Amdocs joined the telco agentic fray this week too – some academics warn that claims of double-digit productivity and multi-trillion dollar economic potential gains need a closer look
Deloitte has unveiled a new framework designed to help telecom operators deploy and scale advanced AI systems across their businesses, with the potential to unlock up to USD 150 billion in value over the next five years.
However some academic research questions the credibility of large productivity or economic impact claims made for agentic AI. A recent paper on arXiv highlights that many of these forecasts rely heavily on narrow technical metrics like model efficiency or task performance, while often overlooking real-world factors such as human-centred usability, safety, governance and economic viability
Agentic AI Blueprint for Telcos
Deloitte’s Agentic AI Blueprint for Telcos offers a model for integrating so-called agentic AI into telecom operations. Unlike traditional AI, which typically supports narrow, task-specific improvements, agentic AI is capable of reasoning, adapting and acting independently – enabling end-to-end automation of complex functions such as network management, billing and customer care.
The blueprint is designed to help telecoms move beyond pilot projects and implement AI systems that generate tangible operational improvements. Built on the TM Forum’s enhanced Telecom Operations Map (eTOM) and powered by Deloitte’s proprietary industry data and AI tools, the framework aims to give telecom companies a scalable path to embed AI across their existing infrastructure in line with industry standards.
“Agentic AI presents a USD 150 billion opportunity for the telecom industry – and the race to capture that value has already begun,” said Deloitte Consulting telecom, media and entertainment and technology group AI leader Baris Sarer. “To move beyond pilots and drive modernisation, telcos can utilise a clear, scalable path to embed AI across their operations.”
Aligns with TM Forum’s standards
The blueprint offers telecom-specific support, including a prioritisation model for AI use cases, a library of workflows aligned to TM Forum standards, automated process engineering tools, and practical implementation guides. Deloitte says the approach also draws on its relationships with leading cloud and AI technology providers to help ensure smooth integration into telco environments.
TM Forum AI & data innovation mission lead Guy Lupo said the blueprint complements the Forum’s own efforts to accelerate AI adoption across the industry. “Deloitte’s Agentic AI Blueprint will provide TM Forum members with a clear path-to-production solution using TM Forum’s Open Digital Architecture,” he said.
“Together, we’re helping telecom organisations quickly and strategically integrate agentic AI into their operations to deliver intuitive, intelligent experiences.”
The launch builds on Deloitte’s broader AI strategy, including the formation of its Global Agentic Network and the establishment of a Global Simulation Centre of Excellence. Both initiatives aim to help organisations scale AI-enabled workforce and operational solutions consistently across global markets.
Amdocs joins the agentic AI fray for telcos
Earlier this week Amdocs release what it is calling “a new standard defining what makes a telco AI agent an advanced, verticalized enterprise offering”. In a whitepaper, AI Verticalization for Telcos, the vendor outlines telco-grade agents that are “skilled, brand-engineered, trustworthy, and autonomous-capable of transforming how CSPs engage with customers and operate their networks,” according to the company.
Key aspects of a telco-specific AI agent include:
Ontology – a formal representation of domain knowledge that enables consistent data interpretation across platforms and tasks. By leveraging in-context learning and specialise telco LLMs, agents gain reasoning capabilities and adaptability. This structured framework is to ensure AI agents deliver accurate, relevant and uniform responses, while allowing for collaboration across systems and sub-agents.
Autonomy and predictive reasoning
Beyond access to telco data, agents must use it reason, applying domain-specific logic, workflows, and best practices for decisions and actions aligned with industry expectations. This, Amdocs says, is what enables “truly intelligent, telco-effective autonomy”.
Digital twin and simulation
A digital twin is a real-time virtual replica of their physical network. By continuously ingesting data and carrying out accurate simulations, it can provide actionable insights leading to trusted, efficient execution. Data, analytics, and AI/ML form the core foundation of any effective digital twin.
Brand engineering – the digital face of the enterprise
As AI agents become the first touchpoint between companies and their customers, how they sound, look and respond must reflect the brand they represent. Agents cannot be generic; they must embody the voice, tone, and personality of the brand in action.
In a global study* commissioned by Amdocs in collaboration with McCann Tech agency and Coleman Parkes, 80% of consumers said they trust AI agents to resolve service issues, and 60% believe AI can positively shape their perception of a brand—when done right. That trust is earned through intentional design.
Trust is not trivial
Enterprise-grade Agents must also possess robust non-functional capabilities like trust and security, not only enhance service delivery but to ensure operations are secure, compliant and trustworthy, to foster stronger customer relationships and confidence.
*Survey, interviews and focus groups of 120 CSP leaders and 7,025 consumers aged 18-69 in 14 countries across North America, Europe, and Asia.
