Unlike previous revolutions that took decades to unfold, generative AI (GenAI) is changing how value is created, captured, and delivered – and with breakneck speed and impact. Across verticals and geographies, GenAI is reshaping labour markets by enhancing human capabilities, improving productivity and increasing efficiency. For example, SoftBank plans to deploy 1 billion AI agents. That’s a whopping 1,000 agent ‘co-pilots’ for every Softbank employee. Masayoshi Son, CEO of the Investment Group at Softbank, believes that ultimately, his company could have trillions of agents making decisions and performing tasks.
Softbank, however, is just one example of a much wider trend. GenAI is making its mark and changing the nature of work practices from familiar to futuristic.
We are living through a profound shift in the way people and organisations accomplish tasks. The world of links, dashboards, and reports are diminishing. In its place, human users will write instructions in plain text, or just simply say what they need done, and the AI agents will determine how best to facilitate the request. As Gartner so aptly forecasted – by 2026, 30% of enterprises will automate more than half of their network activities. What does this mean for telcos?
Are telcos ready to be intent-driven?
In the words of industry analyst, Sebastian Barros, “AI changes every layer of the network: radio, transport, core, compute, and economics. The Telco Big Bang is when machines, not humans, become the primary customers.” Since GenAI shrinks the distance between intent and action, this very premise poses a significant challenge for telcos.
Intent represents more than a new user experience layer; it’s a new operating model that current business support systems (BSS) and operating software systems (OSS) struggle to handle. Agentic AI, the next evolution in AI adds autonomy, decision-making, and task execution. This evolution will further fracture today’s rule-based systems, requiring a full stack re-architecture that translates language into action.
While the opportunities are endless, today’s telcos have gaps to close before they can make the leap to being intent driven. First there’s the network gap: 5G standalone rollouts need to happen in greater numbers to enable sufficient network capacity/capability for personalisation and monetisation, which can only be accomplished through real-time slicing or latency control. Then there’s the software gap: today there’s a lack of agentic orchestration layers that can interpret intent in real time. Telcos need to close these gaps to build agent layers into their ecosystems that understand user intent and can act on the intent autonomously.
For example, a user opens an app to play a video game. The telco AI agent understands the user’s request and acts on the intent. It deploys optimum bandwidth, allocates necessary network resources, and communicates with other parts of the network such as fraud detection (to prevent in-app smishing) and billing (to enable accurate charging, depending on the subscription type). User experiences in the future will declare intent – explicitly via chat/app or implicitly via app signals. The Intent-Net auto-allocates the right slice/latency/bandwidth and updates charging if necessary – only for that window.
In this scenario, the telco is not just enabling the user to play the game, it is delivering a user experience at a much higher level of quality. The user has everything they need, including transparent charging and protection from online fraudsters. This new normal won’t require the user to tell the AI agent how to accomplish the goal, instead agents will compute the intent, then work out the best way to accomplish the task. Essentially, AI agents will be a combination of search engines, personal assistants, and co-pilots.
From telcos to Intent-Nets
While well placed to become Intent-Nets, telcos must first re-architect their systems around the hyper-personalised intent of their users – whether it’s a consumer wanting to book a holiday or an enterprise seeking to onboard 1,000 IoT devices.
Becoming an Intent-Net is a multi-stage process of moving from rule-driven to intent-driven systems. It consists of anti-fraud, network operations, and customer experience. This process, however, needs to be tackled in the following sequence.
• Fraud and Assurance – Security becomes more complex as AI agents interact with networks, systems, and other agents. Scammers are using deep-fake audio and video to perform account takeovers via SIM swapping, smishing and cloning, however GenAI is also leading the defense against fraudulent activity. They now have the capability to equip agents with the tools, knowledge, and memory to battle a variety of fraudulent activities.
Example: A telco partner decides to launch a new product. Prior to the launch, autonomous agents can put in place the necessary assurance controls to ensure that the product is activated and billed without leakages. When it comes to fraud, agents can conduct a deep exploration of misuse and practices associated with that product, ensuring the necessary preventative measures are in place.
• Networks – The Network Layer translates high-level users or application intents into fixed, actionable network behaviours. An Intent Layer interprets each request and maps it to measurable services requirements and resource configurations.
Once resolved, the network autonomously orchestrates the necessary slices, routing paths, QoS profiles, and power-optimised modes. In parallel, assurance, security, and fraud-detection functions adapt to the session context, validating charging, preventing spoofing, detecting credential misuse, verifying device integrity, and monitoring for anomalies.
The above reflects the core principle of the Intent-Net: a single device may create multiple intents, each triggering a precise, fully automated network, assurance, and protection response.
Example: A user initiates an intent to stream a live soccer match. The Intent Translation Layer interprets this as a requirement for sustained HD or 4k video delivery with sub-30-millisecond latency for the duration of the match. Subsequently, the orchestration framework allocates an appropriate QoS profile, instantiates a dedicated network slice, prioritises video packets, and dynamically optimises a content delivery network (CDN) path selection to maintain the necessary performance throughout the duration of the stream.
In the not-too-distant future, telcos will work towards intent-based network operations. For instance, a network engineer may have a goal to ensure high availability for a specific application. Autonomous agents will orchestrate the objective across the network – with minimal manual intervention. These types of scenarios will take place across three layers. The Business Layer, where the objective is set, the Intent Layer which will translate the intent into specific network policies and configurations, and the Network Layer which will execute the objective across the infrastructure, e.g., routers, switches, firewalls, etc.
• Customer Experience – The final step for the intent-led telco is customer experience. With GenAI as the new operating layer, telcos will be able to bridge the current siloes that span BSS, OSS, and the customer experience. In doing so, they will be able to convert the customer’s intent into what the systems need to do fulfill the request.
Empowering telcos through generative agentic AI
Through self-learning, self-healing, and self-optimisation, AI-driven networks are in a constant state of evolution, enabling a security posture that becomes progressively stronger over time. These systems proactively detect, prevent, and adapt to emerging fraud vectors and cyber threats, ensuring a resilient and trustworthy service environment that can deliver seamless service to customers.
Telcos already possess the key ingredients to become Intent-Nets: copious amounts of data about their customers, their network activity, and their location, a distributed infrastructure, and highly automated systems. This data can be used to train their AI models to fully understand how the entire network is performing and what is needed for better customer experiences.
As an Intent-Net, telcos have the potential to tackle widespread fraud. In an Intent-Net every action by an AI agent refines the next. Since this happens in the background, users never see the process, only the seamless outcome. Finally, AI enables self-learning, self-healing, and self-evolving systems that continuously strengthen their defenses against fraud and cyber threats. With fraud out of the equation, telcos can take more risks on disruptive innovation.
AI-driven networks empower telcos to innovate and take strategic risks with greater confidence. However, to benefit from this empowerment, telcos must re-architect for intent, building systems that understand, anticipate, and act. Doing this challenges telcos to undergo a complete mindset change. This change requires telcos to transition from just delivering static partnerships, where data is just data to creating real-time and context-aware experiences, as well as individual session SLAs that are based on customer intent.
The GenAI revolution is underway: will telcos be at the forefront?
The GenAI revolution is underway and hyperscalers and OTTs are already investing heavily in these areas. This technology is not just an overlay: it’s a new operating layer that fuses customer intent with network intelligence. By evolving into Intent-Nets, telcos have a unique opportunity to reinvent themselves into an entity that boasts seamless operations, is customer-driven, and over time be truly fraud free. Ultimately, these telcos will become fearless, seamless, and fraud free.
As the revolution continues, telcos that successfully integrate Agentic AI into their core operations and business models will be better positioned to reduce costs, innovate faster, and capture new monetisation opportunities. Those that lag behind risk becoming commoditised providers in an increasingly intelligent and automated digital world. The question, however, remains – will telcos be at the forefront of autonomous, agentic AI?
Author
Harsha Angeri is Head of AI Business & Corporate Strategy, Subex
