Arova
Proactive AI risk radar for DeFi — one standardized Graph query, every protocol.
Demo
About
AROVA is a proactive AI agent that continuously monitors DeFi lending and DEX protocols — Aave, Compound, Uniswap, and Balancer — for early warning risk signals like sudden TVL drops or unsustainable yield spikes. Instead of manually checking separate dashboards for each protocol, AROVA runs live audits using one reusable query pattern across every protocol it tracks, flags anomalies in plain English, and lets you ask natural-language follow-up questions through a built-in AI Copilot to understand exactly why something was flagged. It also includes a protocol comparison view and a yield calculator, all powered by live on-chain data.
How it's made
AROVA is built with Next.js on the frontend and deployed on Vercel. Its core data layer uses The Graph's Messari Standardized Subgraphs — because lending protocols (Aave, Compound) and DEX protocols (Uniswap, Balancer) share the same schema, AROVA runs a single query per category, unmodified, against every protocol in it, rather than writing protocol-specific query logic for each one. The Subgraph MCP powers the natural-language layer, translating a user's plain-English question into the right live query. Google's Gemini model handles reasoning — turning raw TVL, revenue, and yield deltas into clear, explained risk flags, and answering AI Copilot follow-up questions grounded in the same live data. The trickiest part was designing the signal-detection thresholds to generalize across both lending and DEX categories despite their different underlying metrics, without needing category-specific tuning for every new protocol added.