Managing AI system development in critical large scale applications

It is shown that it is possible to manage the introduction of an AI-based technology into an existing nonstandard mainframe-based computing environment within tight timescales and under the requirement of a very high level of quality. The introduction of this automation to SWIFT´s (Society of Worldwide Interbank Financial Telecommunications) core business has necessitated the use of a project management approach which could ensure efficiency of resource usage and maintain close quality control (from the user´s standpoint) as the system is developed. These challenges have been met by a combination of factors, the most important of which have been the use of a small but high-quality development team and the incremental cycle model of software development. These techniques are made possible by the nature of today´s expert system development tools and represent a considerable departure from traditional software development practices