Development of a design system for an AI call intake assistant - a user-centered reverse engineering approach

Abstract

Although user-centered design (UCD) advocates for early user involvement, practical constraints often result in Human Factors integration taking place only when prototypes already exist. This paper proposes a solution to this challenge by employing a "reverse-engineered" UCD approach to an existing AI call intake assistant prototype. Deviating from the traditional linear approach, this approach reconstructs the problem space directly from the solution space. To apply this approach, a multi-stage method was utilized. First, a group interview was conducted with five developers of the existing AI prototype to elicit their implicit design rationales and assumptions. From this interview, preliminary usage requirements were retrospectively derived. Secondly, these requirements were validated and refined through a focus group involving 14 dispatchers, followed by a Kano analysis to identify basic, performance, and excitement factors. The study revealed critical discrepancies between developer assumptions and actual user needs. For instance, dispatchers required the capacity to intervene and cancel the AI assistance at any moment. The consolidated set of requirements now serves as the foundation for the next development phase of the AI assistant, ensuring better alignment with actual work practices. This demonstrates how a prototype can serve as a probe to deliberately step back into the problem space. Furthermore, the reverse-engineered approach holds particular promise for future rapid AI-based prototyping, especially when user involvement occurs late. This ensures the continued efficacy of human-centered design principles.

Keywords: User-Centered Design Process, Reverse Engineering Approach, Ai Call Intake Assistant

DOI: 10.54941/ahfe1008142