- Amdocs creates human-like AI agents for customer service
- The company wants its AI agents to speak with the brand personality of each particular telco
- It’s also developing AI software to create digital twins for each telco customer
DTW IGNITE 2026, COPENHAGEN, DENMARK – Some of the presentations at TM Forum’s event this week started sounding a bit repetitive, with operators talking about being "trusted entities" and telecom technologists fretting about AI and how it’s turning their worlds upside down. But the conversation between Amdocs’ Gil Rosen with Anna Maria Blengino, CIO of Switzerland’s Sunrise, was more lively. It turns out that Rosen is the chief marketing officer of Amdocs, and his perspective was a breath of fresh air.
On stage, Rosen said, “Before ChatGPT launched, they worked for seven years. This was a company that was working on something that nobody used. And then, within a year of launching, they created a revolution. And now, we have the opportunity to really use this incredible capability to create the kind of experiences that we always thought belonged to maybe another industry — that they could do things that we were not able to do.”
One of the first ways that telecoms are using AI is in customer experience. And this is right up the alley of all the OSS/BSS companies that attend DTW Ignite.
Rosen thinks that AI will enable telcos to create individual relationships with their customers. “We’re talking about a cognitive layer, a relationship layer that understands the customer, talks to the customer, reacts to the customer.” He encourages telecom operators to talk more about the human aspect — the psychological aspect — of the interactions.
Speaking with Rosen after his on-stage conversation, he explained that each company has a brand, even a personality, if you will. For instance, T-Mobile has an image as a rebel. It’s the un-carrier that trash-talks its competitors. AT&T, meanwhile, does not trash-talk; rather its personality is more of a caregiver.
These companies, and ideally all companies, need their AI agents to represent their brand personality.
On top of making telecom AI agents speak in the company’s voice, Amdocs is working on agentic software that takes into account the personality attributes of the customer it’s speaking to.
“Will the agent talk the same way to my 80-year-old dad as it talks to my 20-year-old daughter?” asked Rosen. “Probably not. Humans do that intuitively. Now, agents need to be taught the same.”
Amdocs is working on “personality engineering,” which is the art and science of designing AI agents that adjust their communication based on whom their talking to.
“So, I think the point is we are moving from customer experience to customer relationships, and it's a huge conversation," said Rosen. "CSPs need to aspire to have a long-standing relationship with their customers. That's what made us also invent a new entity that we call the customer digital twin.”
Amdocs’ customer digital twin is a representation of each individual customer with attributes of the actual person. The company has created more than 100 personality attributes, which can be assigned to a customer based on things the telco might know such as age, gender and whether the person seems to understand technology. That's the starting point. But then, as an operator interacts with that customer, it learns more about them and continually refines its interactions — just as humans do when they get to know each other.
But how does this all work in real life? Are telcos prepared for such sophisticated AI agents?
Blengino, the CIO with Sunrise, said, “We all know that the biggest challenge is complexity.” She said telcos are dealing with years of mergers and acquisitions along with migrations of data and applications scattered across several platforms. This is the technical debt challenge of virtually every telco in the world. “You can imagine how our frontline agents have to invest their time to navigate across all these platforms even before they start serving the customer," said Blengino.
But Sunrise has built a new customer relationship management (CRM) system to engage with its subscribers via human-like conversations. “We are planning to adjust our architecture, not yet with a huge, long-year transformation, but we want to preserve the investments we’ve made in our platform and layer on top of this core a layer of engagement,” she said. "The back-end represents our stability, and the layers on top represent our adaptability. And this will have a massive impact on our customer experience.”
This isn’t to say that Sunrise has not prepared and laid the groundwork for AI customer agents.
A few years ago, Sunrise decided to create a platform to concentrate all its data. “This is now the biggest treasure we have in the company,” Blengino said. “All the system and record data, the network data, the interactions of our customers, the behavior is centralized in a single platform that is secure. And it’s the basis for our personalization. But we can only do this thanks to this solid data foundation.”
Amdocs updates its own image
In February, Amdocs repositioned itself as an Agentic Operating System (aOS) provider, putting less emphasis on its image as a traditional OSS/BSS vendor. Its aOS is designed to ride on top of its traditional telco OSS/BSS stack.
Rosen said, “When we really thought deeply about this evolution that we're going through, we understood that this is not the next version of BSS/OSS. This is providing our customers with the capability to orchestrate agentic interfaces. So, we're not ditching our heritage; we're building on it.”
He said customers are really resonating with the aOS pivot because they’re so familiar with operating systems from the likes of Apple and Android. It’s an elegant descriptor that people understand.
OS is cool, like Apple. Nobody ever thought of OSS/BSS as being cool.