Action Research on Enhancing Record Keeping with GenAI at The Salvation Army

Abstract

Generative AI (GenAI) is rapidly transforming industries and business, offering unprecedented opportunities for automation. Organizations are often burdened by administrative tasks, such a record keeping. Record keeping is essential for accountability, care quality, and regulatory compliance, yet professionals often experience it as time-consuming and burdensome, diverting attention from client engagement. To address this challenge, the aim is to test in practice how an GenAI powered app can enhance the record keeping capability of an organization.This research was carried out in the Department of Juvenile Protection of the Dutch Salvation Army in the Netherlands. Following an action research approach, we tested the Luisterlinie application, a GenAI-powered tool designed for transcription and summarization of client conversations and meetings. During seven weeks (April 2025), 47 youth protection professionals participated in the pilot. Data were collected through interviews, three surveys, application usage logs, and feedback sessions. Quantitative and qualitative analyzes were combined to assess impacts on time spent, quality and completeness, and user satisfaction.The findings demonstrate that GenAI reduces record keeping time, eases administrative workload, and improves user satisfaction, while allowing professionals to remain more focused on client interaction. Participants reported perceived efficiency gains and similar completeness compared to humans. However, challenges emerged with respect to transcription accuracy, speaker attribution, and integration with existing workflows, underscoring the continued need for human oversight.The study concludes that GenAI applications like Luisterlinie can enhance organizational record keeping capability, particularly in contexts where minor inaccuracies are tolerable. Although promising, these tools should complement rather than replace professional judgment. Broader adoption requires improvements in accuracy, workflow integration, and reliability.

Ähnliche Arbeiten