Hi everyone, I’ve been working on a computer vision approach to a specific security problem in the "Agentic Economy": identifying malicious transaction patterns that are mathematically obfuscated but topologically distinct. The Problem Traditional rule-based security engines and even standard GNNs o