- Telcos are shifting from GenAI experimentation to pragmatic, production-focused deployments
- Amdocs is positioning domain expertise and open, modular architectures as critical to AI success
- Early GenAI wins are coming from targeted use cases layered on top of existing OSS/BSS systems
MOBILE WORLD CONGRESS 2026, BARCELONA — After a few years of GenAI buzz, communications service providers are starting to recalibrate expectations and focus on what actually works in regulated, mission‑critical environments.
Ilan Sade, division president for GenAI and data at Amdocs, told Fierce Network at Mobile World Congress 2026 in Barcelona that many customers now recognize that deploying AI in telecom is fundamentally different from consumer use cases. “There is a clear understanding that deploying AI in an enterprise environment is probably a bit different than our experiences as consumers,” Sade said, pointing to stricter requirements around accuracy, determinism and policy control.
He added that the idea of AI as a shortcut to instant transformation is fading. “Most of them understand already there is no magic,” Sade said. “There was some kind of hype…they start to understand that this is not the case.”
Domain expertise matters more than generic AI
Rather than trying to be a horizontal GenAI provider, Amdocs is doubling down on telecom‑specific expertise across OSS and BSS. “We are not going to be the best GenAI company,” Sade said. “We’re going to be the best GenAI in telco specifically.”
That focus reflects customer demand for solutions that understand telecom business processes end-to-end. “You cannot boil the ocean,” Sade said. “If you want to really bring value, you need to look at business processes.”
At the core of Amdocs’ approach is what it calls a “cognitive core,” built on decades of telecom knowledge. “This is where we documented everything that we know—four decades of knowledge and experience,” Sade said. “We looked at the different business processes, the roles, how you combine human with agents.”
Overlay, not rip-and-replace
A key theme from customer conversations is caution around replacing existing OSS/BSS systems. Sade said AI alone is not a compelling enough reason for wholesale transformation.
“System of record is mission critical,” he said. “It’s the backbone, and it’s the financial backbone.”
Instead, Amdocs is promoting an overlay model it developed called aOS that extracts value incrementally. “Cognitive core is an overlay,” Sade said. “We are doing it on top of our systems…and start extracting value over time.”
Most customers, he added, are taking a step‑by‑step approach. “Even the most innovative…this is the way they are thinking,” Sade said. “Let’s take step by step. Let’s be very pragmatic about it.”
Early wins come from focused use cases
Rather than broad transformations, early success is coming from targeted, high‑impact use cases. Sade cited an example where AI proactively identifies billing issues before customers contact the call center.
“They identify in advance which customer is likely to have an issue with this bill,” he said. “They proactively approach him.”
In that deployment, most interactions were resolved without human agents. “Ninety‑eight point seven five percent of the interactions are ending with an agent,” Sade said, referring to AI agents.
AI transformation starts inside Amdocs
Sade said Amdocs is applying the same principles internally that it promotes to customers, including workforce readiness. “I’m leading both internal transformation and what we offer to our customers,” he said.
The company has rolled out an internal “AI fitness” program to help employees adapt. “You need to be fit all the time,” Sade said. “It’s not a one‑time thing.”
Looking ahead, Sade expects AI to reshape roles rather than eliminate them outright. “Some people will need to shift what they’re doing,” he said. “We will move more people to be supervising an environment that has both people and agents.”
Read all of our coverage from Mobile World Congress 2026 in Barcelona here.