AT&T Turns to Autonomous AI Agents to Slash Fraud and Customer Wait Times

The telecom giant’s early efforts include a network-integrated “digital receptionist” that engages unknown callers in real time to identify spam and fraud, disconnect suspicious interactions or take messages. Behind the scenes, AT&T is also building out agent-driven tools with Microsoft, including “Ask AT&T” and an builder that automates back-office workflows and synchronizes data across systems to reduce wait times and manual handoffs.

Digital Receptionist: Fighting Fraud and Spam in Real Time

In mid-2025, AT&T began testing an AI-powered digital receptionist, AT&T ActiveArmor, built into its core network to help customers screen unwanted calls and cut through the noise of spam and fraud calls that have become increasingly sophisticated. Unlike traditional spam filters that merely flag calls, this agentic artificial intelligence engages directly with unknown callers using advanced voice-to-voice and natural language capabilities to determine legitimacy. If the caller meets predefined criteria, the call is connected; if not, the system can disconnect the line or take a message. During the interaction, the customer can watch a live transcript of the conversation and choose to intervene at any time.

This agentic approach builds on AT&T’s longstanding network-level efforts against robocalls and fraud. The company already blocks or labels billions of unwanted calls monthly, but elevates the capability by incorporating autonomous decision-making and conversational interaction rather than simple pattern matching.

Early trials suggest the tool can reduce the disruptive impact of unwanted calls, improve overall user experience and offer a new layer of fraud defense without requiring customers to install additional apps or hardware.

Agent-Driven Tools for Internal Efficiency

While the digital receptionist addresses a direct customer pain point, AT&T’s agentic AI strategy also extends into internal workflows and employee tools. Working with Microsoft and leveraging Azure, the company introduced “Ask AT&T,” a generative AI platform tailored to employee use cases, from answering complex HR and policy questions to aiding in software development tasks.

In late 2025, AT&T further expanded this with Ask AT&T Workflows, a graphical drag-and-drop agent builder that lets business teams create custom autonomous agents to automate repetitive tasks that once required time-intensive manual work.

The first production agents built with this system handle customer service update requests by automatically synchronizing data across backend systems and pushing real-time updates without human handoffs.

According to a CIO report, this has already improved customer satisfaction and reduced service wait times, freeing employees to focus on higher-value activities. For network operations, agents can now assist engineers by correlating telemetry, identifying root causes for alerts, opening trouble tickets and even proposing code patches, all while a human engineer oversees and approves outcomes.

Governance, Oversight and the Path Forward

AT&T has institutionalized its agentic AI push through a Generative AI Transformation Office, which funnels potential use cases through a centralized review process that evaluates business value, customer impact and risk, per CIO. Each initiative must pass muster with the CFO and tie into formal business cases, emphasizing measurable ROI and customer benefit before deployment.

Human oversight and audit logging are baked into every stage of the agentic AI lifecycle. Actions taken by agents are logged with data isolation, retention policies and role-based access enforced to maintain security and compliance. Human checkpoints are required throughout complex workflows, ensuring that autonomous executions do not overstep oversight bounds.