- Nokia targets real-world telco pain points with agentic AI across operations, rollouts and customer experience
- Early use cases show measurable gains in automation, accuracy and lower support costs
- Full autonomy remains constrained by trust and regulation, with humans still in the loop
FIBER CONNECT 2026, ORLANDO, FLORIDA — Nokia is ramping efforts to deliver agentic AI capable of fixing persistent problems inside broadband and home networks.
The company recently announced an agentic AI framework for home and broadband networks, enabling the rollout of AI agents that can retrieve data across systems and recommend actions. The focus areas —network operations, fiber rollouts and customer experience — are not new priorities.
Stefaan Vanhastel, VP of marketing and innovation for fixed networks at Nokia, told Fierce at Fiber Connect the move came in response to a growing disconnect between AI enthusiasm and operational reality.
“AI is at the risk of being a science experiment," he said. "I mean, it's cool, right? So, then, what do you do? You know, what does it actually do for you? So, we're getting to the point where it actually does something for that."
That sentiment underpins Nokia’s approach. Rather than positioning agentic AI as a broad platform play, the company is starting with defined operator use cases that target cost structure and customer experience.
One example Vanhastel pointed to is fiber deployment verification. Today, operators often rely on limited human spot checks to ensure installations are done correctly. Nokia’s AI agent can analyze images of fiber connections and verify installations so that installation verification can reach almost 95%.
The implications are less about incremental efficiency gains and more about eliminating human error. Improving verification rates doesn’t just reduce errors — it helps restore customer confidence and satisfaction and improves the data that powers operations, said Vanhastel.
The same logic applies to customer support. High churn among help desk staff and the complexity of troubleshooting multi-domain networks remain persistent cost drivers. Nokia’s agents act as a real-time guide for less experienced engineers, surfacing likely root causes and recommended next steps within seconds.
More importantly, those agents can also correlate data across different parts of the network to pinpoint issues faster. That reduces resolution times, minimizes repeat calls and shifts support from reactive to more proactive models.
Telco challenges are still the same
Here at Fiber Connect, the spotlight may be on agentic AI for margin improvement, but telcos remain focused on the same operational challenges as always: faster troubleshooting, fewer truck rolls, more accurate networks and reduced training overhead. And, of course, executing on lofty automation dreams — now with the help of AI.
The industry has spent years talking about Level 4 autonomous networks, where systems can make decisions and execute changes with minimal human intervention. Vanhastel doesn’t dismiss that vision, but he frames it more as a matter of trust than technology.
For now, he expects humans to remain firmly in the loop, particularly for high-impact decisions. Low-risk optimizations, such as tuning Wi-Fi parameters, are more likely to be automated first. But actions that could disrupt large portions of the network or affect critical services will continue to require oversight.
Vanhastel's thoughts align with industry realities. Regulatory constraints, reliability requirements and the sheer complexity of telecom networks risk putting the brakes on full automation. Even as AI takes on more operational tasks, the idea of completely hands-off networks still looks distant for many operators.
The takeaway from Fiber Connect is that AI in telecom is moving out of the experimentation phase and into execution. The challenge isn’t proving that AI can work, however. It’s proving that it can deliver real world results and that operators can actually trust it enough to let it take the wheel.