Using digital twins and agentic AI to enable level 4+ autonomous network operations

Commercial context

Recent research from STL Partners suggests that the shift to autonomous networks could result in more than $800 million in annual value for CSPs. These gains come from a combination of CAPEX savings, OPEX reductions, and new revenue growth. However, the most significant uplift—about 30 percent—comes only when CSPs reach levels 4 and 5 of the TM Forum’s Autonomous Network model. At these levels, networks can monitor themselves, act on intent, and recover from faults without operator intervention.

The 'BIND' Catalyst—short for Bridging Intelligence, Networks and Digital Twins—aims to make that leap. It proposes a standards-based framework to move CSPs toward level 4+ autonomy using AI, digital twins, and agentic AI-driven automation. The goal is not just to automate tasks, but to redesign how networks operate—so they can anticipate problems, respond intelligently, and serve customers more effectively.

The solution

The project team focused on building a flexible data and AI foundation that ingests inputs from siloed inventory and assurance systems. These include service and resource inventory, alarm and performance data, service quality metrics, and topology insights. All are aligned with TM Forum Open APIs such as TMF638, TMF639, TMF642 and TMF628.

This foundation enables the creation of digital twins that accurately reflect network and service states, updated in real time. The digital twins do more than just visually represent the network—they enable analysis, prediction, and decision support through simulation.

A layer of genAI-powered network agents runs on top of the system. These agents apply domain-specific knowledge and collaborate using Google’s open Agent-to-Agent (A2A) Protocol. Over 50 organizations support and contribute to this standard to ensure interoperability. The agents proactively identify service-affecting issues, simulate potential responses and apply them to maintain the customer intent.

Rather than waiting for complaints, the solution addresses anomalies before the customer feels the impact. Agents recommend, simulate and execute actions, maintaining quality of service while reducing the need for human oversight. This preemptive, intent-driven mode of operation represents a fundamental shift from today’s reactive models.

The project also aligns closely with TM Forum assets such as the Autonomous Operations Maturity Model (GB1042), Digital Twin for Decision Intelligence (IG1307, IG1310), and GenAI Use Cases (IG1369). These standards provide a shared framework for industry-wide adoption and scale.

Wider application and value

For customers, the most immediate gain is enhanced experience. Fault prediction with up to 90% accuracy ensures fewer service disruptions. End-to-end process automation—up to 95% in some use cases—supports seamless, on-demand services without delays or escalations.

For CSPs, the impacts are significant. Automated assurance reduces operational costs by up to 40% and cuts mean time to repair (MTTR) by 50%. CSPs can also reduce truck rolls—an expensive, environmentally taxing activity—and save around $150 per roll. They gain better visibility, resolve issues faster, and deliver more consistent service. This, in turn, boosts Net Promoter Scores (NPS), reduces churn, and improves agility.

The solution also simplifies internal workflows. AI-based prompt engineering and no-code configuration let employees interact more naturally with systems. Staff can focus on high-value tasks instead of repetitive troubleshooting.

According to Andreas Quiskamp, Senior Manager Network Engineering - Inventory, Discovery & Asset Management of Vodafone: "The commoditization of network services has resulted in a reduction in average revenue per user (ARPU) for conventional offerings and increased susceptibility to service interruptions. Meanwhile, massive network infrastructure investment is required to support new technologies and rising demand, with internet traffic up by 60% compared to pre-pandemic levels."

Optimised networks reduce energy consumption and carbon emissions, especially through fewer truck deployments and smarter resource usage. Network reliability improves for essential services—critical in an era where digital infrastructure supports everything from public safety to remote healthcare.

Moreover, by freeing telco talent from manual processes, the industry can redirect skills toward innovation and service development. This helps CSPs play a more vital role in national digital agendas and accelerate broader digital transformation goals.

BIND demonstrates that with the right combination of digital twin architecture and interoperable AI agents, higher-level autonomy is no longer aspirational—it’s within reach. Boosting customer satisfaction through 90% fault prediction precision and cutting assurance costs by 40% isn’t just a technical achievement. It’s a blueprint for how the telco industry can scale smarter, serve better, and step confidently into a truly autonomous future.