TLDR: AT&T is significantly expanding its integration of agentic AI capabilities throughout its entire business, empowering employees to build custom AI agents using an updated internal tool called ‘Ask AT&T Workflows’. This strategic move aims to automate complex tasks, improve customer service, and drive innovation by shifting AI from an ‘information economy’ to an ‘action economy’.
AT&T is making substantial advancements in its artificial intelligence journey, particularly with the widespread adoption of ‘agentic AI’ across its business operations. Andy Markus, Chief Data Officer at AT&T, highlighted the company’s focus on autonomous assistants, emphasizing their role in the next wave of AI innovation. These AI agents are designed to move beyond generative AI’s content creation capabilities, instead planning and executing multi-step tasks from inception to completion, with human oversight when necessary. This involves goal understanding, multi-step planning, and tool orchestration within a single package.
Building on its existing efforts to leverage AI for customers, AT&T is now extending this ‘agentic power’ to its employees. A key development is the enhancement of the internal ‘Ask AT&T’ tool, which now includes ‘Ask AT&T Workflows’. This new feature provides teams with a graphical drag-and-drop agent builder, enabling them to create customized AI agents. The goal is to automate time-consuming tasks, thereby freeing up employees to concentrate on strategic customer service, complex problem-solving, and future planning.
The impact of this initiative is already evident. The first in-production tool developed using Ask AT&T Workflows is demonstrating significant value. This tool, a collaborative effort between AT&T Business and technology development teams, utilizes AI agents to manage customer service update requests. These agents can synchronize data across various systems and automatically install information in real-time, streamlining processes and enhancing efficiency.
Markus provided a hypothetical example of an AI agent’s potential in network operations: an agent could identify a network issue, another could propose and even write code for a resolution, and a third could compile a summary and artifacts for preventative measures. This illustrates the transformative potential of these tools, allowing employees to convert natural language prompts into artificial assistants capable of executing detailed, step-by-step instructions.
AT&T’s commitment to AI is further underscored by its recent achievements in industry benchmarks, including the Spider 2.0 text-to-SQL accuracy leaderboard and the GSMA Open-Telco LLM leaderboard. These accolades demonstrate AT&T’s proficiency not only in utilizing AI technologies but also in developing them. The company views AI agents as crucial for fulfilling the promise of AI for businesses, transitioning from an information-based economy to an action-based one.
This enterprise-wide integration of agentic AI is a testament to AT&T’s long-standing history of technological innovation, dating back to Bell Labs. The company aims to reduce costs by $2 billion over the next few years, with generative AI playing a significant role in achieving this goal. AT&T’s strategic investment in AI has already yielded a 2X return on investment for every dollar invested in the same year, impacting free cash flow from multi-year business cases.
Also Read:
- Next-Generation AI Agents and Co-pilots Poised to Revolutionize Devices and Enterprise Operations
- Cisco Revolutionizes Customer Experience with Pervasive Agentic AI Integration
Beyond internal applications, AT&T has also been testing customer-facing agentic AI tools, such as a digital receptionist designed to screen and manage incoming calls, reducing unwanted interruptions and improving customer experience. This network-based solution uses multiple large language models (LLMs) and advanced voice-to-voice technology to identify spam and fraud calls, determine call urgency, and even take messages autonomously. This digital receptionist is currently being rolled out to select customers, with plans for broader access as testing continues.
