “expert systems.” Such systems give computers the ability to mimic the reasoning of human experts who make complex decisions after sifting through reams of imprecise and sometimes conflicting data.
Computing’s first wave automated routine clerical functions--such as bookkeeping--that involve manipulation of precise data. Jacobson says the second wave will automate white-collar jobs--professions in which coming up with the right solution to a given task or problem depends on hunches, guesses and rules of thumb as much as on concrete data.
“Our software doesn’t make a thinking machine,” Jacobson says. “It is only an emulation, but it is a good one.”
The Gang of Four companies, in short, are among the pioneers in the commercialization of the nascent field of artificial intelligence. “We’ve all heard about the ‘Buck Rogers’ aspect of artificial intelligence,” says Lee Hecht, president and chief executive of Palo Alto-based Teknowledge. “We’re looking to the bread and butter.”
And while skeptics dismiss Hecht and Jacobson’s enthusiasm as just so much industry hyperbole, the fact is that expert systems are steadily winning adherents in an ever-widening range of businesses and government agencies.
Consider:
- Composition Systems Inc., using Inference’s Automated Reasoning Tool, is designing an expert system to take the crisis atmosphere out of newspaper publishing. “Intelligent” systems will tie together page design, advertising and production functions, tracking the flow of news and advertising copy and suggesting page layouts based on priorities set by editors.
- Phillips Petroleum engineers, using a system built around Teknowledge’s S.1 program, can tap out the characteristics of a rock sample on a computer keyboard--and almost instantly learn the probability of finding oil.
- When Digital Equipment customers pick components for new computer systems, an expert system incorporating Carnegie Group’s Knowledge Craft software determines whether the new equipment will fit together, suggests alternatives if it will not and even draws diagrams for Digital’s field service engineers to use during installation.
- The Securities and Exchange Commission is using an expert system built around IntelliCorp’s Knowledge Engineering Environment program to help in the analysis of corporate financial information.
The market for expert systems software, while tiny compared to the multibillion-dollar data processing business, is growing fast. Industrywide revenue, which totaled $74 million last year, will roughly double to $140 million in 1986, according to DM Data, a Scottsdale, Ariz., consulting firm.
Although software makers say the prediction for 1986 seems likely to be borne out, DM Data’s projections for future growth seem a good deal more speculative. Indeed, there are signs that the market is chilling as a result of the cooling economy and the computer industry slump.
Revenue Forecast
The firm predicts industry revenue of $245 million in 1987, $385 million in 1988, $570 million in 1989 and $810 million in 1990. Such long-range projections for many other promising high-tech products and services have proven to be much too optimistic in the past.
But there is no denying that corporations, both here and abroad, are keenly interested in expert-systems technology. Corporations holding equity investments in the leading expert-systems companies include General Motors, Procter & Gamble, FMC, Nynex, Framatone SA and Societe Nationale Elf Aquitaine (Teknowledge), Lockheed and Ford Motor (Inference), and Digital Equipment, Ford and Boeing (Carnegie Group).
Under the typical “strategic partnership” arrangement, the corporate giant buys stock in the software company, gets a seat on its board and agrees to use its products and services. GM, which made a big commitment to the computer services industry when it purchased Electronic Data Systems, is using Teknowledge-based computer programs to train dealers and design more efficient electric motors.
“Our corporate investors gain a window on this new technology,” explains Teknowledge’s Hecht. “We get capital and a measure of independence from traditional sources of venture capital.”
With revenue still small--the largest of the four, IntelliCorp, booked $18.6 million in revenue in its fiscal year ended June 30--the firms are concentrating on positioning themselves for the future and on differentiating their products and services. They must also brace to fend off a host of smaller competitors as well as possible future competition from such industry giants as International Business Machines.
No University Ties
Inference, headquartered on West Century Boulevard near Los Angeles International Airport, is the oldest of the four, having been founded in 1979. Unlike the other three, which boast of ties to major universities, Inference was not incubated in an academic milieu.
Though some may view that as a minus, Jacobson sees it as a plus. “From inception, we developed our product in a commercial environment,” he says. “The commercial world wants commercial goods. All the others are academic adaptations.”
Inference, whose stock is closely held, is widely acknowledged to have the most powerful reasoning program. But it is also the most difficult to learn how to use, and its greatest market acceptance has come in the technology-hungry aerospace industry. “It’s going to be a more thorny penetration in other markets,” Jacobson acknowledges.
IntelliCorp’s strengths lie in its powerful product and large support and training staff. “We will hold the customer’s hand clear through the delivery of their application,” says Gene Kromer, president of the Mountain View, Calif., firm, many of whose scientists come from nearby Stanford University.
Can Limit Market
Alone among the four, IntelliCorp has shunned investments in it by big companies; instead, the firm went public in 1983 at $6 a share and closed Friday at $8 a share. “If GM is a strategic investor in you, you may not conduct much business with Ford, Chrysler and Toyota,” Kromer says.
Teknowledge, based in Palo Alto, also has its roots in Stanford’s artificial intelligence laboratories.
Teknowledge has a high profile as a result of its affiliation with GM, Procter & Gamble and others, but observers say its S.1 program is the least powerful among those sold by the four. “It is a product that is very narrow in scope,” says Kromer, echoing sentiments by others in the industry.
Teknowledge is also faulted for being overly dependent on service revenue. Its four largest customers accounted for 47% of total revenue in fiscal 1985. Such customers often require expensive customized service.
But Teknowledge’s programs are also very easy to use, in part because they’re written in a language known as “C” that is widely understood by programmers. Historically, artificial intelligence programs have required a knowledge of such arcane computer languages as LISP and PROLOG.
To broaden the market, Teknowledge also offers an entry-level program called M.1 that runs on the IBM PC and sells for just $1,500 a copy, compared to as much as $65,000 for some of its competitors’ products.
Many Intimidated
“Too many users are intimidated,” says Teknowledge’s Hecht, “because they view AI (artificial intelligence) as a technology dependent upon a high priesthood. We want to make the technology accessible to mainstream users.”
Finally, there is Carnegie Group, which has specialized in producing software that helps solve problems in manufacturing and engineering. Both are specialties at Pittsburgh’s Carnegie-Mellon University, which owns a small stake in the firm and has licensed its technology to the company.
“We’re the only one of the four with such a tight focus,” says Larry K. Geisel, president and chief executive. Like Inference, Carnegie Group hopes to go public later this year, though the market has been chilled by the general decline in stocks.
In addition, artificial intelligence companies have been hurt by the announcement this month by Symbolics Inc., a maker of computers used in AI applications, that its fourth-quarter earnings would be lower than last year’s, and by Teknowledge’s projection of a fourth-quarter loss of $270,000.
Still, says Jeffry Canin, an analyst with Hambrecht & Quist, AI stocks such as Teknowledge and IntelliCorp “are selling at stratospheric multiples.” Teknowledge, which went public at $13 a share earlier this year, posted net income of about $710,000, or 13 cents a share, on revenue of $14.6 million in fiscal 1986, which ended June 30. Yet its stock closed Friday at $12.87 1/2, for a multiple of about 100.
Concludes Canin: “An investment is a real bet on the future.”
