Next Session #REDeFiNETOMORROW2025 Fireside chat: Safeguarding DeFi: Managing Risk in a Decentralized Economy @omeragoldberg of @chaos_labs @melmelmelting of @TheSpartanGroup Free Ticket:

1/ Chaos Labs CEO @omergoldberg joined @MelodyHe_ from Spartan Group to discuss how they're safeguarding DeFi through advanced risk management. The company powers risk parameters for major protocols like Aave, Jupiter, GMX, dYdX, EtherFi, and Renzo. 2/ Chaos Labs has evolved from manual risk consulting to building three core products: Chaos AI (financial intelligence), Edge (oracles), and Analytics. Their risk platform calculates and streams real-time financial parameters across 20-30+ blockchain networks. 3/ When you interact with protocols like Aave, the loan-to-value ratios, liquidation thresholds, and interest rates are determined by Chaos Labs' AI-powered risk engine. If you've used these protocols, you've likely interacted with Chaos technology under the hood. 4/ Initially, protocol teams had engineers on-call 24/7 to manually update risk parameters. Each change required writing Solidity payloads, getting reviews, and audits - a tedious process for thousands of updates per year across protocols. 5/ This led to building Chaos Oracles with three components: price oracles (with risk context), risk oracles (streaming financial parameters), and proof of reserves. They're the only provider of risk oracles in the market. 6/ Key differentiation: Chaos doesn't just relay price data - they apply business logic and risk assessment. Example: A stablecoin shouldn't trigger mass liquidations from a 3% price spike on a single pool that reverts in one block. 7/ Real impact: Last week, a Chainlink feed on Avalanche caused over $500k in losses on Euler Protocol due to temporary stablecoin price dislocation. Over the years, oracle attacks have resulted in hundreds of millions in losses industry-wide. 8/ Chaos AI emerged from their core technology - an agent-based Monte Carlo engine that evolved to incorporate LLMs for processing unstructured data. They now use 15-16 different models in their AI system, each optimized for specific tasks. 9/ Major announcement: Over 1 million people are on the waitlist for Chaos AI. Initial rollout has begun with 100 users, with broader release planned for summer. Estimated inference costs would be $500M-$1B annually if released to all waitlist users today. 10/ Their approach to decentralization is pragmatic - focus on product-market fit first, then progressively decentralize. As Goldberg notes: "It's all fun and games until it really matters" - referencing the importance of alternatives when centralized systems fail. 11/ On building for crypto vs TradFi: Once data is properly normalized and abstracted, the financial instruments are fundamentally similar. With 10-15% of global crypto market cap now in tokenized assets, any serious AI must understand both worlds. 12/ Chaos is positioning itself as a leader in AFI (Artificial Financial Intelligence) - systems that can reason about both on-chain and traditional finance as first-class citizens, understanding the relationships between them. 13/ With new oracle partnerships to be announced in coming weeks and claims of being the fastest-growing oracle in the market, Chaos Labs is building the risk infrastructure layer that could enable the next wave of DeFi innovation.