The use of interface agents for email notification in critical incidents

This study reports on several typical scenarios of the use of email notification interface agents under the influence of critical incidents. An interface agent is a reactive, collaborative, and autonomous visual computational system, which communicates directly with a person offering assistance and advice in performing computer-related tasks. The critical incident technique was employed to survey the actual users of an interface agent-based email notification application. Respondents were asked to provide the last most significant either positive or negative incident of the usage of interface agents in their email application. They were also asked to offer recommendations for designers and marketers of this technology. Sixty critical incidents were obtained and analysed. With regards to positive-outcome situations, one representative scenario was constructed. With respect to the negative-outcome events, three distinct scenarios were identified. Based on the critical incident technique, it is concluded that users acknowledge the quality of an agent when it acts reliably, an agent's intrusive behaviour results in an immediate agent usage termination, operability issues sometimes force people to reject the technology, and users attempt to preserve the employment of an agent under the negative impacts of external factors. A number of other practical recommendations for manufacturers and marketers are also outlined.