![]() |
|||
![]()
|
![]() |
![]() |
![]() |
5. THE TOOL MARKETUp to the mid-1980s, sales of expert system development software were made primarily to universities, the U.S. military, and a few research groups within major companies. Users of this software were principally researchers who were exploring the potential of the technology. Being used mainly for experimental reasons, rather than focused on solving practical commercial problems, the number of tool sales during this period was modest. This situation changed dramatically in the late 1980s as the commercial value of a deployed expert system was realized.
Figure 5 shows total sales figures for expert system tools during recent years, as reported in the monthly newsletter Intelligent Software Strategies, one of the best publications for monitoring the state of the field (Harmon, 1993; 1994; Hall, 1996). These sales figures include tool sales, upgrades, maintenance, and runtime licensing. All sales are for worldwide sales of commercial intelligent software development tools for North American vendors. The figure shows that total sales have grown at an average of approximately 17% per year since 1988. According to personal correspondence from Timothy Biebelhausen, associate vice president for investments at Kemper Securities, many economists argue that a healthy growth rate for any industry is around 10% per year. Clearly, the expert system tool vendors have had a good run. One event worthy of note is the dip in sales between 1990 and 1991. It was during this period that many critics were pointing to the decline of tools sales as an indicator of a field that had fallen out of favor. Some concerned tool vendors even began to drop the AI label from their products, while maintaining the same product features. As Figure 5 shows, sales have rebounded quite well. Vendors now do not hesitate to advertise their products as containing AI capability. In fact, they often use it to their advantage to distinguish their product from competitors who market products without this capability. Figure 6 shows total sales of LISP tools (Harmon, 1993; Hall, 1996). Sales of LISP tools had a dramatic drop in 1991 but have maintained a relatively constant level over the past several years. When compared with Figure 5, it is clear that the LISP market is relatively small. However, LISP offers symbolic processing capability that cannot be found in other languages. It should, therefore, continue to remain popular among many expert system developers, particularly in AI research activities. Figure 7 shows sales of domain-specific tools, which shows a substantial growth in this tool category (Harmon, 1993; Hall, 1996). Representing only 4% of total tool sales in 1988, these tools rose to capture 39% of the market in 1995 -- a tenfold increase. This is dramatic evidence of the evolution of the vendor market, with vendors moving from providing general-purpose tools to ones tailored to the specific needs of their customers. This trend is very likely to continue.
Figures 8 through 10 show, respectively, tool sales for the PC and Mac, workstation, and mainframe. Recognize that, in general, the larger the platform the larger the tool cost. Therefore, these figures don't show how many tools are sold for each platform, only their total sales. Sales of PC and Mac tools showed a dramatic decrease coming out of the 1980s. Since then, sales have shown a small but steady increase. As these small platforms continue to become more powerful, this category of tool sales should show modest but sustained growth. Interesting, though, from a perspective of the technology, the decrease in small platform sales when entering the 1990s is a positive sign. Most organizations purchasing a small platform tool are looking to test-the-water; that is, can this technology help us? They will often take this initial step before committing to the purchase of a more costly tool that will run on a larger platform. Given the many success stories of expert system applications, many organizations are now confident in the technology and are shifting to larger platform tools.
Tools sold for workstations offer more features than ones that run on a smaller platform, such as a PC. They are designed to run in Unix, OS/2, or Windows NT environments. They are chosen for applications that require higher speed and integration needs, and are better suited for client-server development. As Figure 9 shows, sales of workstation tools have increased each year. This trend should continue. Purchasers of mainframe tools are usually large organizations who presently have a large traditional application, such as a database management system found in a financial or insurance organization, who are looking to the technology for improving the use of the application. In general, these types of organizations are conservative by nature and move slowly to adopt a new technology. As seen in Figure 10, there was a large decrease in mainframe sales between 1990 and 1991. This is the same period when it became popular to criticize the technology. Being conservative, these organizations began to back away. Recently, mainframe sales have increased and should continue as the commercial benefit of the technology is further realized.
Sales of CBR tools have shown a dramatic increase in recent years. From approximately $6 million in 1993, sales increased by a factor of over three to around $20 million in 1995 (Hall, 1996). 6. THE FUTUREDuring the past 2 decades, the field of expert systems has come far. From a technology confined to research circles in the early years, we find it today being used commercially to aid human decision making in a wide range of application areas. Supporting this transition has been the availability of expert system development tools that permitted a large audience to participate in this emerging field. The expert system tool market should continue to play a major role in future developments in the field. The following is a discussion on what might be expected in several areas.
|
![]() |
|
Use of this site is subject certain Terms & Conditions. Copyright (c) 1996-1999 EarthWeb, Inc.. All rights reserved. Reproduction in whole or in part in any form or medium without express written permission of EarthWeb is prohibited. Please read our privacy policy for details. |