AT&T has shone a light on its agentic AI activities, including network engineering use cases.

In an update on its agentic plans as laid out last year, AT&T’s chief data officer Andy Markus highlighted Ask AT&T Workflows, a drag-and-drop-based tool that allows users to graphically build custom AI agents for the automation of time-consuming tasks.

Markus cited the use case of network engineers using AI agents to automatically diagnose, handle, and fix network issues, as well as generating tickets and incident reports.

These actions are spread out across three agents, with one correlating telemetry to pinpoint an alert, gathering recent change logs, checking for existing issues, before opening a ticket.

A second agent then suggests a path resolution by writing new code, while a third compiles artifacts and generates a summary for the engineer, who stays 'in the loop' at all times.

“A human always oversees the “chain reaction” of agents (and has required human checkpoints throughout the process) … Every action the agents take is logged, and when one agent passes off a workload to the next, data isolation, retention policies and role-based access are enforced, keeping data privacy and security top of mind along the full process,” Markus explained.

The telecom giant also touted two recent AI industry benchmarks to its name. Its Ask Data with Relational Knowledge Graph received plaudits on the Spider 2.0 text to SQL accuracy leaderboard, while its fine-tuned large language model (LLM) based on Google’s Gemma, Gemma-3-4B-IT, came out on top for root-cause analysis (RCA) tasks in 5G networks.

Recent months in telecoms have seen O2 Telefónica Germany launch an agentic AI solution in partnership with Tech Mahindra and Nvidia, while a Nokia AI agent solution was released this summer to help telecom operators manage their networks.

In a recent blog, Nvidia developers gave a deeper look at the chip giant’s new reference architecture in support of AI agent-centric 6G networks.

The framework pushes data-plane intelligence from centralized data centers to the network edge, helping provide instantaneous response times for agentic AI models by preventing transport delays in data between radio sites and distant core locations.