Telcos want more than efficiency from agentic AI

retained text

Featured Story
HPE claims to have beaten Cisco to a proper self-driving network
Marvis, described as an AI brain, now unifies all the various parts of the Aruba and Mist portfolios to create a truly autonomous system, says HPE.
Orange, Telenor and Vodafone explored how they plan to turn agentic AI into measurable operational and customer outcomes during a recent webinar.
Root-cause analysis as well as incident coordination and detection emerged as key use cases for applying AI agents and automation in network operations during a recent panel featuring executives from telcos Orange, Telenor and Vodafone as well as software provider Polystar.
Thomas Nilsson, chief product officer at Polystar, part of Elisa Industriq, called root-cause analysis a "key area," noting that agents can help identify an issue in minutes rather than the hours it may take a person.
Afnan Ahmed, director, technology strategy and architecture at Telenor, said he tended to agree with the "gold medal from [Nilsson] on cause analysis. It's probably something that would help a lot in the near term."
However, Ahmed said operators should also think beyond the immediate use cases.
Indeed, the general view of the panelists taking part in a FutureNet webinar titled "From AI pilots to business impact: What will define the next generation of telecom operations?" was that technical efficiency alone is not enough: AI and network automation should ultimately improve customer and business outcomes. (See Agentic AI has already seeped into network operations, but trust remains critical – study.)
As noted by Alexis Koalla, director operations strategy and transformation and autonomous network at Orange, operators need to connect technical measures to business results, such as customer experience, revenue growth, NPS, reduced churn, and customer trust.
"The key challenge is how to translate these technical KPIs into business outcomes," Koalla said.
Ahmed stressed that efficiency matters, but treating autonomous networks as "just an efficiency game" would miss the bigger opportunity. More programmable, open networks could help telcos defend their connectivity business and enable future offerings, including security and smart-city services, he said. (See AT&T braces for the agentic AI wave.)
Mabel Pous Fenollar, global director of digital and zero touch operations at Vodafone, also said ensuring that "the experience to the customer is better is super essential to us."
"At the end of the day," said Nilsson, "it's about the customer. It's the customer experience you want to manage and bring up to spend more with you and stay with you."
Meanwhile, the panelists identified other use cases including employing agents to coordinate teams to rectify breaks in fiber networks and then following up to help meet the terms of service-level agreements.
According to Fenollar, "when you have a fiber break and you need to coordinate and ring different teams, agents play a pivotal role there."
Koalla also pointed to agent-based detection as another promising operational use case. "Detection with a good level of accuracy from agents could be a very good thing for network operation," he said.
In addition, for change management preparation, he said agents could review and organize large configuration files, classify parameters and recommend changes, while a human retains responsibility for sensitive network execution.
"This could be something where we can use agents in change management processes tomorrow," Koalla said.
Overall, the panel noted that telcos have operational experience and are at an early stage in automation, but are still working through data, organizational and governance challenges before AI-enabled operations can scale. (See Omdia survey results: The rise of agentic AI in transport networks.)
"We are moving from pure network operations engineers to what we call network automation engineers in the operational side of things," Fenollar said.
Scaling the various use cases, the panelists cautioned, will require much more than new technology. Better data quality, new skills, clear governance and changes to operating models and company culture will all be needed.
Ahmed said the industry does not yet know exactly what products customers will need in the future, but warned that today's operating model will not be enough.
"So whether you want to call it the introduction of agents and agentic AI, whether you want to call it autonomous networks, whether you want to give it some other name, all these things need to come together. Otherwise, I believe we as telcos will not only lose our right to take the space in the connectivity business, but we'll also lose the right to bring the next generation of products to the customers. Somebody else will do it," he said.

document fetched 2026-10-29 03:02 · canonical https://lightreading.com/ai-machine-learning/telcos-want-more-than-efficiency-from-agentic-ai