BLOG | Autonomous networks won't be built by AI agents alone

The telecom industry isn't short of AI innovation. What's becoming increasingly clear is that innovation alone won't deliver autonomous networks. NGMN's latest report makes that point forcefully, and on this issue, the industry is increasingly aligned.

Operators have already demonstrated that AI can deliver value. From customer care and service assurance to network optimization and energy efficiency, AI is improving how telecom businesses operate. The challenge now is different. It's not about whether AI can automate individual tasks. It's about whether AI systems can work together well enough to support autonomy at scale.

That's where NGMN's report is on point. It advocates for conversation beyond individual AI capabilities to focus attention on the foundations for autonomous operations: interoperability, governance, trust, shared context and operational control.

What’s next is the challenge of autonomous flows that cross network domains, operational silos, cloud platforms and supplier ecosystems, and these are not technology challenges alone. They are industry challenges, requiring alignment across operators, vendors, hyperscalers and the wider ecosystem.

Beyond the agent

The telecom industry is entering a new phase of AI adoption. The conversation is shifting away from what individual agents can do and toward how multiple agents work together. An agent that can resolve a customer issue, identify a network fault or recommend a remediation action is valuable. Agent orchestration that brings decisions and resolution into alignment, even more so. But that level of autonomy depends on whether – and how – individual capabilities can operate as part of a coordinated system spanning across enterprise domains.

NGMN highlights many of the barriers that can stand in the way of that vision in the network context. And one of the most encouraging aspects of the NGMN report is how closely its findings align with work already underway across the industry, including here at TM Forum with our Member community. Many of the challenges the report identifies are no longer theoretical. Operators, vendors and industry organizations are actively collaborating on the foundations required to scale autonomy.

Multi-vendor interoperability, agent coordination, context sharing, governance and trust all become more significant as operators move beyond isolated use cases and begin scaling AI across increasingly complex operational environments. And, of the 12 major standards gaps identified by NGMN, TM Forum’s Open Digital Architecture already has explicit workstreams or roadmap items to provide a concrete response addressing the operator requirements and ecosystem gaps.

TM Forum's AI-Native ODA launched at DTW Ignite 2026, delivering an agent-ready operational architecture that provides standardized interfaces, shared operational context, governance controls, policy enforcement and observability to operators investing in autonomy across their IT and network operations. Those capabilities are essential if agents from different domains and vendors are to work together consistently and safely at scale.

An AI agent working in service assurance may identify degradation. Another may assess customer impact. A third may recommend corrective action. The value isn't created by any one of those agents individually. It's created when they can operate with a common understanding of the situation, exchange information consistently and act within the same operational framework. Without those foundations, operators risk creating disconnected islands of intelligence rather than autonomous operations.

In short: autonomous operations require more than AI capability. They require common operational foundations that allow systems to coordinate decisions, share context and work toward common outcomes.

Building the foundations

The NGMN report also emphasizes the importance of shared semantics, context sharing and common understanding between agents. Through Open Telco AI, a collaborative initiative between TM Forum and GSMA, operators and technology partners are developing telecom-specific models, ontologies and operational assets that create a common language for AI systems across the industry. This work helps agents interpret information consistently, reducing the friction that emerges when different systems operate with different assumptions or definitions.

Operational maturity is another major theme throughout the report. TM Forum's collaboration program is addressing this through AN Levels, AN Level 4 implementation work and high-value scenarios that help operators move from isolated automation initiatives toward scalable autonomy. These frameworks provide a common approach for measuring progress, prioritizing investment and translating ambition into practical operating models.

Taken together, these initiatives demonstrate that many of the foundations required for autonomous operations are already being developed collaboratively across the industry. The challenge is no longer defining what good looks like. It's accelerating adoption.

Turning autonomy into outcomes

It's easy to become absorbed in discussions about governance, standards and interoperability, but operators ultimately invest in outcomes rather than architectures.

The foundations matter more than ever because they enable capabilities that have direct operational and commercial impact. Cross-domain service assurance, autonomous fault remediation, RAN optimization, improved customer experience and greater operational efficiency all depend on systems being able to coordinate decisions across organizational and technology boundaries.

The business value of autonomy lies not in the sophistication of individual agents but in the collective outcomes they help deliver as an autonomous enterprise. As such, the debate about what needs to happen is largely settled.

The challenge now is making it happen.