Improving the Custom Color Ordering Process with Artificial Intelligence for a Manufacturing Firm

Abstract

Purpose: Practitioners who try to apply AI in Business Process Management (BPM) often face a gap between theory and real-world implementation, because most prior work stays at a high level and offers limited replicable implementation blueprints or measurable evidence of value. This study addresses that gap by reengineering a custom color ordering workflow, following the BPM lifecycle and integrating two AI components: machine learning for palette extraction and generative AI for reference image previews.

Methods: The as-is process was mapped in BPMN 2.0 and analyzed using root-cause and value-added analysis. A to-be process was then designed using BPM redesign levers (task elimination, resequencing, integration, and enabling technology). An integrated web platform prototype was built and tested to measure four Process Performance Indicators (PPI): Cycle Time, Response Time, administrative staff hours per order, and Automation Ratio.

Result: Prototype tests showed significant improvement. Response Time reduced from 8 days to 8 minutes (-99.4%). The overall Cycle Time reduced from 92 to 74 days (-19.5%). The administrative workload per order was reduced from 88 to 0.4 staff hours (-99.5%). The redesigned workflow automated 8 of 15 activities, increasing the Automation Ratio from 0% to 53.3%.

Novelty: This study delivers a practical, replicable roadmap for embedding AI capabilities into an operational workflow while staying aligned with BPM principles and tracking measurable PPI. The blueprint can help organizations respond faster to customers, improve throughput, and reduce administrative burden, so staff can focus on higher-value work.