U Mobile uses Amazon Bedrock AgentCore to automate telco operations
U Mobile has built an AI-driven helpdesk on Amazon Bedrock AgentCore that coordinates investigations across telecom systems, supporting more than 150 documented operational scenarios. According to AWS, the Malaysian 5G provider's internal measurements show trouble ticket volumes falling by up to 84.5 percent in key categories.
- AiDaH covers more than 150 documented operational scenarios.
- U Mobile says trouble ticket volumes fell by up to 84.5 percent in key categories.
- The platform uses eight specialist agents across telecom and business operations.
- AiDaH went live to 50 L2 call center and branch users in Q3 2026.
- The system keeps human confirmation before changes to live subscriber accounts.
From routine lookups to coordinated investigations
U Mobile developed AiDaH, short for AI-Driven Agentic Helpdesk, with AWS and AWS Partner Axrail. The platform addresses a problem created by the operator's hybrid IT environment, which includes business support and customer relationship management systems, mobile number portability platforms, big data analytics and infrastructure monitoring tools.
Before AiDaH, diagnosing an issue could require an agent to inspect interface logs, query subscriber eligibility through business support system APIs and search internal knowledge bases for relevant policies. These steps took place across separate systems and portals. If the issue remained unresolved, it was escalated to backend IT teams.
A traditional AI assistant could help draft a response or suggest troubleshooting steps, but would not navigate those systems or carry out the analysis. AiDaH is designed to do so. For a case involving mobile data that does not work overseas, for example, it can check roaming configuration, verify international data eligibility, identify the failure point and assemble a remediation plan. The support agent reviews and confirms actions before they are applied to a live subscriber account.
A six-layer platform built for multiple divisions
AiDaH runs in the Asia Pacific (Singapore) Region and gives staff access through Microsoft Teams, alongside a web portal and a REST API for programmatic triggers. Its architecture has six layers covering the staff interface, orchestration, specialist agents, gateways and backend APIs, data and knowledge, and operational systems.
The orchestration layer performs pre-flight security checks for role-based access control, identity validation and guardrails. AWS says roughly 70 percent of requests are resolved through deterministic keyword matching without model invocation, 20 percent are classified by a lightweight model, and 10 percent require the full orchestrator.
Eight specialist agents cover areas including customer relationship management, customer problem management, number portability, order and SIM operations, infrastructure monitoring, big data analytics, cloud cost management and finance operations. They communicate through the Agent2Agent protocol, enabling them to delegate subtasks and complete workflows across systems.
Amazon Bedrock AgentCore Gateway exposes U Mobile's backend systems through a unified Model Context Protocol interface. Amazon Bedrock Knowledge Bases provides governed content for responses, while AgentCore memory maintains conversational context across turns. Amazon DynamoDB stores the six-tier role-based access control model and session metadata. Connections to U Mobile's on-premises and cloud systems use AWS Transit Gateway and an AWS Site-to-Site VPN with no public internet exposure.
From pilot to production deployment
U Mobile began conceptualizing AiDaH in late 2025 with two pilot agents and Amazon Bedrock Knowledge Bases. A Q1 2026 pilot brought the internal IT helpdesk online with 4 agents and 20 operational scenarios. During 2026, the platform added three-tier routing and reached 6 agents by mid-year.
AiDaH went live to 50 L2 call center and branch users in Q3 2026 with 8 agents. It now covers more than 150 documented operational scenarios across six business divisions, according to AWS.
The deployment shows how U Mobile is applying agentic AI to operational work rather than limiting it to response drafting. The source describes a production case where AiDaH traced a failed data add-on purchase to the error code “Subscriber has pending order.” It also found that the subscriber had tried the same purchase more than 30 times across multiple channels over three days, and presented the root cause and remediation steps in one response.
For users, the approach is intended to reduce manual investigation and escalation while retaining human approval before changes are made to subscriber accounts. For U Mobile, Amazon Bedrock AgentCore provides a managed runtime and built-in memory, allowing the operator to expand the system across operational domains without operating separate agent infrastructure or state stores.
