PT Tristan Digital Teknologi is an Internet Service Provider (ISP) facing challenges in managing customer service, particularly when network disruptions trigger a simultaneous surge of complaint messages via WhatsApp and Telegram. Manual handling causes message pile-ups, delayed responses, and inconsistent complaint-ticket recording. This study aims to design and build an Artificial Intelligence-based automated helpdesk system that integrates the n8n workflow engine with WhatsApp (via Gowa Gateway) and the Telegram Bot API. The system employs Natural Language Processing (NLP) powered by the Google Gemini model to identify customer message intent, extract key entities, generate natural automatic responses, and record complaint tickets into a Google Sheets database in real time, all hosted on the Sumopod cloud server for 24/7 operation. Black box testing was used to validate system functionality. The results show that the system responds within 2-3 seconds on average, compared to 15-20 minutes for manual handling, with 85-92% intent-identification accuracy, 80-88% entity-extraction accuracy, and 99.8% server uptime. The implementation is shown to improve operational efficiency and customer-service quality at PT Tristan Digital Teknologi.
Copyrights © 2026