ChainPilot AI

• 

  Control your wallet using natural language to send funds, manage contacts, and teams.

• 

  Real-time portfolio insights, spending trends, and blockchain activity tracking.

• 

  Control your wallet using natural language to send funds, manage contacts, and teams.

• 

  Automate transfers with date, amount, token, and built-in security validation.

• 

  Save and manage wallet addresses manually or through AI chat commands.

• 

  Generate QR codes and shareable links for transaction verification.

• 

  AI-powered smart contract analysis with risk scoring and on-chain audit proof.

ChainPilot AI

Autonomous AI Risk Intelligence for Web3

1) Inspiration

Web3 unlocks financial freedom — but it also exposes users to massive risk.

Every week, we see:

• Smart contract exploits
• Rug pulls
• Malicious token contracts
• DeFi protocol vulnerabilities

Despite billions lost annually, most users still interact with smart contracts blindly.

Even experienced developers struggle to:

• Read raw bytecode
• Interpret proxy patterns
• Evaluate reentrancy risks
• Simulate transactions safely

We asked a simple question:

What if Web3 had an AI co-pilot that thinks before you transact?

ChainPilot AI was built to become that autonomous security layer.

What It Does

ChainPilot AI is an AI-powered Web3 security and transaction intelligence platform that:

• Converts natural language into blockchain actions
• Performs static contract vulnerability analysis
• Generates AI-powered human-readable risk explanations
• Simulates transactions before execution
• Logs audit hashes immutably on-chain
• Monitors contracts continuously every 5 minutes

It’s not just a scanner. It’s an AI security operator.

How We Built It

🔹 Frontend

• Next.js 14 + TypeScript
• Tailwind CSS
• Privy wallet authentication
• AI chat interface powered by Gemini

🔹 Backend (Modular Service Architecture)

Built with Node.js + Express + MongoDB.

Six core services:

• ContractAnalyzer → Static bytecode + source inspection
• RiskEngine → Weighted vulnerability scoring
• AiEngine → GPT-4o security explanations
• TransactionSimulator → Pre-execution risk modeling
• BlockchainLogger → On-chain SHA-256 audit logging
• MonitoringEngine → Cron-based automated surveillance

🔹 Autonomous Agent

Built using ElizaOS framework.

The agent can:

• PREPARE_TRANSFER
• CHECK_RISK
• CREATE_TEAM
• Parse natural language into blockchain workflows

This transforms ChainPilot from a tool into an autonomous AI operator.

Smart Contract Design

We deployed an immutable audit registry (AuditLog.sol) on Sepolia.

Functions include:

• logAudit()
• verifyAudit()
• totalAudits()
• getAuditHashes()

Security Model:

• Only a trusted backend logger can write
• Audit hashes are immutable
• Emits AuditLogged event for indexing

Each security report is hashed using:

H=SHA256(report_data)

This ensures cryptographic integrity and verifiability.

How It Works (Full Flow)

1. User connects wallet via Privy
2. User types: analyze contract 0x...
3. Gemini parses intent
4. Backend fetches contract bytecode
5. riskEngine evaluates 15 vulnerability patterns
6. GPT-4o generates a structured explanation
7. Hash of report optionally stored on-chain
8. Monitoring engine re-checks contract periodically

Risk Score is computed using a weighted model:

RiskScore=∑(Severityi​×Weighti​)

Output: 0–100 vulnerability index.

Challenges We Faced

• Designing a scalable modular architecture
• Separating risk scoring logic from AI explanation layer
• Ensuring deterministic scoring before AI interpretation
• Secure private key handling for audit logging
• Building a cron-based monitoring engine without spamming RPC
• Coordinating multi-AI pipelines (Gemini + GPT-4o)

One major design decision: We intentionally separated AI reasoning from scoring logic to prevent hallucinated risk values.

What We Learned

• AI should assist deterministic systems — not replace them
• On-chain logging adds real credibility
• Autonomous agents require strict action validation
• Web3 UX improves drastically with natural language interfaces

Most importantly:

Security intelligence must be explainable.

Why It Matters

ChainPilot AI introduces:

• AI-native Web3 UX
• On-chain verifiable security reports
• Continuous contract surveillance
• Agent-based transaction execution

It’s a programmable AI firewall for decentralized finance.

Built With

Languages

• TypeScript
• Solidity
• JavaScript

Frameworks

• Next.js 14
• Express.js
• Tailwind CSS
• Hardhat

AI APIs

• OpenAI GPT-4o
• Google Gemini API

Blockchain

• ethers.js v6
• Sepolia Testnet

Database

• MongoDB

Auth

• Privy

Agent Framework

• ElizaOS

Built With

• content
• devpost
• documentation
• html5css3
• javascript
• json
• language
• markdown
• mongodb
• query
• shell
• solidity
• typescript
• yaml