backBack

Building the Intent-Driven Future with Network as Code and Agentic AI

Nokia is advancing connectivity to intelligent, adaptive and intent-driven networks powered by Network as Code and Agentic AI.

Artificial intelligence is expanding how applications interact with mobile networks, creating a more intuitive environment. For telcos, this expansion means a more efficient way to expose and monetize network capabilities. For developers, it creates a more streamlined path to utilize the network capabilities. Instead of telling the network how to behave through manual configuration, developers can now simply state what their application needs and let the AI system execute those actions autonomously.

Welcome to the era of Agentic AI.

What is Agentic AI?header link

If you’ve used AI to write an email or summarize a report, you’ve experienced Generative AI. It’s brilliant at processing information and giving you a reactive response.

Agentic AI goes a step further. These are goal-driven systems that can reason through a problem, plan the necessary steps, and take action autonomously using the tools available to them. In short, Agentic AI doesn’t just describe the work–it executes it.

Agentic AI with Network as Codeheader link

Nokia’s Network as Code launched with a mission to make 5G network capabilities easily accessible through open, standardized APIs, including CAMARA APIs supported by GSMA Open Gateway. It has successfully turned the network into a programmable environment for developers with more than 20 use cases from advanced fraud detection to bespoke ‘quality on demand’.

By integrating Agentic AI into the Network as Code platform, we provide a powerful abstraction layer. Developers define the intent (the "what"), and the AI agent takes care of “how”: identifying the necessary APIs, determining the execution path, and performing the required actions autonomously. This approach massively reduces complexity for the developer and ensures the network is always performing exactly as the application requires.

Pathways to the Networkheader link

Network as Code provides several pathways for enterprise applications to interact and access the network capabilities:

• Agent-to-Agent (A2A): Multiple agents, such as an enterprise application agent and a network-specialized agent, can be deployed to coordinate workflows and negotiate network resources autonomously.

• Agent-to-MCP: Using the Model Context Protocol (MCP), agents can discover and access available network APIs in a specific region. The agents remain portable and can operate consistently across different global networks.

• Agent-to-API: An AI agent calls specific Network as Code APIs to fulfill requests based on real-time application needs.

Governance, Trust, and Validationheader link

Networks must not only work autonomously–they must also be trustworthy. We have built safety rails into the system to ensure that every action taken by an AI agent is safe and secure. Every request is checked against pre-approved rules in real-time, ensuring the network always stays within its operational limits.

The result is a network that feels responsive, one that understands what your business is trying to achieve and automatically aligns itself to help you get there.

What does intent-driven connectivity look like?header link

Consider a drone operator remotely managing a fleet of inspection drones. Traditionally, this requires constant manual oversight: checking coverage maps, monitoring real‑time connectivity, and requesting bandwidth boosts during network congestion.

In an agent-to-agent setup, the drone operator uses their enterprise AI agent and simply states:

“Ensure stable, high‑quality connection for this drone’s inspection route.”

From there, the Network as Code agent takes over. It identifies connectivity requirements, discovers the right APIs and tools to monitor the drone in real-time and autonomously trigger a bandwidth boost when network congestion is detected.

The result: Network Code agent understands the network and manages its complexity, so the drone operator can focus entirely on the inspection results rather than network operations.

Preparing for the agentic future with Network as Codeheader link

The transition to Agentic AI marks a turning point where the network becomes an intelligent collaborator. As we work towards launching Network as Code AI agents, you can prepare by:

• Exploring existing Network as Code APIs and MCP capabilities

• Identifying workflows where autonomous execution reduces complexity

• Designing agent-ready applications

• Connecting with the Network as Code team to pilot Agentic AI workflows

Together, we are advancing connectivity towards intelligent and adaptive networks.