Understanding AI in Telco

AI is here, so what’s coming next? The evolution of Agentic AI may be the next big game changer for telcos.

Over the past year, we’ve been writing about AI with increasing frequency. And with good reason, since the subject has been and remains on everyone’s lips. To date, we’ve considered the impact of the technology on the telco sector (you can read that blog here), investigated how AI might affect your future (read here), and we’ve given a broader overview of the subject that you can read here. Now, let’s turn our attention to perhaps the next influential leap forward; the shift from “traditional” AI to Agentic AI.

Agentic AI explained

Agentic AI systems differ from traditional ones in that have agency. That means that they can autonomously make decisions, take actions toward goals, and interact n a purposeful way. The traditional AI systems that you’re likely most familiar with are generally static or reactive in how they operate, only doing what they are directly programmed or asked to do. Agentic AI systems, on the other hand, are different, as a number of features underline.

Key Features

Looking at those features, the difference is immediately clear. Agentic AI systems are hallmarked by:

• Goal-directed behaviour: Agentic systems can pursue objectives, and can have the ability to plan and adjust to information as needed.
• Autonomy: These systems can operate independently. They don’t need constant human supervision.
• Persistence: Systems work in a “cumulative” way, over time, toward a task, even when it involves multiple steps or adapting to new information.
• Reasoning and decision-making: Systems can choose between different options, prioritize tasks, and solve problems.
• Interaction with tools or environments: Many agentic systems can use software tools, APIs, or even manipulate virtual or real-world environments to complete tasks.

AI in telco

Agentic AI represents the shift from AI being a passive tool to an actual collaborator in carrying out complex tasks. In a sense, it’s AI made “smart”. What impact will it have on telco, though? To find out, let’s consider one possible use case: A telco using an agentic AI system to autonomously monitor, diagnose, and optimize network performance across thousands of cell towers and network nodes. How would an Agentic AI system help?

1. The agentic AI could be purposed with the goal of maintaining optimal Quality of Service (QoS) and minimizing downtime across the network. As we’ve seen, it will then work autonomously towards that goal.
2. Rather than just reacting to predefined triggers (like traditional monitoring systems), the AI will continuously scan real-time data from network nodes (e.g. traffic loads, latency, packet loss, signal strength). If an anomaly is detected—for instance, a cell tower that’s showing signs of congestion or interference—the Agentic AI doesn’t just flag it; it automatically starts diagnosing the issue.
3. Say it identifies a degraded signal-to-noise ratio in a specific location. The AI will automatically check neighbouring towers for handover performance and load balancing metrics. It will then simulate possible solutions—e.g., reallocating spectrum, adjusting antenna tilt, offloading users to adjacent cells – and pick the best one based on trade-offs (minimizing disruption, power usage, compliance with regulations).
4. It will execute the chosen reconfiguration directly via APIs into the network management system and then monitor post-action effects. If the implemented solution didn’t work as expected, it may roll back or try another strategy.

In conclusion, the Agentic AI agent will log actions and outcomes, learning over time which fixes work best in which contexts. Eventually less human intervention will be needed for routine optimizations.

The benefits to the operator are numerous. Faster response times to network issues (from hours to minutes or seconds). Lower operating costs (reduced need for manual network engineering intervention). Improved customer experience (fewer dropped calls, faster data). And scalable management of 5G or edge infrastructure without linear increases in headcount.

In conclusion, while end-to-end agentic AI is still emerging, telecom companies including Vodafone, Ericsson, and AT&T are already implementing elements of it in AI-driven network automation platforms. The direction of travel is key telco systems reaching a point where they can act independently.