Out of all the news that poured out of this year’s Mobile World Congress (MWC), perhaps one overlooked announcement was Huawei’s launch of the A2A-T (Agent-to-Agent for Telecom) protocol.

Agent watchers will likely be reminded of the Agent2Agent Protocol (A2A), Google’s widely embraced bid from last year to establish industry-wide standards for AI agent interoperability. Huawei’s spin on the protocol is also open-sourced, and aims to provide a unified interaction architecture for multi-agent collaboration, addressing operator challenges in collaboration, reliability, and security within autonomous networks.

According to Sam Wang, general manager and a founding member of Huawei’s autonomous driving network (ADN) unit, A2A-T aims to improve communication between both the network and service layers.

“We’re building in features like security, a controllable negotiation process as well as features to support collaboration, for example fault optimization,” Wang told this title, adding the Linux Foundation, which took stewardship of A2A from Google last June, is "very welcoming" about the project.

This will likely not only spur the advance of agents in telecoms but also help the industry go beyond current agentic use cases. As recently explored by SDxCentral, most telecom agent usage revolves around back-office examples: IT service desks and energy bill analysis were examples noted by BT and Vodafone in our feature. According to ABI Research, that’s likely to change, with analysts forecasting the 5G RAN automation market will reach $6.6 billion by 2029, while open radio access network (RAN) automation will account for around 42% of that revenue by 2030.

In the Huawei vision, agents are an integral part of autonomous networks. With telecom’s traditionally laggard pace, that convergence may seem as far-off on the horizon as a universal leap to level 4 (L4) automation, a network framework that existed before agentic became the buzzword it is today.

But that may perhaps be the Western view on autonomous network progress. According to Wang, China and the wider Asian market are already ahead of the pack when it comes to automation. This tallies with recent TM Forum research showing Asia-Pacific (APAC) as the highest region aiming for Level 4 this year, at 36%.

“We released scenario-based autonomy over a year ago, which is now deployed across more than 100 live networks from over 20 operators,” claimed Wang. These are high-value use cases, such as fault handling or optimization.

Huawei’s interview took place at London’s FutureNet World event, where Wang detailed the second phase of the Chinese giant’s ADN efforts. The plan, he explained, is to better calibrate such scenarios through upgrading legacy architecture using the power of agents.

"Some customers ask me, if we build an agent, what about my existing systems? Each of those carriers has more than 150 different systems for each province of their networks [which are] operated manually. This means the relationship between the existing system and the new agent system has to be calibrated.”

What it doesn’t mean is ripping out and replacing current systems. Under Huawei’s new approach, all existing systems can be reused following optimization from the vendor.

“We need to upgrade the existing system to provide model-friendly kind of APIs, for example … Because if you look at the industry, almost every agent, the open cloud, the harness, each of the architectures behind each agent, they need to take actions. They need to use tools. So you will see all existing legacy systems reused by the agent."

Chinese ‘AI to the room’

Alongside an examination of all things agentic, Wang gave an overview of telecom market developments in China, as helping to spur L4 evolution.

Following an upgrade from fiber-to-the-home (FTTH) networks to fiber-to-the-room (FTTR), operators now want to evolve to a service level agreement (SLA) form of the latter, where specific performance benefits are contractually guaranteed.

Operators like China Mobile and China Telecom, meanwhile, are piloting “AI plus home broadband” offers, e.g., elderly‑care services that detect if an older person has fallen by accident using camera-based computer vision plus connectivity. These are sold as value‑added home services, often powered by partner applications.

Wang also mentioned instant packages designed for stadium-based events, where users can scan a QR code to purchase two hours of guaranteed uplink, as enabled by strong 5G/5G‑Advanced networks and supporting systems.

Together, such market options see the autonomous networks use case in China move beyond cost savings and energy efficiency, and into new revenue opportunities for operators. Wang believes that operators need to address customer needs with stronger value propositions that customers are willing to pay more for.

“The operator CFO or the CEO, they care about cost, sure. But they care more about revenue. So for [Huawei ADN’s second phase], the other aim is to drive revenue growth. At the same time, you need to have support in your network systems to sell these new products to customers. If you don't have this kind of capability, it's not a real, sellable product, right?” Wang said.