Vicci-Web3

Core Concept VICCI is a decentralized visitor information system that uses AI agents to connect blockchain users with relevant protocols through targeted rewards and recommendations. Think of it as a smart concierge service for Web3. Key Components

1. Multi-Agent System The system consists of three primary AI agents: Campaign Agent Helps protocols design and deploy targeted promotional campaigns Creates smart contracts that generate EIP-712 signed reward permits Uses Claude, Langchain, OnchainKit, and AgentKit to optimize campaign strategies Analyzes campaign performance and adjusts targeting parameters Visitor Agent Acts as a personal Web3 concierge for users Analyzes wallet history and on-chain behavior to understand user preferences Provides personalized recommendations for protocols and opportunities Matches users with relevant reward permits based on their profile Indexer Agent Powers the data intelligence layer using The Graph protocol Maintains a comprehensive database of user interactions and protocol activities Feeds data into a Cohere RAG pipeline for enhanced analysis Supports decision-making for both Campaign and Visitor agents
2. Value Proposition For Users: Personalized Web3 guidance based on their interests and history Access to exclusive rewards and discounts from protocols Privacy-preserving preference sharing (only when claiming rewards) Curated discovery of relevant blockchain opportunities For Protocols: Targeted user acquisition based on demonstrated on-chain behavior Cost-effective marketing (pay-per-claim model) Rich analytics on campaign performance Access to qualified users with verified interest
3. Technical Architecture Data Flow: Protocols share their target audience and reward offerings Users connect their wallets and optionally share preferences Indexer Agent aggregates on-chain data and protocol information Visitor Agent matches users with relevant opportunities Campaign Agent generates and manages reward permits Users claim rewards through EIP-712 signed messages
4. Innovation Highlights AI-Powered Matching: Uses advanced AI to understand both user behavior and protocol needs On-Chain Verification: All rewards and claims are verifiable on-chain Privacy-First: Users maintain control over their data and only reveal information when claiming rewards Scalable Infrastructure: Built on Base with Graph Protocol integration for efficient data indexing Composable Design: Uses standardized components from OnchainKit for seamless integration
5. Future Potential Creation of a rich dataset for Web3 user behavior Enhanced protocol discovery mechanisms Development of sophisticated targeting algorithms Cross-chain expansion opportunities Integration with traditional marketing systems VICCI represents a novel approach to protocol growth and user acquisition in Web3, creating a win-win ecosystem where users discover relevant opportunities while protocols find qualified users efficiently.

Core Architecture

1. Microservices Infrastructure The project is built using a microservices architecture with Docker Compose, consisting of:
2. Technology Stack Breakdown Backend Services API Service Built with Fastify for high-performance API endpoints Integrates AgentKit and OnchainKit for blockchain interactions Uses LangChain for AI agent orchestration Implements WebSocket support for real-time updates Uses Prisma as ORM with PostgreSQL Indexer Agent Utilizes The Graph Protocol for blockchain data indexing Implements GraphQL Mesh for data aggregation Uses LangChain with Cohere for data processing Integrates with RabbitMQ for event-driven architecture Frontend Next.js Application Uses OnchainKit (@coinbase/onchainkit) for Web3 interactions Implements comprehensive UI components with Radix UI Uses TailwindCSS for styling Integrates wagmi and viem for blockchain interactions Uses React Query for data fetching and caching
3. Notable Integrations AI/ML Stack Blockchain Stack
4. Interesting Technical Solutions Vector Database Implementation Used pgvector for efficient similarity searches and embeddings storage, crucial for the AI recommendation system. Message Queue Architecture Implemented RabbitMQ for reliable communication between services and async processing of blockchain events. Multi-Database Setup Clever use of multiple databases for service isolation while maintaining data consistency.
5. Partner Technologies Integration Coinbase Technologies OnchainKit: Provides robust Web3 components and utilities AgentKit: Enables sophisticated Web3 agent behaviors AgentKit-Langchain: Bridges AI capabilities with blockchain functionality The Graph Protocol Used for efficient blockchain data indexing Enables complex queries across multiple protocols Powers the recommendation engine with historical data
6. Notable Technical Challenges Agent Orchestration Created a complex multi-agent system with different specialized roles, coordinating through LangGraph. Real-time Updates Implemented WebSocket connections for real-time updates on permit status and recommendations. Environment Management Managed complex environment configurations across multiple services while maintaining security.
7. Development Workflow The project uses pnpm workspaces for package management: This architecture allows for rapid development while maintaining scalability and reliability, with each component serving a specific purpose in the larger ecosystem.