AI Agents and the Future of the Contact Centre in Indonesia
Brands are racing to put AI agents on the front line of customer service.The question is no longer whether, but when a customer should be handed over to a human.
AI-assisted translation from Bahasa Indonesia, reviewed by Raka Mahendra, Managing Editor on 6 October 2026. The Indonesian original prevails in case of discrepancy.
Three years ago, an AI agent in Indonesian customer service still meant a keyword-based chatbot that frustrated customers all too easily. Today, a number of major banks and telecoms operators have deployed large language model-based assistants as the first layer of service, answering more than half of all inbound interactions.
This shift is not simply about cost efficiency. According to several CX leaders interviewed by CXinsight, the pressure comes from customer expectations shaped by using AI assistants in other apps — fast, personal and available around the clock.
From answering questions to resolving problems
Early-generation chatbots were designed to reduce ticket volumes by answering repetitive questions. The new generation of AI agents is designed to actually complete transactions: changing a card limit, lodging a complaint or rescheduling a delivery — without escalating to a human.
The measure of an AI agent's success is not how many sessions it handles, but how many problems are genuinely resolved without the customer having to repeat their story to someone else.
— A CX practitioner at a digital bank, interviewed on condition of anonymity
62%
of customer service interactions at five major Indonesian digital banks are now handled entirely by AI agents without escalation, up from 24% in 2024.
Source: CXinsight estimate based on interviews with five digital bank CX teams, September 2026
The risks few are talking about
The pace of adoption leaves a question that is rarely answered openly: what happens when an AI agent misreads a customer's emotional context? Some brands have begun introducing a "fast lane" to a human as soon as the system detects repeated frustration, but implementation remains uneven.
CXinsight Angle
AI agents are moving from simply answering questions to completing transactions in full. The evidence lies in the sharp rise in resolution rates without escalation at digital banks. For other brands, this means the bar for customer service speed is quietly rising — and customers will carry that new expectation into every subsequent interaction, including with brands that are not yet ready.
CXinsight will continue to monitor how regulators and consumer associations respond to this trend, including on transparency about when customers are talking to a machine.