Beyond the hype: How telcos can win with agentic AI

Telcos are perfectly positioned to reap the benefits of agentic AI, but the possibilities of this new technology can be overwhelming. Here's how to find and focus on what moves the needle.

Agentic AI is opening doors for telcos to streamline both customer-facing and internal operations, improve CX, and enable smarter, faster decisions. It’s all good news, but the potential applications of agentic AI are so numerous and varied—touching everything from network optimization to service personalization—that it’s easy for leaders to feel daunted by the sheer range of possibilities.

Help is at hand. Telcos can cut through the complexity by identifying the most suitable internal processes for agentic AI, centralizing agent development, and establishing the right data and API frameworks. Below, we outline three practical steps that—whether used singly or in tandem—can help telco leaders speed adoption and drive meaningful business results. 

Why telcos are fertile ground for agentic AI

Telecom operations have long relied on a mix of structured, rule-based systems and human-driven oversight—precisely the conditions in which agentic AI thrives. Intelligent agents perform best when they’re given clear rules to follow, but also the flexibility to seek human input when needed.

That makes them a natural fit for telcos. Agents excel at orchestrating complex, cross-functional tasks, and can quickly add value by connecting siloed functions like network operations, billing, and customer service. They can even help generate a unified customer view by gathering and integrating data currently scattered across OSS, BSS, CRMs and ERP systems.

3 steps to accelerate agentic AI for telcos

For many telcos, the biggest hurdle is knowing where to begin. With agentic AI able to support everything from customer service to field operations, it’s easy to get lost in the forest of possibilities. 

These three practical steps can help telcos discern the path ahead—and take the first steps forward.

1.    Identify the use cases that genuinely add value

Not every telco process is suited to agentic AI. You wouldn’t want AI agents crafting enterprise strategy, for example, or carrying out sensitive HR actions such as layoffs. Those processes require intuition, creativity, holistic reasoning and emotional intelligence, which all still lie beyond the agentic wheelhouse.

Where agents shine is in those processes that combine clear structure with human decision-making. Billing is a prime example. Its workflows follow standardized, rule-based sequences spanning ERP, CRM and payment systems —all aimed at ensuring accurate payments and preserving customer trust. Yet billing also involves tackling disputes or missing information that call for contextual awareness of the “big picture.”

Those attributes put billing in the sweet spot for agents. Imagine a reconciliation agent that could analyze logs, payment histories and plan details to flag anomalies—such as overbilling due to a missed discount. With the ability to “reason” and act, the agent could also recommend adjustments or even initiate a refund, helping resolve issues faster while preserving customer trust.

Quantifiable impact is another key marker of a strong use case. Agents deliver the most value in processes that have standard measures of success. Billing fits this profile with KPIs like accuracy rates and day sales outstanding (DSOs). Field operations is another good fit for the same reason: metrics, such as mean time to repair, make it easy to track agent-driven improvements and efficiencies.

2.    Centralize development of AI agents

Business units are often first to experiment with new tools, especially when they can see clear, immediate value that doesn't require deep IT integration. CRM platforms, videoconferencing, and social media management tools like Hootsuite all took root and proliferated in this way. But while early enthusiasm among business users can help seed adoption, a decentralized approach to agentic AI can quickly become chaotic, leading to duplicated efforts and inconsistent results.

A centralized, modular strategy, on the other hand, delivers order and scalability that benefits everyone: By creating agentic AI modules and maintaining central agent repositories, telcos can accelerate development, reduce redundancy, and enable reuse across functions. Take a knowledge retrieval agent: initially developed for field operations, the same core agent can be adapted for deployment in customer service. In both cases, the agent’s job is to scour datasets for details. Each use case leverages different data sources and generates different outputs, but the intent and functionality are the same: faster, better resolution.

This plug-and-play approach not only speeds up deployment, it also reinforces step one: allocating resources by the speed and probability with which each use case’s value will be realized.

3.    Build strong data and API foundations

Agentic AI is only as good as the data it can access. To operate intelligently and effectively within a telco environment, agents need to connect—securely and seamlessly—to the systems, applications and data sources that drive business operations.

Consider the example of agentic AI for network capacity planning and optimization. An agent built for this task is only useful if it can access real-time information from network monitoring systems. Without API access to usage and performance metrics, the agent can’t adjust network resource allocation—in real time or near-real time—to balance supply and demand. 

Robust data pipelines and well-structured APIs are essential. Telcos that invest in these foundations, which let agents “connect the dots” across a large business ecosystem, will be best positioned to capture the full value of agentic AI.  

Agentic AI isn’t just a new tool. It’s going to reshape the future of work, requiring a comprehensive strategy to manage the upheaval. Telcos will need to leverage their existing partnerships, and forge new ones—with advisory firms, technology providers, system integrators, and academic institutions—if they’re to keep pace with innovation, and agentic AI’s emerging capabilities.

The journey is just beginning. But those who act early, and with clarity, will be best positioned to lead in an increasingly AI-native world.