For most speech applications, the greatest cost incurred relates not to the equipment itself, but to the process of designing and integrating the system into new or existing work practices. This represents the biggest hurdle for speech system vendors. Recent developments in speech technology now provide opportunities for improving productivity across a broad spectrum of commerce and industry. However, the level of user satisfaction is often disappointing, despite a continuing reduction in price and improvements in hardware and software. The interface to be established between user and computer when using speech technology equipment is more complex than other user interfaces. Issues range from intricate systems and programming topics to human factors problems in physiological, psychological and sociological fields. Many industrial and commercial applications typically require vocabularies ranging between twenty and a hundred words, with a low level of syntax. There is a frequent requirement for an effective human-machine dialogue, though the amount of data transferred is small compared to an office system. The technology to achieve this type of activity exists and has high performance. This paper restricts itself to this area of interest. The development of concepts presented here has largely been possible through work carried out under European Economic Community ESPRIT Research & Development Project 449 [1]. This project was conducted over the period from August 1984 to September 1986; its primary objective was to study areas of actual or imminent application of speech technology in commerce and industry and the corresponding requirements for hardware and software. The project Consortium was comprised of: Voice Systems International Ltd, Cambridge, UK; International Computers Ltd, Reading, UK; British Maritime Technology Ltd, Wallsend, UK; Fincantieri, Trieste, Italy.