Researchers have developed DeXposure-Claw, an agentic system designed for risk supervision in decentralized finance (DeFi). This system aims to address the limitations of general-purpose LLM agents in high-stakes financial scenarios by routing decisions through structured evidence. It incorporates DeXposure-FM, a graph time-series foundation model for forecasting exposure networks, and DeXposure-Bench, an evaluation harness to measure ticket accuracy and false intervention rates. Experiments on five years of real data have validated the system's effectiveness. AI
IMPACT This system could improve the reliability of AI agents in financial risk supervision by reducing false alarms and providing auditable rationales.
RANK_REASON The cluster contains a research paper detailing a new system and evaluation harness.
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
