Augmenting Administrative Capacity with Generative AI: Best Practices and Outcomes of Claude AI Training at IDS Jakarta

Abstract

This community engagement initiative evaluates the implementation of a structured sharing session under the Claude Partner Network Learning Path to train administrative staff and interns at International Design School (IDS) Jakarta. Grounded in andragogical principles, the training prioritized interactive demonstrations, practical prompt engineering, and hands-on problem-solving over traditional lecturing to address repetitive administrative workload bottlenecks. The intervention was executed across four structured stages: large language model orientation, best-practice demonstration, account setup guidance, and independent task completion. Pre- and post-session evaluations demonstrated a substantial increase in overall mean comprehension scores from 42 to 81 out of 100, with interns achieving higher post-training scores (87) than full-time staff (78) due to greater initial digital exposure. Furthermore, perceptual surveys revealed that 90\% of participants perceived Claude AI as effective in accelerating administrative tasks such as email drafting and document summarization, while 85\% reported an improved understanding of prompt engineering. Additionally, 80\% expressed self-efficacy in utilizing Claude independently, and 88\% requested division-specific follow-up workshops. Overall, this project highlights the efficacy of interactive, task-oriented AI training in bridging technical literacy gaps and augmenting workplace productivity within creative higher education environments.