How AT&T Uses Azure OpenAI to Deploy 71 GenAI Solutions at Enterprise Scale

AT&T serves tens of millions of customers and employs thousands of engineers and developers, creating a scale challenge that no manual process could solve. Employees were spending too much time searching for answers across fragmented systems, developers were debugging code line by line, and data was flowing faster than human capacity could keep pace. The imbalance between data velocity and human ability to act became the core problem the company needed AI to solve.

Before the AI platform was in place, AT&T ran isolated pilots and experiments. While early results showed promise, the company recognized that proof-of-concept demos were not the finish line. AT&T needed AI that could move from controlled experiments to production-grade, enterprise-wide operations—handling sensitive data in a regulated, compliance-heavy environment without sacrificing security or transparency.

In 2023, AT&T launched Ask AT&T, a generative AI platform built on Microsoft Azure. The architecture was purpose-built for enterprise reliability: Azure Kubernetes Service orchestrates containerized agents, Azure API Management governs every interaction, and Azure Cosmos DB stores chat history and configurations that feed Azure AI Search. At the core is a multi-agent framework coordinated through Azure OpenAI, where specialized agents collaborate in near real time—one retrieving verified content, another summarizing customer context, and a third automatically updating call notes.

The results at enterprise scale are concrete. AT&T now processes approximately 9 billion tokens per day, has deployed 71 generative AI solutions, and has 100,000-plus employees actively using AI-powered agents daily. Customer care agents resolve issues 33% faster, and the time developers spend debugging code has dropped dramatically. The platform has delivered more than double year-over-year return on free cash flow through efficiency gains and automation.

The broader transformation is architectural: AT&T built a reusable, governed AI ecosystem where skills developed for one workflow can be redeployed to another, compressing months of development time. Every AI agent clears legal, security, and finance reviews before launch, and Azure Monitor with OpenTelemetry-compliant Arize AI provides full auditability. AT&T's approach has become a repeatable model for how large enterprises can operationalize generative AI while maintaining governance at scale.