Democratizing DeFi trading. Create trading strategies via a chatbot -- no programming required!
Prize Pool
The chatDeFi.app AI agent democratizes DeFi by enabling anyone to create and execute DeFi investment strategies by simply typing in their desired DeFi investment strategy into a user-friendly chatbot interface. Our AI agent then executes these strategies automatically.
For example, you could instruct our AI agent to:
“Please allocate my wallet funds to the highest yielding DeFi lending pools that earn at least 5% APY. Also, to manage my risk, split my wallet funds across 2 lending pools with 70% of my funds in one pool and 30% in another. If my portfolio loses 10% of its funds, then stop managing my portfolio.”
Our AI agent will parse your natural language request into the corresponding DeFi strategy and actively manage and rebalance your portfolio according to the parameters you define.
Key Components
Key Features
Future Roadmap
Our project is a decentralized platform that allows users to create and publish personalized vaults based on their own investment strategies. Other users can then deposit into these vaults, and the management of funds is delegated to an AI agent. This agent operates within predefined rules set by the vault creator and autonomously executes DeFi actions on their behalf.
One of the standout features of this platform is the ability for users to customize strategies by setting specific constraints and rules. The AI agent adheres strictly to these parameters, ensuring personalized yet autonomous vault management.
Architecture & Technology Stack Smart Contracts: Each vault is represented by a dedicated smart contract deployed when a user creates a new strategy. These contracts can be updated or rebalanced based on changing strategies. The agent periodically evaluates the vault’s strategy and triggers rebalancing actions as needed.
AI Agent: We utilized the Elize framework to build our DeFi-native AI agent. It has been fine-tuned for our use case to interact with major DEXs and lending protocols. The agent can:
Dynamically generate calldata for protocol interactions.
Execute contract functions using a generic execute() function.
Handle complex transaction flows such as approving tokens before depositing.
The agent continuously seeks to maximize yield. For instance, if it detects higher APY opportunities on Aave for a different asset, it automatically swaps the current asset and reinvests into the higher-yield pool—always staying within the creator-defined strategy.
Frontend: We used Next.js with TypeScript and React to develop the user interface, providing an intuitive experience for creating, customizing, and monitoring vaults.
Blockchain Integration: The platform is deployed across multiple chains to ensure accessibility for a broad user base.
Storage: We use IPFS to store and retrieve strategy metadata in a decentralized manner.
Smart Contract Development: We leveraged the Foundry framework for writing, testing, and deploying our contracts efficiently.
Notable Hacks & Challenges We initially explored using Crossmint’s agent wallet for full wallet delegation to the AI agent, but integrating it within the limited timeframe of a 2-day hackathon proved too complex. However, the flexible execute() function in our contract allowed the AI to operate across diverse DeFi protocols without needing extensive additional integrations.
Architectural Diagram https://excalidraw.com/#room=dbf02233d7f50bceb730,J-UOJTDqkBcGEYCBMvabiw