ZyFAI

What it is

ZyFAI is an AI-powered DeFi yield agent that helps users earn better returns by automatically moving funds across lending protocols. Instead of forcing users to chase APYs manually across half a dozen tabs, ZyFAI continuously observes the market and rebalances when the math justifies it.

What I built

• Protocol data infrastructure — APY, TVL, liquidity, utilization, incentives, risk signals, and historical performance across Aave, Morpho, Silo, Euler, Moonwell, and more.
• Rebalancing logic — decides when funds should move, when they should stay, and how to compare higher yield against risk, liquidity, and cost.
• Portfolio tracking and earnings systems — calculate user performance even when funds continuously move between protocols and chains.
• Smart-wallet execution flows — session keys, GuardExecModule, and TargetRegistry to restrict automated execution to approved DeFi actions and trusted protocol targets.
• Backend pipelines and Web3 infrastructure — real-time APY tracking, automated rebalancing, reliable execution.
• Internal models and tools — lifetime earnings, extra rewards, points system, Merkl-based incentives, rebalance backtesting.
• AI-agent workflows — LangChain + LangGraph + PGVector so users can analyze DeFi opportunities and execute actions through chat.

Why it's interesting

DeFi yield optimization is a deceptively hard problem: APYs lie, protocols have hidden risk, and execution costs can eat any edge. ZyFAI treats it as a continuous control problem rather than a one-shot deposit, and pairs an AI front-end with hardened smart-wallet execution so the agent only has the permissions it needs to do its job.