Bringing agentic service assurance closer to autonomous networks

PRISM-AI Phase II brings agentic service assurance closer to autonomous networks

As communications service providers move toward autonomous operations, one of the biggest challenges is ensuring service quality across increasingly distributed, multi-domain and multi-operator environments. Modern telecom services can span edge cloud, IT systems, transport networks such as All Photonic Networks, and home Wi-Fi, often managed by different operators and orchestration systems. That makes it difficult to gain end-to-end visibility, correlate performance data, detect degradation in real time and coordinate corrective action across organizational boundaries.

PRISM-AI – Phase II, a TM Forum Open Innovation Catalyst project showcased at DTW Ignite in Copenhagen in June 2026, is addressing this challenge by combining AI-driven analytics with standards-based orchestration to automate service assurance and recovery across multiple CSPs and technology domains. The project builds on the concept of agentic AI for intelligent fault correlation and autonomous resolution, extending it into a holistic closed-loop approach across business, service and resource layers.

The challenge: from fragmented assurance to cross-domain autonomy

Today, service assurance is still too often reactive and manual. When an issue affects an end-to-end service, operations teams may need to piece together information from multiple domains and operators before they can identify the root cause and take action. In multi-CSP scenarios, that coordination can be complex and slow, increasing mean time to repair and limiting the scalability of assurance processes.

For CSPs, the commercial pressure is clear. Service outages and performance degradation directly affect customer experience and revenue, particularly for OTT services where users expect consistent quality, low latency and reliable bandwidth. At the same time, operators must collaborate more effectively to deliver end-to-end services, but often lack standardized and interoperable mechanisms to do so efficiently.

The solution: collaborative AI agents and standards-based orchestration

PRISM-AI Phase II introduces collaborative AI agents that operate across federated CSP environments. These agents are designed to detect service degradation, identify root causes and recommend or execute corrective actions autonomously. The Catalyst integrates multiple Service Order Managers to coordinate multi-CSP access orchestration, while domain-specific orchestrators, including All Photonic Network controllers, support control across transport and network domains.

By bringing together edge cloud, IT and network resources, All Photonic Networks and home Wi-Fi domains, the project aims to enable proactive issue detection and automated recovery without manual intervention. For example, when service performance begins to degrade, the solution can identify the affected domain and reroute traffic through an optimized alternative path to maintain service continuity and improve OTT user experience.

Importantly, PRISM-AI Phase II was demonstrated as a live implementation rather than a conceptual design. At DTW Ignite 2026, ODA-based services, AI capabilities and orchestration functions were actively running and interoperating across a heterogeneous environment involving multiple vendors and CSPs. Rather than showcasing future concepts through presentations or architecture diagrams, the Catalyst demonstrated real-time service assurance and recovery scenarios using integrated systems, standards-based interfaces and working network functions. This practical implementation highlighted the complexity of achieving interoperability across independently developed solutions and showcased how collaborative integration can support autonomous operations in real-world environments.

Stefano Talamo, Technology Plans and Innovation Programs at FiberCop, and Norihiro Fukumoto, Core Researcher at KDDI, explain that PRISM-AI Phase II demonstrates how AI-driven, standards-based orchestration can transform complex multi-domain, multi-operator networks into intelligent, autonomous ecosystems. They say the Catalyst is valuable because it enables proactive multi-domain service assurance and automated recovery across CSP boundaries, improving service reliability and customer experience while reducing operational complexity and cost for operators.

Using TM Forum assets to support interoperability

The project uses TM Forum assets to provide a common foundation for interoperability, data consistency and autonomous operations. These include Open Digital Architecture, TM Forum Open APIs, including Service Ordering and Service Inventory APIs, the Autonomous Networks Framework, the AI Management and Governance Framework, service orchestration models aligned with TM Forum standards, and the Information Framework (SID) to support consistent data across domains.

Using these assets helps the Catalyst move beyond isolated automation toward a standards-native model for multi-CSP coordination. Delivering the demonstration required extensive use of TM Forum frameworks, Open APIs and architectural principles to connect independently developed capabilities across operators, technology domains and vendors. While integrating solutions from multiple vendors is challenging in itself, PRISM-AIPhase II extended this further by enabling end-to-end interoperability across multiple CSP environments. The project team also used complementary industry-standard APIs beyond the TM Forum framework where required to facilitate interoperability between participating operators. The resulting implementation demonstrated how standards-based collaboration can support autonomous assurance and recovery at scale while providing a practical pathway toward Autonomous Networks Level 4 and above.

Business impact and industry benefits

According to the project team, PRISM-AI Phase II has the potential to reduce operational costs by automating fault detection and resolution, with estimated OPEX savings of 20% to 30%. Faster service recovery could reduce mean time to repair by 40% to 60%, while standardized interfaces can improve the efficiency of collaboration between operators.

For the wider industry, the Catalyst supports interoperable orchestration across federated networks, promotes secure and scalable AI collaboration across organizational boundaries, and accelerates the adoption of higher levels of autonomous operations. For end users, the benefits could include more resilient digital services, reduced latency, improved bandwidth reliability and a better overall experience through proactive issue resolution.

The project team summarizes the business impact as reducing operational costs and improving service recovery times through AI-driven autonomous network orchestration across multi-domain, multi-CSP environments. Beyond the operational benefits, the Catalyst demonstrates that complex interoperability challenges across vendors, domains and operators can be addressed through a combination of TM Forum standards, AI-driven automation and collaborative integration. In practical terms, that means helping operators move from fragmented, reactive assurance to intelligent, closed-loop operations that can protect customer experience at scale while accelerating the industry's transition toward autonomous networks.