Table of Contents
Telecom is entering a new era where supervising AI agents may be the most important skill to have, says Markus Nispel, Extreme Networks
As adoption of AI-driven automation accelerates across core and radio access networks (RAN), telcom engineers are finding themselves in the middle of a tricky transition. They are going from managing networks to managing AI agents.
“That’s the new skillset that everybody needs to learn,” said Markus Nispel, chief technology officer, EMEA, at Extreme Networks.
Speaking on Pulse this week, Nispel said that the excitement around prompt engineering has cooled off as agentic AI entered the scene sparking new conversations around AI orchestration . “Every experienced engineer becomes an orchestrator and manager of agents, and like you manage people, you need to manage your agents,” he said.
And what does supervising agents entail? According to Nispel, an important starting point is setting boundaries to ensure that the agents act exactly as they should — an essential step to building trust in AI.
“We always say that adoption comes with trust. As trust increases, adoption increases. And as this becomes a flywheel, we see that users are more and more willing to give up control,” he noted.
Another essential piece is context and intent. Much like human employees, AI agents must be fed the right contexts and clear instructions so that they can successfully accomplish the tasks assigned to them.
“Managing agents or a field of agents is like managing a team. You need to be clear on instructions, how you define goals, and [provide] enough context so that when they get stuck, they come back to you and ask for clarification and not go off and do something that they think is the right thing to do,” Nispel argued.
This, he said, will be the “ultimate skillset” for maximizing engineers’ own capabilities and outputs using AI and agentic technologies.
Blending curiosity with experience
Nispel added that AI-driven autonomy is most likely to deliver successful outcomes when it happens at the intersection of people, processes, and technology. This prompts a reset of the current telecom operating model which happens to be fragmented and disconnected.
He noted that, at present, a significant adoption gap exists in the employee base with younger employees showing greater enthusiasm to embrace AI, while senior-level personnels remain cautious. A poll conducted by the International Workplace Group confirms this. They found that Gen Z employees are quicker to adopt AI and are now even helping their older colleagues do the same.
“People need to become comfortable,” Nispel said, but in order to get there, he said, there needs to be higher levels of AI literacy among the staff to understand what’s possible with AI. He urged operators and enterprises to find the pain points and frictions that engineers face every day and design AI solutions around those use cases to encourage adoption. The important thing is to bridge the gap between curiosity and execution with experience.
At the same time, Nispel said that customers too need to rethink the workflows and processes at their end and bring those that are outdated for the agentic world in sync with the rest of the pipeline.
As operators continue advancing toward higher levels of network autonomy, these adjustments will be critical to achieving more controlled and effective use of agentic AI, while uprooting the culture of resistance from within the organization.
“It will take us as a market and as a society a couple of years to really figure it out, but it’s certainly starting. People should not wait for [it] to happen. Just start using [AI], experiment with it, get familiar, and that’s the best guidance that we can give to everybody across markets and technologies” he said.
