Neuron Data, the Palo Alto-based provider of expert system tools for commercial and industrial integration, recently made its first official UK statement. The UK was thus honoured because Neuron Data is pulling out the stops to increase the British business community’s awareness of expert systems products. This is not exactly a novel ambition among companies touting artificial intelligence these days, but then Neuron Data does claim to have around 40% of the US market for expert systems under its belt, so perhaps its founder member and chairman Patrick Perez (sporting fashionably long hair and baggy suit) had a special insight to offer its competitors. The reasons for Neuron Data’s successful sale of 5,500 licences worlwide and its profitability since Year One (ie 1985) are varied. According to Perez the main reason for his company’s leadership lies in the fact that its flagship product, the Nexpert Object toolkit, is written in C and has full interoperability across all main operating systems. In fact, one of the themes of this press briefing was to tell the UK that the Lisp and Prolog languages in which the majority of British offerings are written are dead or dying in the US and France. In marketing terms, Neuron Data has clearly benefitted from its strategic alliances with companies such as IBM, DEC and Sun. Indeed, DEC distributes the Nexpert product directly in the US, France and Germany, while Sony will soon be selling the product in Japan. However, about 66% of Neuron Data’s sales worldwide come from its distributors in the OEM and value-added networks, and it is left to them to focus on placing Nexpert in niche markets. Neuron Data itself prefers to pursue a general marketing strategy and its latest ambition is to have its expert system embedded in all computing systems in the mainstream computing market such as spreadsheets, computer-aided software engineering tools, desktop publishing systems, Hypertext programs and operating systems. The idea behind this project is that: Artificial intelligence is the intelligent agent provid ing the technological glue between existing computing paradigms to enhance functionality and create greater power. Basically this translates into the broad concept that if ordinary desktop facilities acted more like people than computers then their malleability would make them more useful; so long, that is, as powerful enough user interfaces are available, and the AI technology is meshed with the existing computing environment in a seamless fashion. Neuron Data clearly thinks it is approaching the practical implementation of this philosophy with the launch of Nextra, a knowledge acquisition tool for visual thinking. Inductive capabilities Nextra conducts interactive interviews with human experts or users to gain facts and opinions from them. This knowledge is then represented graphically and can be edited and reviewed by the user while the tool’s inductive capabilities construct and integrate rules, classes and objects from the knowledge bases which can then be imported into Nexpert. Nextra is initially available on Apple Mac II, but the knowledge base it generates can run with Nexpert Object on Sun, Hewlett-Packard, Apollo, and Apple Mac machines as well as on DEC VAXs and IBM PS/2s, RTs and mainframe operating systems. All well and good on the product front, but the problem of penetrating the UK market still remains. The Neuron Data solution is to heighten awareness of expert systems by opening a marketing office in Central London. It has no intention of selling direct to the British market, however, which will continue to be served by Bechtel AI Institute, Megatron Computer Systems and Software Sciences. Rather Neuron Data believes that it’s all a quetion of product potential UK clients see expert systems as toys produced by small research-based companies, but once they spot the real thing there will be no stopping their desire for embedding artificial intelligence technology. Possibly, but strategic alliances for the UK market with firms like DEC and Apollo probably wouldn’t come
amiss.
