Agentic AI dominates today’s technology conversation, but AI is not new to telecom. Telecom networks are too large, complex and data-intensive to operate without machine learning, analytics and automation. What’s changing is the capability of AI, and the urgency to apply it more broadly.
ChatGPT created an “iPhone moment” for AI, making the technology tangible to millions of users. For telecom operators, the bigger opportunity lies beyond the hype: automating operations, simplifying complex systems, and improving decision-making.
Automation is becoming essential
Operators are managing increasingly complex networks, often with fewer people. They still need to monitor services, identify problems and manage hundreds of daily network changes, without customers noticing.
AI can help detect issues before they affect customers, accelerate root-cause analysis and automate routine changes. But greater automation does not mean removing humans from the network.
Autonomy will happen incrementally
A fully autonomous network is an attractive vision: define the desired outcome and let the network manage itself. The reality is more nuanced. Operators run mission-critical services for millions of customers. When something goes wrong, they must be able to identify the cause and regain control quickly. Some network functions are deterministic enough for extensive automation. Others carry consequences too significant to remove human oversight.
Consider a cloud-native network. Automatically adding an instance when capacity reaches a defined threshold may be relatively low risk. Automatically terminating a network function because a system determines it is unhealthy is a totally different situation, particularly if the decision could have widespread consequences.
The level of automation must therefore match the risk. Operators can start with specific services and network domains, learn from those deployments and expand as confidence grows.
Progress requires calculated risk
Historically, Telecom has always prioritized reliability. But demanding zero risk from automation can become a roadblock towards fast progress and innovation. Human-led operations are not risk-free. Engineers can mistype commands, miss preconditions or make incorrect decisions. Automation introduces different risks, but eliminating every possibility of failure is unrealistic.
The challenge is to take calculated risks: automate where confidence is high, maintain safeguards where the stakes are greater and learn from each deployment.
Making complex technology easier to use
AI can also change how people interact with telecom systems. Many of the platforms currently deployed around the world were built by engineers for engineers. Today, business analysts, roaming managers, and other non-engineering users increasingly need access to the same systems and data without mastering the technical complexity underneath.
Natural-language interfaces can bridge that gap. Users can discover features, retrieve reports and interact with sophisticated platforms conversationally. AI can also support root-cause analysis automation, helping users understand what failed and what action to take next. The underlying technology remains complex. AI simply hides that complexity away from the end user.
From fraud alerts to explainable decisions
Fraud management and revenue assurance show how this can work in practice. Traditional systems flag suspicious activity when certain indicators or thresholds are triggered. An analyst then investigates and decides what to do. AI Agents can go one step further. They can provide a verdict, assemble the supporting evidence and recommend the next action.
High-confidence cases could eventually trigger automated actions. Ambiguous cases can remain with human analysts, who validate the recommendation and refine the underlying logic. The shift is from simply detecting a problem to explaining it and recommending what happens next, therefore arming the analyst with the tools to make better and faster decisions.
Creating value beyond internal efficiency
AI also creates opportunities for new customer services. Small and medium-sized businesses are a clear example. Unlike large enterprises, many lack the technical resources to build and manage AI services themselves.
Operators can help close that gap. A small business could deploy an AI agent through its telecom provider to handle customer enquiries or online interactions without building its own AI infrastructure or working directly with model APIs. The value is practical: a missed enquiry or delayed response can mean lost business. Making AI simple to adopt gives operators another way to create value beyond connectivity.
Using AI to rebuild trust
Scam and spam protection presents another opportunity. Fraudulent calls and messages are already pervasive, and generative AI is making them more convincing. At the same time, AI can strengthen the defense. Protection across voice, messaging, and data can help identify suspicious interactions before customers fall victim to them.
This could become more than an operational capability. If operators can help consumers trust the voice channel again, or better protect children and elderly family members, that protection could become a new revenue source.
From AI hype to practical value
The strongest AI opportunities in telecom are not necessarily futuristic. They address problems operators already face: network complexity, operational efficiency, difficult-to-use platforms, fraud and declining trust in communications.
Full network autonomy remains some distance away. It may not even be the right objective for every part of the network. What matters is applying automation where it delivers clear value while keeping human oversight where the consequences demand it. AI is already making that possible. The opportunity now is to move beyond experimentation and put it to work.
👉 Explore more in the original podcast here
Give us your comments
Let us know what you thought about this article