Malaysia's fifth-generation mobile network operator U Mobile has disclosed that its in-house AI ticketing system, AiDaH, now covers six business departments and more than 150 operational use cases, with critical incident tickets down 84.5% in some areas. The significance of these figures goes beyond improved customer service efficiency: enterprise AI is beginning to enter the core workflow of troubleshooting, invoking systems, and proposing fixes.
What this is
AiDaH is an AI-powered intelligent help desk developed by U Mobile with AWS and partner Axrail, built on Amazon Bedrock AgentCore. When an Agent—an AI program capable of autonomously invoking tools and completing multistep tasks—receives a ticket, it queries interface logs, verifies customer eligibility, searches internal policies, and assembles recommended actions, escalating complex issues to backend teams when necessary.
AgentCore runs these Agents and stores the state information, including memory, required for multistep conversations. U Mobile adopted a specification-driven approach: it first defines task rules and expected outcomes, then develops and tests the workflow. Unlike traditional automation, the system does more than execute against fixed conditions; it gathers information across multiple systems and determines its next action.
Industry view
What stands out is that U Mobile did not simply add headcount. It inserted AI into a workflow already marked by fragmented systems. If the ticket reduction proves sustainable, the benefits will extend beyond saving customer service time. They include shortening the impact of network outages, reducing repetitive investigation work for backend teams, and shifting employees' focus toward complex problems.
That said, the 84.5% figure is an internal company measurement and applies only to "critical categories." It cannot be extrapolated directly into an overall cost reduction. Telecom systems involve customer records and the core network, where the risk-reward profile is unusual: an incorrect judgment, unauthorized query, or inappropriate commitment to a customer can turn efficiency gains into compliance and trust costs. Agents also depend on the quality of logs, policies, and interface data. If underlying systems are not standardized, intelligence can amplify disorder rather than solve it.
Impact on regular people
For enterprise IT: The focus of enterprise AI procurement may shift from chat assistants to workflow execution, but data, access, and audit mechanisms must be governed first. Whether Agents can operate across systems will bear more directly on real business value than the quality of any single response.
For individual careers: Roles such as customer service and operations that have clear rules but require cross-system queries will be affected first. The emphasis of the work will move from gathering information to reviewing judgments, handling exceptions, and assuming final responsibility.
For the consumer market: Mobile users may receive faster answers to faults and more accurate service handling. Whether complex cases actually deliver shorter waiting times, however, still depends on whether backend staff, system interfaces, and accountability boundaries improve in step.