A primary appeal of employing agile software development approaches is the ability to tailor them to any specific project, giving rise to numerous techniques for identifying software features and prioritizing the features for inclusion in process automation software. A literature review found that the majority of existing models equate a software feature’s value to its potential to deliver financial benefit. As agile approaches are applied to the automation of increasingly complex, safety-critical, human-in-the-loop processes there is need for a model that instead considers potential to reduce the process’ human error risk as the basis for defining a software feature’s value. This thesis addresses this need by proposing a framework for feature elicitation and valuation using established Human Reliability Assessment (HRA) task analysis methods including expert elicitation and fuzzy linguistic error frequency estimation. A feature scoring method that rewards potential for human error reduction was developed and the results shown to be compatible with popular feature prioritization models. The framework was employed during automation of NASA’s robotics mission design process;​ a complex, safety-critical, human-in-the-loop industrial process. To assess its impact, process operators were surveyed prior to and following introduction of software developed using the framework. Through the framework, NASA successfully elicited features and prioritized them using Popli’s value versus cost model. As a result of the automation effort, NASA realized a 3157 worker-hour/year time savings, a statistically significant decrease in frequency of 12 human error modes and complete elimination of an additional 10 human error modes. It took approximately 376.75 worker-hours to perform the feature elicitation, valuation, and prioritization, accounting for approximately 4.7 percent of the automation project’s cost. The case research approach adopted for this project was unable to compare the effectiveness of the proposed framework against other techniques;​ however, the research successfully demonstrated the framework’s feasibility when applied to a project typical of those in the community of practice. Further, the work offers a novel application of HRA methods to the field of software design decision making and provides a benchmark for future replication and expansion.