LiquidAI

LiquidAI is an autonomous liquidity orchestration protocol that uses AI reasoning and on-chain data to actively manage DeFi liquidity positions. It continuously analyzes APR, volatility, and risk across pools, then reallocates liquidity intelligently to maximize yield and minimize risk — just like a self-driving portfolio manager for DeFi.

Problem

DeFi users constantly face:

Fragmented liquidity across chains and pools

Volatile markets that make manual rebalancing inefficient

Lack of transparency in why liquidity decisions are made

Static yield strategies that fail to adapt to real-time data

Manual liquidity management is error-prone, time-consuming, and often leads to missed opportunities or unnecessary risks.

💡 Solution — LiquidAI

LiquidAI introduces an AI-powered liquidity manager that automates portfolio rebalancing across DEX pools. It integrates:

Pyth Network for live price feeds

Envio HyperIndex for on-chain liquidity metrics

MeTTa reasoning (from ASI) for intelligent decision-making

Hardhat 3 for testing, deploying, and simulating smart contracts

Users interact through a dashboard and AI chat interface that transparently explains every liquidity move, confidence level, and reasoning rule behind it.

How We Built LiquidAI – The Nitty-Gritty

LiquidAI is an AI-driven liquidity orchestration protocol that combines real-time data, reasoning-based AI, and a live dashboard for portfolio management. It’s built around three layers — Data, AI, and Frontend — stitched together for real-time interactivity and explainability.

Tech Stack Overview

1. Data Layer

Pyth Network for real-time price feeds

Envio HyperIndex for on-chain pool metrics

Uniswap V3 APIs for APR, TVL, and liquidity stats

Aggregated via Next.js API routes, updated every 3 seconds to simulate live market data

Hacky Part: We faked a real-time trading environment entirely in the browser using React state updates and interval-based “pseudo-streaming.”

1. AI Layer (MeTTa Reasoning Engine)

Custom-built mettaReasoner.ts mimics rule-based symbolic reasoning (e.g., volatility_high → rebalance_to_stables).

Generates confidence scores and actions in real time, shown on the dashboard.

Hacky Part: Built a lightweight MeTTa-like engine from scratch in JS to simulate explainable decision logic live in the browser.

1. Smart Contracts

LiquidityVault.sol (Solidity 0.8.24) handles deposits and strategy execution.

Integrated OpenZeppelin libraries for ERC-20 safety and compliance.

Deployed via Hardhat with mock price updates from Pyth scripts.

1. Frontend (Next.js + Tailwind)

Next.js 14 + TypeScript for fast, modular frontend

TailwindCSS + Framer Motion for sleek UI/animations

“Demo Mode” simulates complete AI-driven rebalancing workflows client-side

AI Agent Architecture

Observer Agent – Collects and analyzes data

Allocator Agent – Runs MeTTa reasoning and suggests allocations

Executor Agent – Simulates or executes transactions

They communicate through lightweight Node.js event emitters for real-time updates.

🚀 What’s Notable

Real-time simulation without backend

Custom symbolic AI engine in JS

Transparent “reasoning cards” showing how AI makes decisions

Fully modular AI-agent design

End Result

LiquidAI is a self-contained, AI-powered liquidity manager that reacts to market changes in real time, explains its reasoning, and demonstrates how AI + DeFi automation can coexist with human transparency and control.