HÀ NỘI — GoTyme, a multi-country digital banking group, has developed a generative AI-powered Root Cause Analysis (RCA) solution on Amazon Web Services (AWS), cutting the time needed to identify the cause of incidents from a median of 34.5 minutes to 10 minutes.

Developed by GoTymeX, the group’s Product and Technology Development Hub, the solution was tested using incident samples collected over six months. It is designed to reduce customer disruption, infrastructure costs and alert fatigue across GoTyme’s banking platform.

“RCA now starts investigating the moment an alarm is triggered, and incidents are diagnosed within 10 minutes," said Hiếu Tạ, director of Engineering at GoTymeX.

"That means fewer disruptions for our customers, less escalations across teams, and more time for our engineers to add value by building, not searching.”

The new solution, powered by AWS DevOps Agent, automatically investigates incidents when an alert is triggered. It correlates metrics, logs, traces, deployments, infrastructure health, network telemetry and application performance to identify the most likely root cause and provide supporting evidence.

GoTyme said the solution has helped it achieve 3.5 times faster service recovery while reducing operational overhead and customer impact.

All recommendations remain subject to human review. Engineers validate and approve actions before any changes are made to production systems.

“GoTymeX exemplifies this shift. Their engineers built a root cause analysis solution that cut incident resolution times dramatically, directly protecting millions of customers,” said Eric Yeo, Country General Manager for Việt Nam at AWS.

Since 2016, GoTymeX has grown to more than 900 engineers, designers and technology specialists in Việt Nam, developing digital banking platforms serving customers across multiple markets.

The company said the RCA solution is part of its broader effort to develop AI-assisted operations capable of improving reliability and customer experience across complex, distributed systems. — VNS