AT&T Bets on AI Agents and 500 Workflows

AT&T is deploying autonomous agents at scale with supervision and precision criteria, reporting fivefold returns on its production projects. The operator combines governance, data consolidation, and centralized logic to transform its processes, with a plan to expand workflows starting next year.

Thresholds of 90 to 95% and Announced Fivefold ROI

Each production workflow begins with human supervision and only operates autonomously after achieving an accuracy between 90% and 95% within an acceptable risk profile. Performance is continuously evaluated to detect any potential drift. Success criteria focus on reducing labor needs, generating revenue, operating expenses, and customer satisfaction levels. Luke Corbin notes that investments in production AI yield five times greater returns over the year.

Centralized Governance and Controlled Use of Agents

Workflows that may affect customers go through the Business Agent Foundry and undergo security, privacy, and compliance reviews, while employees can create agents for their personal productivity. According to Luke Corbin, the use of agents is democratized for individual efficiency but cannot drive workflows with material impact without oversight. Established in early 2025, the Business Agent Foundry maps processes with industrial frameworks and aims to eliminate unnecessary steps, transfers, and dependencies on legacy systems. Corbin believes that in the long term, transformation responsibilities should be distributed throughout the organization to the point where the dedicated team is no longer necessary.

Lean Architecture in Agents and Context-Stored Business Logic

At AT&T, the architecture relies on a small number of agents tasked with executing workflows, with business logic centralized and retained in context or memory. Some agents are responsible for retrieving relevant knowledge, while others, dedicated to planning, select the required actions, tools, and data. Developing these capabilities required consolidating scattered data and ensuring stable connections between data centers, clouds, and mobile teams, in an environment characterized by numerous mergers and acquisitions.

Reinvented Processes: From Mainframe to Halved Provisioning

AT&T has automated interactions with its mainframes by eliminating tasks previously handled manually, which has reduced lead times and accelerated service delivery. Luke Corbin emphasizes the importance of integrating recent technologies with legacy systems. In another use case, the time to provision dedicated Internet connections has been halved in buildings already equipped with fiber. In prospecting, agents validate locations and fiber availability to target promising customers before representatives intervene.

Scaling Up and Ambition for Agentic Workflows

Nearly 500 agentic workflows are in production, and AT&T plans to triple that number next year. According to Luke Corbin, the company processes about 45 billion AI tokens per day, and its platform manages over 1 million transactions per month. Many workflows already operate autonomously and cover tasks such as customer provisioning and sales prospecting. The approach prioritizes the reinvention of complete processes, following a rapid transition from experimentation to systematic identification of redesign opportunities. AT&T is now directing its efforts toward launching services, entering new markets, and supporting new sales channels. A 151-year-old company supported by costly legacy systems, AT&T believes that simpler generative workflows can offset some of these costs. Employee involvement in proposing ideas has helped mitigate resistance related to replacement fears and has been facilitated by the ability to eliminate inefficient processes that many were familiar with but did not appreciate. Luke Corbin presented this approach at an AWS event in Boston. This initiative is part of a broader effort to rethink the company's operations through autonomous agents.