Hypergraph AI

Hypergraph AI is a cybersecurity company developing an automated Security Operations Center (SOC) platform based on Graph Neural Networks (GNNs) for network threat detection. The company originates from a research group at the Polytechnic University of Catalonia (UPC) in Spain, with over 8 years of research in GNNs applied to network analysis.

The core technology models network infrastructure as a graph topology, analyzing relationships (edges) between devices (nodes) to detect complex attack patterns. This approach differs from traditional AI methods that process sequential log data, as the graph-based model captures structural and contextual relationships across the network.

Key capabilities include:

• Agentless deployment via NetFlow/IPFIX and network-level traffic analysis, requiring no software installation on endpoints
• A pre-trained foundational model designed to generalize across different network environments without per-client fine-tuning
• Automated Level 1 SOC analysis, including alert correlation, false positive reduction, and mitigation action proposals
• Detection of lateral movement, anomalous behavior, and complex multi-stage attack patterns

Target sectors include mid and large enterprise IT, medical IoT (e.g., MRI machines, infusion pumps), and industrial/OT environments with legacy SCADA systems and PLCs. The agentless architecture is specifically positioned for environments where agent-based tools cannot be deployed.

The company is in an early access stage and is targeting MSSPs, SMEs, medical and industrial organizations, and investors. It was a winner of the Wayra by Telefónica 2025 incubation program.