On August 5, Swiss telecom operator, Sunrise, announced the extension of the partnership with Amdocs to modernize its core business support systems (BSS) by deploying its specialized agentic operating system (aOS) for telcos.
Led by Sunrise Chief Information Officer (CIO) Anna Maria Blengino, the collaboration focuses on the evolution of Sunrise’s Customer Service Representative (CSR) CRM interface. The long-standing partnership between Sunrise and Amdocs going beyond fragmented AI pilots into automation on a production scale.
Sunrise has built on its work with Amdocs to modernize the CRM interface for staff, and the project will use Amdocs’ aOS to place AI analytics and automation inside governed workflows.
The Swiss telco is moving AI into systems running customer care and operations. Production, not promises, will decide whether it is useful, safe, and accountable.
From Chatbots to Core Workflows
An agentic operating system is not one chatbot, but a control layer that coordinates AI agents, gives them business context, applies rules, and lets them complete parts of a process. Agents may read complaints, check billing records, and recommend or carry out approved actions.
For a telco, this matters because of problems with the cross agent operating system. A billing dispute can touch CRM, charging, product catalogues, order management, service fulfilment, and network records. Traditional assistants may summarize the case.
Agentic AI telecom aims to move it through these systems, under defined permissions, with each action recorded.
This marks the point where AI agents in telecom move from experimentation to operational automation. Sunrise is not placing a language model besides its CRM. It is preparing to embed AI into the daily work of its call center and store employees.
“CRM evolution is central to our operational strategy at Sunrise,” said Sunrise CIO, Anna Maria Blengino.
“Extending our collaboration with Amdocs allows us to build on a strong foundation while modernizing our CRM. By introducing new capabilities such as AI analytics into our CRM environment, we are equipping our call center staff with the tools they need to work more efficiently, resolve issues faster, and spend less time navigating systems,” Blengino continued to state.
The intended result of the agent operating system is shorter handling time and faster answers. Yet the larger test is whether Sunrise can connect to fragmented systems without allowing agents to make unchecked changes.
AI agents in telecom often operate across estates that include old software, custom integrations, and data held across separate platforms. Agentic AI in telecom can move quickly, but they can also repeat an error across thousands of accounts.
Industry specific CRM is gaining attention because general platforms do not always understand telecom products, service lines, network relationships, or billing structures. According to Forrester’s 2024 survey, 69% of business and technology professionals planned to increase CRM investment.
Security’s the Real Test
Moving AI into core Business Support Systems (BSS) expands the security boundary. An AI agent operating system that can read customer information, change an order, apply a credit, or trigger a workflow which becomes more than a writing tool.
Agentic AI in telecom becomes an operational user in this case. Sunrise will need role-based access, limited permissions, encryption, testing, logs, model monitoring, and a way to stop or reverse unsafe actions.
Within an agentic operating system, human-in-the-loop governance means people remain involved in important decisions.
As agentic AI in telecom takes on more operational tasks, staff can approve sensitive actions, review unusual cases, correct mistakes, and intervene when confidence is low.
Human presence shouldn’t be a ceremonial click. Reviewers need enough context, authority, and time to be able to challenge the system.
Human supervision of high-risk AI systems is mandated by the EU AI Act, requiring understanding limitations, identify automation bias, interpret outputs, and stop a system when needed.
The General Data Protection Regulation (GDPR) also limits the ability to make certain decisions solely on the basis of automated processing.
European Union Agency for Cybersecurity (ENISA) guidance says AI security must cover data, models, infrastructure, access, monitoring, and incident response, not only the chatbot interface.
Sunrise is based in Switzerland, where the Federal Act on Data Protection (FADP)applies to AI-supported processing. Swiss guidance links transparency with the right to object to automated processing and request human review of automated individual decisions.
For software for telecom, privacy controls, clear explanations, and audit trails are therefore central to the project, not optional additions.
Amdocs presents aOS as open, cloud-agnostic, and able to work across different Business Support Systems (BSS) and Operations Support Systems (OSS) environments.
The flexibility of the agentic operating system may reduce dependence on one model or cloud provider, but it also creates more connections to be secure. Each model, Application Programming Interface (AP)I, agent, and data source adds another point that must be governed.
“Industry CRM builds on the foundations of horizontal CRM and delivers workflows, experiences, data models, and extensions that are unique to each industry, providing a faster path to value,” according to Forrester.
Success will not be measured by how many of the agentic operating systems are launched.
Telco software solutions will be measured by fewer errors, faster resolution, lower handling time, secure data use, and employees who understand when automation should act and when it should stop.
Inside Telecom provides you with an extensive list of content covering all aspects of the tech industry. Keep an eye on our Telecom sections to stay informed and up-to-date with our daily articles.
