Verizon is drawing a clear line between automation and autonomy. The operator wants AI systems that reason, coordinate, and act across domains. This marks a shift from scripted network tasks to agentic operations.
Today, many network tools follow predefined rules. They adjust settings when known events occur. Verizon says that model still matters. Yet it cannot handle every unknown network condition.
At the Intelligent RAN Forum, Anil Guntupalli explained the company’s direction. “Automation is about what we know, and scripting around it,” he said. “Autonomy is about what we don’t know, and reasoning [around it] – and [then] working autonomously within the network and the domain.”
That distinction matters for modern mobile systems. A radio issue may come from transport, backhaul, or software behavior. A single AI agent may miss that wider picture. Verizon wants agents to share context before taking action.
Guntupalli summarized the challenge clearly. “It has thousands of KPIs, and dependencies on transport [and] everything else in the environment. So you have to not only predict, but you have to act dynamically in real time… This is where the agentic world kicks in.”
The opportunity is significant. Verizon already uses closed-loop automation at large scale. In 2025, its platforms processed more than 70 million configuration changes. That saved many technician hours and reduced manual effort.
However, the company does not want uncontrolled AI inside the network. Engineers still define intent, limits, and desired outcomes. Agents may reset nodes or handle sleepy cells. Major incidents still require human authority.
This approach reflects a core telecom reality. Networks carry critical services and customer traffic. A smart action in one domain can harm another. For example, traffic rerouting can overload transport capacity.
Therefore, Verizon wants to own the intelligence layer above vendor tools. Vendors can provide RAN controllers, rApps, and AI models. But they do not know Verizon’s full topology. They also lack its operational history.
Guntupalli said traceability will become essential. Verizon wants to know which agent changed what. It also wants the data, state, and context behind each action. That record can improve training and future decisions.
Standardization now becomes a key industry question. Open RAN works with defined interfaces and predictable behavior. Agentic AI behaves differently because context changes decisions. “The short answer [it is] absolutely needs standardization,” Guntupalli said.
Verizon also links this shift to future traffic patterns. Devices, drones, sensors, and robots will send more upstream data. “Physical AI is about to hit us; all these things need a symmetrical uplink capability,” he said.
The company does not plan to wait for 6G. It sees major advances starting with 5G, Open RAN, sensing, and general-purpose compute. Agentic AI may not replace engineers. Instead, it could help them solve harder network problems faster.
