project screenshot 1
project screenshot 2
project screenshot 3

MeowFi

MeowFi is an AI agent designed to optimize liquidity pool strategies in DeFi. By combining on-chain data analysis with machine learning, MeowFi helps liquidity providers maximize yield and manage risk across Layer 2 ecosystems like Arbitrum and Base.

MeowFi

Created At

Agentic Ethereum

Winner of

The Graph

The Graph - Best Use of The Graph with an AI Agent 🥉 3rd place

Project Description

MeowFi is an advanced AI agent created to enhance the liquidity provider (LP) experience within DeFi. It specializes in liquidity pool analysis and strategy optimization across decentralized exchanges (Dexes) like Camelot, Uniswap, and Aerodrome. MeowFi uses machine learning to analyze over 30 parameters, including volatility, liquidity concentration, and fee APR, ensuring that liquidity providers (LPs) can achieve the highest possible returns while effectively managing risk.

With its ability to dynamically adapt strategies based on real-time market conditions, MeowFi is a versatile tool for novice and experienced LPs. The system supports cross-chain optimization and offers native integration with Arbitrum and Base networks, with plans to expand further.

Key features of MeowFi include:

  1. AI-Driven Pool Selection: Automatically selects optimal liquidity pools based on real-time data. Risk-Adjusted Yield Scoring: A proprietary system for balancing APY with impermanent loss risk.
  2. Advanced Analytics: Volatility-weighted ranges, liquidity health metrics, and machine learning models for precise predictions.
  3. Auto-Compounding Strategies: Provides users with automatic reinvestment of earnings to maximize returns.
  4. Leverage Optimization: Suggests dynamic leverage adjustments (1-50x) based on the market’s volatility.
  5. Natural Language Interface: Allows users to interact with complex strategies using plain English.

How it's Made

Core AI functionalities are built using the CrewAI Multi-Agent System and GPT-4o Mini Language Model. We have also built a custom ML model for volatility prediction. Utilizing The Graph protocol to fetch the data from Camelot (Arbitrum) & Aerodrome (Base) subgraphs and Pandas/Numpy for Analysis.

The frontend interface is deployed using Streamlit with the use of custom CSS visualization.

background image mobile

Join the mailing list

Get the latest news and updates