| The following articles, which should be of general interest to those selling
into the insurance industry, appeared in the 3/16/89 VOGON NEWS.
Comments in parentheses are mine.
[OSI COMPATIBILITY WILL BECOME ONE OF THE KEY ISSUES IN THE INSURANCE
INDUSTRY OVER THE NEXT 5 YEARS. WHILE SNA IS TODAYS STANDARDS, IN
DISCUSSIONS, THE LONG TERM BUSINESS IMPORTANCE OF "VENDOR INDEPENDENT"
COMMUNICATION STANDARDS SHOULD BE STRESSED. OSI IS RAPIDLY BECOMING THE
STANDARD IN EUROPE AND MOST US GOVERNMENT CONTRACTS NOW CALL FOR OSI
COMPATIBILITY.]
Digital - Details OSI-based networking strategy
Digital Director of Enterprise wide Networking Lee Sudan reiterated Digital's
commitment to OSI migration for Decnet users, but also indicated that the
company has no immediate plans to discard Decnet's higher level functionality
in favor of the industry standard. Digital now offers a full range of OSI
networking software, including FTAM and CCITT X.400 applications, that allows
users of the existing Decnet Phase IV system to communicate with other
OSI-compliant systems. Decnet Phase V, which should be available next year,
will allow users to communicate both with Digital's existing 300,000 Decnet
devices and with OSI devices. Digital also announced three networking
products, X25Portal 2000, enhancements to existing FTAM software and the
Digital Wide-Area Network Controller 220.
{Computerworld, ?, p. 60}
{Article summarized and distributed by Digital's Market Information Services
Group (MISG)}
[A MAJORITY OF THE INSURANCE INDUSTRY "BACK OFFICE" SYSTEMS RUN UNDER CICS.
THIS IS AN OPPORTUNITY TO MOVE SOME OF THE SMALLER APPLICATIONS (PERHAPS FOR
PRODUCTS THAT ARE NO LONGER BEING SOLD BUT MUST BE SUPPORTED) OVER TO THE
LESS EXPENSIVE VAX BASE]
Digital, IBM - OLTP programs running under CICS migrated to VAX
Though John Akers isn't having any sleepless nights yet, he may if a trend in
the New York area escalates to a national level. Bankers Trust in New York
City and two other large IBM customers have migrated their mainframe-based
OLTP programs running under CICS to a DEC VAX-based Equivalent. This snippet
comes from Ben Rosenberg, president of Advanced Systems Concepts, whose young
company created the vehicle for such a migration, a piece of software named
CICS Liberator. ASC also developed the CICS-equivalent software that runs on
the VAX, a product called INTACT, now owned by DEC. Rosenberg says he and DEC
have had "700 inquires" for CICS Liberator from IBM customers since it was
unveiled in November. The two have embarked on a joint mailing program to
1,500 selected IBM accounts in the New York area to turn some of these
inquiries into firm commitments. There are believed to be more than 30,000
CICS installations worldwide. Many of the programs running under CICS are
between 10 and 12 years old, and typically were written in COBOL. They help
anchor IBM's long-standing dominance of the OLTP marketplace. DEC's sell to
these customers is that by switching their OLTP traffic to VAX's, their
overall costs per transaction will drop significantly. A major erosion of
IBM's CICS base would be very serious for IBM. As Rosenberg notes: "A mere 1%
of these CICS sites buying CICS Liberator would net my company $30 million in
revenue. As the provider of INTACT and large VAX's(as well as numerous
associated products and services) Digital could make considerably
more-billions of dollars." One IBM counterattack is a plan to make CICS more
appealing by offering it under the PC-based operating system OS/2. Such a
strategy helps turn the PC into a major OLTP vehicle in its own right, rather
than just a standalone device or terminal emulator. IBM has also funded its
own "DEC Attract" group in the New York area to lure DEC's minicomputer
customers over to the computer giant's best-selling AS/400, according to
sources.
{Contributed by: Frank Kilsdonk}
{DATAMATION, 15-Mar-89, p. 11}
[THIS ANNOUNCEMENT HAS GREAT POTENTIAL IN THE CUSTOMER SERVICE AREA (A HOT
BUTTON FOR THE INDUSTRY). THE NORTHERN TELECOM - DIGITAL SYSTEM WILL BE
SHOWN AT LOMA]
Computer Integrated Telephony
Canada's Northern Telecom Inc. announced new communication capabilities
designed to integrate voice and data at the desktop. Digital said the
agreement will enable greater flexibility and integration of computers and
telephones for business customers. Among other capabilities, the new product
can connect incoming calls with a company's automatically displayed on
terminals. On-screen dialing, integrated message-taking and electronic mail
are also possible. The software, available in April, will work with Digital's
VAX computers tied to the switching capabilities of Northern Telecom's
Meridian office phone systems. Digital and Northern have a long-standing
strategic relationship, the two companies said.
{The Boston Globe, 15-Mar-89, p. 31}
|
| The following article not only prominently mentions Digital, but also explains
a mission critical expert system application, and positions Nexpert (an expert
system shell used frequently in insurance applications.
ARTICLE
EXPERT INSURANCE
Underwriting Unlimited Via AI
FROM THE INSURANCE SOFTWARE REVIEW JULY/AUGUST 1989
Milwaukee Safeguard, a division of Milwaukee Insurance of Milwaukee,
Wisconsin, is no stranger to new technology. And within Milwaukee
Insurance, the division is known as an innovator in using technology as a
management tool. So it came as no surprise when, faced with a growing
number of complex applications to process, Milwaukee Safeguard coupled
effective management techniques with artificial intelligence (AI)
technology to trim processing time by 60 percent.
The company asked Tom Kruse, a consultant with Milwaukee's Computer
People Unlimited, to help develop an embedded expert system that could
review and approve its insurance applications. Although Kruse was an
experienced developer of decision support systems, he was not a
specialist in AI. But his interest in cognitive psychology aided him in
gathering the necessary expertise from the underwriters. Using Nexpert
Object, an expert system shell from Neuron Data, Inc, of Palo Alto,
California, Kruse translated underwriters' knowledge into a form that the
computer could understand.
"I've been in this business a long time and have gained a lot of
experience which enables me to make decisions quickly," explains Jim
Owens, Milwaukee Safeguard's vice president of operations. "But all that
knowledge is stored away in my gray matter somewhere." Kruse's job was
to take that 20 years' of experience and put it into a form that the
machine could emulate.
Experience is especially important at Milwaukee Safeguard. Its
underwriters insure some 280,000 automobile drivers. Those insurance
applicants range from drivers with squeaky-clean "preferred" records
(which are easy to process) to the more difficult, no-so-clean
"non-standards" which make up the bulk of Milwaukee Safeguard's
clientele. Before the expert system was implemented, it could take 10 to
15 minutes just to review the information necessary to reach a decision
about a "non-standard" applicant. With hundreds of applications to go
over weekly, underwriters were constantly racing the clock.
Because Milwaukee Safeguard had plans to switch from IBM to Digital
Equipment Corporation hardware, portability was an important criteria for
expert shell selection. Since data would be accessed from existing
database files, the ability to interface to those files (particularly the
Digital Rdb files) was critical. Nexpert Object, which runs on
Macintosh, IBM PC, AT and compatible, Digital VAX and various UNIX
workstations, met these criteria, and also supported object-oriented data
structures.
While the system could have been built without object-oriented
support, it would have been more difficult and less elegant. For
example, there is no set number of drivers or vehicles for a policy. To
allow the program to handle all these permutations with an object
orientation would require extensive programming. With objects, the
developer can use generic rules which apply no matter how many drives or
vehicles apply to a policy.
Another Nexpert function, a "metaslot," allowed Kruse to imply certain
rules, rather than spell out every simple step the system had to follow.
"With a minimal number of commands, I caused all sorts of things to
happen," he recalls. "For instance, if I wrote a rule that says a driver
is 'completed,' it implies that the system has already gone through an
in-between step and checked to make sure that all the data is there."
While many expert systems have failed because employees sabotaged the
knowledge acquisition phase, Kruse was able to sidestep this pitfall.
Most opposition comes from fear that the employee is no longer valued and
may be replaced by the system. At Milwaukee safeguard, fear was used as
a positive tool to ensure the system's success. After the underwriters
were reassured that the system was to be a tool, not a replacement, they
were reminded they each had a large incentive for cooperating with Kruse
in the knowledge acquisition phase. Their paychecks depended on it.
At Milwaukee Safeguard, performance is measured by the loss ratio.
Since the system could affect the loss ratio if it was based on
inaccurate information, underwriters were personally affected by the
system outcome. Kruse used other methods to ensure cooperation. He
shared information freely. Every attempt was made to increase each
underwriter's understanding of the system.
In fact, Kruse tried to accommodate underwriters' requests even when
he felt they were unnecessary or redundant, just so the underwriters
would feel more comfortable. It worked, says Kruse. "They wanted the
machine to do more and more and more." At that point Kruse and Owens
realized they were successful. The users owned the system.
While native Nexpert was ideal for Milwaukee Safeguard's data
intensive batch environment, it would not be as effective for a highly
interactive application. Building screens or displaying information is
difficult and cumbersome in native Nexpert. Add-ons from other vendors
can help improve the problem, however. And this product is strong on
data handling. Mapping the Digital Rdb structure within Nexpert was
fairly straightforward.
For Milwaukee Safeguard, Nexpert has been a good investment. This
$5,000 product generated a system which can review 100 applications in a
mere five seconds. The system is consistent too. "Unlike humans," Owens
points out, "the system doesn't argue with its wife or get up in the
middle of the night with crying babies or fight traffic on the way to the
office." And its judgment matches that of the most experienced
underwriters "100 percent of the time."
|