Telcos have invested heavily in 5G, Open RAN, edge, cloud and AI. Yet many networks still operate reactively. Customers detect outages before systems do. Engineers still correlate alarms manually. Change teams still push updates without full confidence in the downstream impact.

The issue is not ambition or telemetry. It is operational debt. Fragmented OSS/BSS systems, inconsistent network data, undocumented expert judgment and disconnected execution layers prevent AI from moving beyond recommendations.

That is why legacy modernization has become a strategic priority. It is no longer back-office cleanup. It is the foundation for turning reactive networks into intelligent, governed and monetizable platforms.

Analyst & Forums like IDC, ISG, Everest, TMForum & McKinsey are putting bets on AI-driven modernized, orchestrated and monetizable platforms

1. Why Legacy Holds Autonomy Back

Telecom networks were not designed as one intelligent system. RAN, core, transport, cloud, IT, OSS and BSS evolved across different vendors, tools and data models. As a result, operators have vast telemetry, but no single machine-readable view of product, service, topology and network state.

That is why many AI pilots remain limited. They can summarize, recommend and alert, but they cannot reason across domains, prove impact or close the loop when the underlying context is fragmented.

For telcos, autonomy is not only an engineering agenda. It is a growth agenda. A telco cannot become a techco if complexity scales faster than its ability to automate, reason and monetize.

2. Four Barriers Leaders Must Break

No Single Network Truth: Alarms, topology, inventory and assurance systems often describe the same network differently. Without a shared source of truth, AI cannot reason across domains or recommend actions with confidence.

Invisible Operational Knowledge: Fault correlation, energy tuning and change-risk assessment still depend on expert judgment. Much of that knowledge lives in engineers’ heads, incident notes and post-mortems, not in reusable decision paths.

Trust Is Not Built Into Execution: A single configuration change can ripple across RAN, transport and core. Without a digital twin to simulate impact first, teams will hesitate to give AI write access to production systems.

Execution Remains Disconnected: A pilot may start in the NOC, but real outcomes depend on provisioning, assurance, capacity and billing. If AI cannot trigger, update and learn from those systems, it remains another dashboard.

This is the inflection point telecom leaders need to focus on. The shift from Level 3 to Level 4 autonomy is not simply a technical upgrade. It changes the operating model from human-validated recommendations to intent-driven execution, where agents act within defined guardrails and continuously optimize toward business outcomes.

From Advisory AI to Intent-Driven Autonomy

TM Forum’s maturity model draws the line clearly. At Level 3, AI recommends and humans execute. At Level 4, operators define outcomes and guardrails. Intent-driven agents then decide and act across domains to maintain uptime, optimize energy use and resolve incidents without waiting for manual intervention on every alarm.

Crossing that line requires three foundations: a knowledge graph that creates a consistent view of product, network and service truth; a digital twin that tests changes before they reach production; and a cognitive intelligence layer that keeps service, network and operational context synchronized with the live environment.

Inside the Runtime Operational Cognition Platform

The three foundations create value only when they work together. Telcos need a runtime platform that sits between the live network and the agents acting on it.

This platform does four things continuously. It ingests live network state. It reasons over a shared model of truth. It governs every decision before execution. It pushes validated actions back into the network.

At Persistent we have built a context aware Telecom R(Ai)G to bring enterprise data with guardrails, governance together and generate relevant cognitive data intelligence to build autonomous operations. Also Persistent brings AI-led accelerators, telecom skills & partner strengths for driving the OSS/BSS modernization strategy

The architecture brings together five runtime layers:

• Data Acquisition normalizes and enriches raw events.
• Services federate context through MCP, A2A coordination and open APIs.
• Memory and Knowledge maintain the knowledge graph, vector store and operational memory.
• Governance applies authority, policy, trust and safety controls.
• Intelligence resolves and projects context so every agent acts from the same picture.

The output is not another dashboard. It is governed autonomous action across orchestration, assurance, optimization and reporting.

Runtime Operational Cognition Platform

Modernization Turns Legacy Networks into Monetizable Platforms

Modernization is where the network stops being only a cost-heavy infrastructure stack and becomes a programmable revenue platform. By abstracting network capabilities through open APIs, NaaS constructs and governed exposure layers, telcos can package capabilities such as quality on demand, location, identity, security, slicing, edge and assurance as consumable services.

The path is practical. Telcos need to clean and federate OSS/BSS and network data, expose reusable service and network APIs, automate fulfillment, assurance and charging and apply policy, consent, SLA and revenue-sharing controls to every offer.

In this model, autonomous operations and monetization reinforce each other. The same knowledge graph, digital twin and governance layer that make the network safer to automate also make it easier to productize. Services can be defined, simulated, exposed, assured, charged and optimized through APIs.

That is the real modernization outcome: a legacy estate transformed into a software-defined, API-first network business.

What Leaders Should Actually Track

Readiness is not a model question. It is an operating question. Leaders should ask how much of their network, topology and service knowledge is current, governed and machine-readable and how much still sits in dashboards, PDFs and individual memory.

The higher that readiness ratio, the more AI can reduce truck rolls, MTTD, MTTR and energy per gigabyte. Without it, AI risks becoming one more screen in an already fragmented estate.

Leaders can use the scorecard below as a practical benchmark, recognizing that ranges vary by network domain, baseline maturity and degree of closed-loop automation:

These ranges should not be treated as fixed guarantees. They are directional leadership benchmarks to test whether modernization is creating measurable shifts in reliability, cost, agility and monetization.

If these metrics improve, modernization has changed how the network behaves and how the business grows.

Ready to move from reactive operations to an autonomous, monetizable network? Connect with Persistent’s telecom modernization experts to explore how knowledge graphs, digital twins and governed intelligence can help turn legacy networks into programmable growth platforms.