JINNI Autonomous DeFi Wallet Agent

Managing crypto opportunities is still a manual and time-consuming process. Users must continuously research markets, monitor positions, manage risk, and execute transactions themselves. While AI agents are becoming more powerful, giving them unrestricted wallet access introduces major security and trust concerns.

JINNI solves this problem by introducing a permission-based autonomous wallet agent powered by Venice AI and MetaMask permission architectures.

With JINNI, users define clear spending limits, trading boundaries, and expiration rules. Once permissions are granted, the AI agent can research opportunities, analyze risk, execute approved actions, and monitor positions while remaining fully constrained by user-defined limits.

Key capabilities include:

• AI-powered wallet risk analysis using Venice AI
• Permission-based execution using ERC-7710 / ERC-7715 inspired architectures
• Autonomous DeFi research and token scoring
• Non-custodial vault-based execution model
• Position monitoring with automated exit logic
• Complete activity logging and transaction transparency
• One-click permission revocation at any time

JINNI demonstrates how autonomous agents can safely participate in DeFi without requiring users to surrender full control of their wallets.

By combining AI reasoning, delegated permissions, and on-chain execution, JINNI creates a practical foundation for the future of Agentic Finance.

Progress During Hackathon

The project was built from scratch during the hackathon.

Day 1

• Designed the JINNI architecture

• Defined the autonomous agent workflow

• Built the smart contract structure for delegated execution

• Integrated MetaMask wallet connectivity

• Created the first dashboard prototype

• Implemented vault management logic

Day 2

• Integrated Venice AI for wallet analysis and token research

• Developed the Research Agent for token evaluation and scoring

• Added permission management flows inspired by ERC-7710 / ERC-7715

• Implemented activity logging and transaction tracking

• Added Sepolia testing environment and token faucet

• Built monitoring infrastructure for position management

Final Improvements

• Improved dashboard UI and user experience

• Added documentation, screenshots, setup guides, and deployment instructions

• Published live frontend deployment

• Published backend deployment

• Recorded and uploaded demo video

• Created public GitHub repository and project documentation

Final Result

JINNI evolved from an initial concept into a complete autonomous wallet agent capable of:

• Analyzing wallet risk

• Generating AI-powered trading insights

• Managing permission boundaries

• Executing actions within approved limits

• Monitoring positions

• Providing transparent on-chain activity records

The final product demonstrates a real-world implementation of permission-based autonomous finance using AI and blockchain technologies.

Fundraising Status

JINNI is currently a hackathon-stage project and has not raised any external funding.

The project was independently designed and developed during the hackathon with a focus on validating the concept of permission-based autonomous finance.

Current status:

• Bootstrapped project
• No external investment received
• No grants received
• Open to feedback, partnerships, and future collaboration opportunities

Future roadmap includes expanding the autonomous agent framework, deeper Smart Account integrations, enhanced AI reasoning capabilities, and production-ready permission management infrastructure for on-chain financial automation.

At this stage, the primary goal is product validation, community feedback, and demonstrating the potential of agentic finance powered by AI and blockchain technology.