Moonshot Catalyst - This Catalyst is empowering self-driving networks with graph AI and digital twins for predictive root cause analysis and Level 4 autonomy. Their solution unlocks the "zero-X" experience - zero wait, zero touch, zero trouble.
Project companies
Same fault. Different response.
Not all network failures carry the same business risk. A performance degradation in a revenue-dense cell represents a materially different operational priority than one in a low-traffic zone. Yet today's autonomous networks have no mechanism to express or act on that difference.
Isolating contributing factors across RAN, Core, and Transport remains a manual and siloed process. Every fault enters the same resolution queue regardless of business impact.
Business-Aware GNN-Healing Networks changes this.
The Catalyst introduces a three-layer closed-loop architecture:
Business Intent Layer Sequences remediation based on infrastructure business criticality and revenue exposure.
Graph Intelligence Layer A Graph Neural Network powered by Google Cloud's SpannerDB Temporal Digital Twin traverses cross-domain topology to identify remediation branches from a single service-layer trigger. No alarm is required.
Execution Layer Standards-based service-layer orchestration. Healing intent travels in TMF921 and decomposes through TMF641 service orders into per-domain actions across RAN, Core, and Transport.
The key innovation is the first Business Intent Extension Model (BIEM) contributed to the TM Forum Intent Ontology (TIO). It gives operators a standards-based vocabulary to declare which remediation branch executes first when recovery resources are constrained, decoupling business priority from network implementation.
The result: same fault, different response.
Level 4 autonomy across RAN, Core, and Transport, on a Google Cloud digital twin, with the business layer exploring the Level 5 boundary through declarative, intent-driven prioritization.
The Catalyst targets:
* More than 95% silent-degradation lead-time compression: detection in sub-minute windows instead of hours (Phase 1 target)
* More than 80% cross-domain triage labour reduction per event, USD 1.08M per tier-1 operator per year (Phase 1 target)
* +$2.475M customer market share revenue uplift per year, protecting revenue at risk in critical coverage zones (champion-validated)
Resources
1.0 Introduction
Project Introduction
2.0 infographic
Infographic
3.0 Demo
The Full Demo
4.0 First business-oriented ANLET
TM Forum contribution - GB1515 First Business Oriented Scenario ANLET
5.0 Article
Catalyst Article
6.0 GNN Model Training
GNN Training Approach Appendix
7.0 TMF641 execution
Decomposition and Config by Ericsson Service Orchestration & Assurance Engine
8.0 Closed Loop Healing
Closed Loop Healing
Contact team
Email the members of the Catalyst team to request more details.