Agentic AI is moving from concept to execution in broadband networks, but its adoption is being shaped by operational realities rather than ambition. At Fiber Connect 2026, Nokia detailed how operators are focusing on a narrow set of high-impact use cases where AI can reduce manual effort, improve customer experience and deliver consistent outcomes. Troubleshooting and customer care have emerged as early priorities.
According to Nokia, broadband operators continue to spend significant time and resources managing trouble tickets and investigating faults across increasingly complex networks. “Operators are always looking for those high-value use cases where they are spending a lot of their time, energy and money,” said Sireesha Kora, head of fixed networks controller applications at Nokia. “One such application happens to be customer care.” Nokia is extending its Altiplano platform so its existing automation and AI capabilities can be consumed by agentic AI workflows that assist with fault isolation, root cause analysis and recommended repair actions. Kora described the approach as a virtual operations engineer designed to reduce resolution time without removing human oversight.
Beyond customer care, Nokia sees agentic AI playing a role in alarm correlation and anomaly detection across broadband networks. Modern access networks generate large volumes of alerts that often point to a single underlying issue. “There are lots and lots of alarms in the networks,” Kora said. “How can we actually correlate all of them without having to have a manual person do that?” Agentic AI can help operators connect those signals more efficiently, but only if the system is reliable and grounded in domain expertise. Trust remains a central adoption challenge, particularly for operators that lack the internal resources to validate AI-driven decisions. “It’s not just AI,” Kora said. “It’s AI grounded with domain knowledge that is important and that’s going to bring value for our operators.”
Tommy Clift:
All right, Sireesha, thanks so much for being here. FiberConnect day one on the floor. So you guys have announced quite a few agentic AI capabilities here this year, Altiplano AI being one of them, which is an existing platform now agentic AI. Can you talk a bit about what's new with that launch this year?
Sireesha Kora:
Altiplano has always been a data-centric platform with open interfaces, and it was always consumable at API levels. And we've had predictive AI and proactive AI applications for a while now. But what we are doing this year is taking it one step further, making all the great tools that we have in Altiplano, the automation, as well as the AI capabilities in Altiplano, consumable by agentic AI. So, as a part of this agentic AI launch, we have looked at one of the high-value use cases, which is for troubleshooting, and we are making it available for operators to use it. Think of it like a virtual operations engineer that is going to help operators improve the customer experience, reduce the amount of time that they're tending to troubleshoot, and resolve the tickets. So it's not just providing you isolation capabilities. Most importantly, it's also providing you capabilities to figure out exactly what's happening in the system, root cause, and also provide you a repair action for that. That too in a trustworthy way with utmost reliability, which is very, very important for the operators.
Tommy Clift:
Right. So speaking of operators, where are you guys from your perspective really seeing operators currently use agentic AI?
Sireesha Kora:
Well, there are so many applications that AI and agentic AI in general could be used. Operators are always looking for those high-value use cases where typically they're spending a lot of their time, energy and money. And on such applications and one such use case happens to be in customer care. Today, operators are spending enormous amount of time trying to figure out and deflect all the trouble tickets that they're getting. So one important thing that we hear from operators is that, "How can you really help us there? How can you help us go one step further?" So I have some AI help me as a virtual engineer to take care of all that for me.
And the other thing that they also ask for is, today you have all these sophisticated AI applications. However, I still want to accurately go and find out what the root cause is. And that's also something which is very, very important for them with the reliability that I mentioned before as well. So customer care is definitely on high-value use case. We do see other use case, like for example, trying to figure out anomalies and see what's going on in the network. There's lots and lots of alarms in the networks. How can we actually correlate all of them without having to have a manual person like a human do that? So, bringing all that in is something which is quite valuable and useful for them.
Tommy Clift:
Right. Yeah. So you mentioned quite a few of the challenges they're looking to solve with agentic AI. I'm also curious to know, in the adoption process, what are the challenges they're kind of facing in actually using agentic AI, all of these tools? And kind of I guess where are you guys fitting in terms of trying to guide them through that process?
Sireesha Kora:
When people talk about AI, the first question that comes into the customer's mind is, "Is it going to be reliable enough? Is it always going to give me a trustworthy answer?" And that's where they're looking for vendors like us, because it's not just AI, its AI grounded with domain knowledge. That is important and that's going to bring value for our operators. So they're coming to vendors like us with questions like, "Is your AI going to be trustworthy? Is it going to be reliable? Is it going to be accurate enough?" So, the biggest challenges for them is they don't have the expertise and skills to look at all these problems and figure out whether the AI is for coming back with the right answer or no. So they're looking at us as vendors to take all this experience that we have built over the last decades and give them a solution that works all the time. And that's something that's very, very important.
Tommy Clift:
Well, Sireesha, thank you so much for joining us and telling us a bit about what you're doing here today.
Sireesha Kora:
My pleasure.