AT&T uses AI to cut network downtime

AT&T has developed an AI-enabled end-to-end incident management system to keep its 145 million wireless and 16 million broadband customers connected. The company began building the platform in 2017 to centralize how it identifies network interruptions, diagnoses causes and coordinates responses before customers feel the impact.

The system, known as EEIM, draws on technologies from MongoDB, Snowflake, Microsoft Azure and Databricks. AT&T reorganized 10 petabytes of data, including network logs, alarms, dispatch tickets and outage details, so the platform could identify problems, predict disruptions and route fixes to the right teams. A related technician app, Atlas, uses machine learning and AI models to recommend repair plans in the field.

AT&T later added generative AI in the first quarter of 2022 and AI agents in the first quarter of 2025. The agents can gather outage details from customers, take steps to resolve issues and pass case information to technicians when field work is needed. Markus said 100,000 employees have access to AT&T’s generative AI tools, and the company consumes over 27 billion tokens per day.

The EEIM has helped AT&T prevent 3.1 million unnecessary field dispatches and reduced customer downtime by more than 12 million hours over the last year, according to Markus. The company has also built more than 30 AI models to forecast configuration issues, weather-related problems, system failures and other potential disruptions.