As it announced its first quarter earnings for its financial year, Amdocs also launched aOS, which it described as a new AI-native operating system for telcos. The launch can be viewed as Amdocs’ play for relevance in the evolution, or even revolution, in operational processes that may be enabled by GenAI and agentic AI. aOS introduces a framework for the deployment and management of AI agents within the gamut of telco’s operational processes, fed by domain-specific data and data insights, and managed by Amdocs’ processes for ensuring trust, security and policy compliance.
While the company announced a quarterly revenue of $1.16 billion, an increase of 4.1% on last year and slightly above the midpoint of its guidance, it also saw its share price drop 8.6%, before rebounding 5% after hours. A new five year deal announcement with T-Mobile (although overall 2026 revenues from T-Mobile will be lower year on years) may have helped – as could the introduction of aOS.
The company described aOS as “an agentic operating system purpose-built for telecommunications” and added that it will enable operators to “accelerate their generative AI strategies on top of any BSS-OSS stack.”
Ronen David, Product Marketing Lead, AI & Data, said, “We see more and more what what customers are are trying to do is to build a governed AI ecosystem that will enable them to harness agentic capabilities from multi vendors into a single unifying AI ecosystem. We completely expect that, and we tried to tailor our new approach into that mindset of how we connect to those existing AI ecosystems.”
“They [customers] are coming with their own plans, their own strategies. They are smart people that understand exactly what they want to achieve. And our role is to help them fulfil this, but to make it in a very scalable manner.”
David said that the big shift agentic AI will bring is that it will move from AI’s current role in specific tasks and jobs, to understanding and managing the end-to-end business objective of the specific process. aOS is Amdoc’s blueprint for deploying and securing that shift.
AmAIz is not dead but we will be calling it Cognitive Core which is important because we don’t want people to think, yes, this is what we already know. This is different in approach.
amAIz in grace
So is aOS a new technology offer from Amdocs, or is it a wrapper and repositioning of existing capabilities?
“It’s a bit of both,” David said. “aOS is how we construct our new portfolio so it is based on a lot of what we have, but it also combines not just new capabilities but also new concepts or a new mindset on how we see the telecom industry going forward and how we can support that.”
One aspect of aOS that is essentially a re-packaging is its “Cognitive Core”, one of the three main building blocks within aOS. David said that this is basically a reworking of everything we previously knew as Amdocs’ AmAIz GenAI suite – with the new capability creating libraries of telco-specific agents and sub-agents. It is also a foundation of telco-specific, context-aware telco data insight that Amdocs can serve to customers in a consumption-based model so they can inject that into their GenAI strategies, using what they need for what they are trying to achieve.
“We shape our agents in the form of what are the business processes they are going to support. It’s great that you have agents. You also want decisioning, we inject insight so an agent can be data informed, and not just be able to answer a question. There’s a lot of insight we can inject into a process based on insight.”
“The is the core brains of aOS,” says David.
“It’s basically amAIz on steroids, we’ve taken amAIz and transformed it into Cognitive Core.” The amAIz branding and product name will now be shelved, as David said, “AmAIz is not dead but we will be calling it Cognitive Core which is important because we don’t want people to think, yes, this is what we already know. This is different in approach.”
Amdocs is not very known as the most open. So this is a very different proposition, I think, coming from Amdocs.
Opening up
On top of the Cognitive Core is CES26, Amdocs’ BSS, OSS, and Network suite. David says that Amdocs’ promise is that any OSS-BSS suite can work on top of Cognitive Core – not just Amdocs’ own offers. That said, customers will get the most benefit out of aOS if they do have the latest OSS-BSS offerings from Amdocs.
“Our promise to the market is that any BSS-OSS stack will integrate and work on that, and will provide the same the same value. Having said that, obviously if you will work on the latest CES version the benefits you will get from the entire aOS will be higher. You don’t have to modernise to get the benefits of aOS but if you modernise you get more.”
For David, this aspect of openness is an important part of the shift in Amdocs strategy – that it wants to place itself within the wider agentic AI transformation within businesses.
“We are the King of OSS-BSS, right? This is where our key value is. But obviously, customers have much more systems in their stack – the ERP, HR, payments – there’s a lot of other systems we don’t necessarily control. So some processes will be based on integrating capabilities from multiple stack and multiple system. It’s not always pure BSS-OSS and we want to be part of that revolution, right? We want to help our customers accelerate that.
“And this is exactly the mindset that we had in this very open approach. Amdocs is not very known as the most open. So this is a very different proposition, I think, coming from Amdocs. Just use us and then leverage the capabilities. Even the business or the commercial model behind it is very dynamic and effective to support this mindset of multi-vendor ecosystems.”
David also points out that the technology and platform for aOS is exactly the same as Amdocs uses internally for its own agentic AI development – with the same product team being responsible.
Understanding Amdoc’s position
You wouldn’t necessarily expect competitors to have kind things to say about Amdocs. But where their criticism comes from can still be instructive in terms of where Amdocs will face objections. One of those taking a view is Totogi CEO Danielle Rios, who is bringing to market AI-native solutions in the BSS and charging area of telco’s business.
For her, the proof point of Amdoc’s new strategy is simple: “Show me this move makes Amdocs’ projects 90% faster and cheaper. That’s what AI delivers and that’s what operators expect.”
As you might expect, Rios doubts that this is a viable route for Amdocs.
“Their business depends on the opposite. The vast majority of Amdocs’ revenue – north of 80% – is professional services oriented. Managed services, implementation, change requests. AI accelerates this work and can do it 90% cheaper. AI doesn’t protect Amdocs’ business model. It accelerates its extinction,” she said.
She also sees aOS as a play for relevance in a world where Amdocs is facing increasingly tough competition.
“Migrating off Amdocs used to be a career-ending risk – five years, hundreds of millions of dollars, and your job if it failed. In the AI era, the math has changed. What was a career bet and a five year project is now a no brainer and a five month cake walk. Totogi is talking to dozens of operators who want exactly one thing from AI: help me get off my legacy BSS. That’s the use case no incumbent wants to talk about.”
Telcos’ Agentic AI expectations
Whatever the success or otherwise of Amdocs’ approach, there’s no doubt that telcos see the Agentic AI era as a departure that requires them to restructure their operational models, with a focus on understanding data context and the effects of agents with decision making capabilities. David points to upcoming research commissioned by Amdocs that shows that telcos are most cautious about agentic AI in the network – with other business areas likely to benefit earlier.
One leading operator thinker on the AI transformation, Orange Fellow and VP of Software Engineering, Philippe Ensarguet, told TMN in a recent briefing that agentic AI involves a systemic change.
He said, “It would be much way too easy to consider that the agentic AI is only a new feature of what we are already having. But that’s totally false. It’s a true revolution, a new pattern.
He describes agentic AI as moving on from the static, knowledge-based Gen AI models to something that acts in a feedback loop, with memory, to make decisions and act. But it is still in a very early phase – key protocol MCP is only a year in the wild. The whole GenAI journey is only 3-4 years old in public.
But one area that will resonate is Amdoc’s professed willingness to open its systems to other vendors, suiting Orange’s shift to a horizontal infrastructure, with a common life cycle, management, observability and operation.
“Guess what?” Ensarguet said, “Agentic AI systems need to be in the same environment. I’m pretty sure that the Nokias and the Ericssons of the world will be able to implement the best agentic orchestration and studio. But at the end of the day, the reality of our networks, they are made of multiple partners. So do we want to stick and rely on the studio of X, Y or Z? No, never. But what is interesting is that when the partners are moving into that space, they are also learning. And the discussions we have today about disaggregation and software behaviour – I’m pretty sure that it will reach exactly the same stage with agentic at a moment in time.”
Ensarguett added, “The reality in agentic is that we will have to synchronise, potentially, agents from vendor A, vendor B, and vendor C, and it’s where I’m very curious to understand how all the ecosystem will level up our ability to test, qualify and validate the system before having them in production.”
Comments
0