How to coordinate multiple agents to fund revenue and growth strategy
The promise of 5G-Advanced technology is not “more bars on your phone,” but “networks that do more with less”: they are smarter in operation & maintenance, optimization, energy efficient, and useful in business. As with any wave of technology, early adopters get seduced by cost savings. Once the foundation is solid, the next phase kicks in: generating new revenue. It’s a recurring theme: first “cost-out,” then “revenue-in.”
The same is true for Huawei’s IntelligentRAN. This new, state-of-the-art solution performs like an athletic athlete. The architecture of intelligentRAN includes three layers. Service layer is for cross-domain collaboration and network layer ensures single-domain autonomy: this is the “brain” of this athlete. And a layer of network elements with digital locations— the “arms and legs”, who achieves more with less effort. First, it reduces energy consumption (OPEX), then it increases muscle mass and develops new movements (new revenues, new market segments). The intelligentRAN agents are the brain of the wireless domain, are based on a foundational telecommunications model and a RAN digital twin system (RDTS). The arms and legs are digital sites that provide coverage and capacity where needed, with lower kWh/TB consumption and fewer field visits. Intelligent RAN agents and digital sites work together with ultra-fast response times, operating decisively within auditable limits.
Brain and limbs: RAN agents and digital sites
In 5G-Advanced, the “brain” consists of agents that observe the network, propose actions, and, when the benefits are clear and the risks are limited, act independently. Their promise is concrete: faster problem solving, improved user experience, and less repetitive tasks that currently consume OPEX. However, introducing artificial intelligence and agents involves a demanding program. The five challenges that every operator must anticipate are:
- Conflicting goals and arbitration: The system must continuously arbitrate opposing goals, such as maximizing energy savings versus guaranteeing network capacity. Without clear, shared priorities, fluctuations and abrupt shifts occur.
- Scale and operational complexity: As maintenance, optimization, energy, and service assurance coexist at a massive scale, O&M must become more flexible and coordinated.
- Precision, latency, and reversibility: The RAN operates in milliseconds, so its automated actions must be correct and timely, with the ability to explain, record, and safely revert.
- Data quality and shared context: To act intelligently, agents require reliable telemetry and a common “language of intent” to build a shared context. Without this “shared memory”, agents make decisions blindly or reactively after the fact.
- Security, functional safety, and compliance: Developing a robust framework requires layered defenses, including strong agent identity, the principle of least privilege, abuse/injection detection, audit logging, human oversight, and regulatory compliance.
The response to these challenges is based on governance and standardization. The interaction between intention and policy is expressed in agent-to-agent telecom protocols (A2A-T), which translate business objectives into technical policies for a given domain, avoiding ad-hoc scripts and enabling collaboration between scenarios. Audit is part of the design: every recommendation and action is signed and traceable, with a timestamp, “why” and “how” the decision was made. Security works in layers: strong agent identity, allow/deny lists, data segmentation, detection of prompt injection attacks, and human override to return to baseline in seconds if something deviates from the norm. In parallel, preliminary verification in a RDTS allows configurations to be tested and errors reduced before going into production. The result: simplification (from intent to frictionless action), high precision (coordination between agents with shared memory and fast connectivity), and high reliability (protective barriers, traceability, and fallback).
Agents: four scenarios with immediate return on investment.
- Maintenance: agents prioritize alerts, suggest diagnoses, and automate fixes, resulting in fewer tickets and lower mean time to repair (MTTR).
- Optimization: agents close the loop with tilt/azimuth/beam recommendations verified in RAN Digital Twin System (RDTS) and implemented in minutes, raising reference signal received power (RSRP) and reference signal received quality (RSRQ) where it is most needed (indoors, hotspots).
- Energy: agents apply precise cell/band/hour rules that reduce kWh/TB without compromising coverage.
- Service quality assurance: agents prioritize users/applications (video, gaming, critical IoT) with visible SLAs. Creating a differentiated experience is no longer a slogan, but a product to be monetized.
What the brain decides, limbs execute. Digital sites (smart antennas, integrated power supplies, and sensors) provide connectivity where life happens, with adequate power and no energy waste. Digitization makes the antenna a “self-aware” limb: the twin of the site knows its orientation and surroundings precisely, so remote adjustments, vertical and horizontal, improve range and uplink without climbing the tower. Energy can be coordinated with the environment in real time (turning off where not needed, cooling better if demand increases), and ultimately field visits and downtime are reduced.
Equipment goes hand in hand with intelligence. Multi-band antennas and panoramic radios reduce the footprint and weight of installations. Meanwhile, lightweight devices, such as EasyAAU (the industry’s lightest active antenna unit), simplify construction work, dissipate heat more effectively, and reduce energy consumption. The introduction of Massive MIMO to FDD (Multiple Input Multiple Output; Frequency Division Duplex) and low bands provide deep coverage and a robust uplink for artificial intelligence (AI)-ready experiences —without excessive density. And because the brain provides demand and context (traffic, events, mobility), investments are prioritized where they yield returns: less CAPEX for a useful location, fewer kWh/TB, and an experience that is visible exactly where the customer is looking for it.
The RAN is no longer “iron that transmits bits”. It is a fully trained body. A brain that understands intentions, controlled and verified before acting. Digital limbs that feel, adapt, and act with precision. This is how a bridge is built between cost-out and revenue-in.
No one wins alone: ecosystem and economics
In 5G-Advanced, technology works when the ecosystem works in a coordinated manner. No operator, supplier, or start-up can cover all areas: spectrum and RAN, core, location energy, OSS/BSS, security, edge, AI applications, integrators, standardizers, and developers must co-create for multi-stakeholder coordination to work from start to finish. This ecosystem can be seen in three layers that complement each other.
Layer 1: a common technical framework. Agents can only work together if they speak the same language. This “common language” is created by three key components: intent protocols (A2A-T), which translate business objectives into domain-specific rules; open interfaces that connect the RAN, core, energy, and edge; and a common digital twin for verifying actions before they go into production. This structure includes consistent telemetry, shared memory between entities, and a priority and arbitration system so that savings, mobility, and experience assurance do not conflict with each other. Once this foundation is in place, automation ceases to be an isolated project and becomes a repeatable capability.
Layer 2: partner orchestration. Results emerge when each participant contributes: operators with network context and customers; providers with RAN, digital sites, mechanical knowledge and single/cross domain agents; external OSS developers and AI companies with cross domain agents; integrators to connect domains; and standardization organizations to set the rules of the game.
Layer 3: the economics that make it scalable. The healthy sequence remains unchanged: first achieve performance, then revenue will follow. As O&M is automated, energy consumption is reduced, the optimization cycle is shortened from days to hours, and the experience is assured from the provisioning. With the generated savings, it is then possible to fund a commercial offensive, which includes guaranteed experience in services (video, gaming, critical IoT) and service APIs so that third parties can monetize latency or submetric position.
For this ecosystem to thrive, adopting a value-sharing model is crucial: This model must be built on three contractual pillars: first, telemetry contracts that define what is measured and who sees the data; second, agent SLAs that specify what the agent promises and the penalties for non-compliance; and third, a clear revenue split defining how the operator, provider, and developer are paid when capacity is sold as a product. Without this framework, hype does not translate into revenue.
When efficiency frees up OPEX, the operator invests in differentiated experiences and opens markets of opportunity. These revenues justify adding more agents and verticals (ports, metro, logistics, retail), which in turn provides more data and better models, closing the positive flywheel. The advantage of 5G-A is not having more features but packaging repeatable movements: “20ms guaranteed at the stadium,” “enhanced uplink during train rush hour,” “optimized energy during heat waves,” each with its own price, API, and audit.
Brain and limbs, the business in top shape.
When combining all the parts of the “body”, from the brain to the limbs, the network then performs like a competitive athlete. First, it reduces “calories consumption” (energy, MTTR, field visits), which then converts that “good shape” into points: SLAs, APIs, and differentiated plans that increase revenue.
This “winning athlete” does not require more features. At the same time, customers don’t buy features; they buy results that impact the bottom line. The key is coordinating the existing agents (the brain) with digital sites (the limbs) according to clear rules (from intent to policy, traceability, security), so that every optimization funds the growth strategy. Huawei has followed this principle: 5G-Advanced with IntelligentRAN is profitable because it lowers costs today and increases revenue tomorrow. When technology is presented this way, it ceases to be a list of acronyms and becomes an economic balance sheet.
The infrastructure discussed so far consists of elements that sense and adapt. A brain that understands intentions, verifies in a digital twin, and operates with safety barriers. This creates clear benefits for the whole organization:
For the CFO, optimization reduces tasks that used to take ten days to a few hours, energy consumption is lowered without compromising coverage, and network downtime is reduced through remote diagnostics.
For the CMO or COO, this same technology serves as a guarantee that at the stadium, on the high-speed train, or in the shopping mall, users will actually get better picture quality, less latency, and more stable sessions—and that this difference can be packaged and sold. Three words (operations, experience, monetization) that tell a story every buyer understands and accepts.
