MeTTaPay

MeTTaPay is a smart crypto payment agent that changes how users send USDC payments. It uses natural language processing along with symbolic reasoning. Users can just say, "Send 5 USDC to alice.eth," and the agent takes care of everything, including ENS resolution and transaction execution across multiple chains like Ethereum, Polygon, and Sepolia.

What makes MeTTaPay unique is its use of MeTTa Knowledge Graphs for symbolic reasoning. This allows for efficient caching of ENS names and balances, detection of patterns for safety, and clear decision-making. The agent keeps a knowledge base of facts and rules. This helps it make smart choices about whether payments can be made, spot suspicious patterns, and offer relevant responses.

The system is completely non-custodial. Users keep control of their wallets and keys while the agent acts as an intelligent interface. It has built-in safety features like transaction limits (up to 10,000 USDC), detection of suspicious activity, and required approvals for transactions. Sending crypto is as easy as sending a text to a friend.

This project leverages the Artificial Superintelligence Alliance utilities, specifically using Agentverse as a marketplace for launching the agent and interfacing with ASI1 LLM for natural language understanding.

Core Technologies:

uAgents Framework: Provides the foundational agent architecture and AgentChatProtocol v0.3.0 for message handling. MeTTa Knowledge Graphs: Implements symbolic reasoning with facts like (balance user123 25.5) and rules for decision-making. Web3.py: Handles blockchain interactions across Ethereum, Polygon, and Sepolia networks. ENS Python Library: Resolves ENS names to wallet addresses. Flask: Provides REST API endpoints for external integrations.

Architecture: The agent combines reactive natural language processing with proactive symbolic reasoning. When a user sends "Send 5 USDC to vitalik.eth", the system:

1.Parses the intent using ASI1 LLM. 2.Queries the MeTTa knowledge graph to validate payment feasibility. 3.Resolves the ENS name to a wallet address. 4.Executes the USDC transfer on the specified chain.

Notable Implementations: Smart Caching: MeTTa facts store previously resolved ENS names and user balances, reducing RPC calls. Safety Layer: Symbolic rules detect suspicious patterns like rapid large transfers or unusual recipients. Multi-Modal Interface: Supports both conversational prompts and structured API calls. Knowledge Graph Queries: Custom endpoint /metta-query allows direct querying of the symbolic reasoning system.

The integration with Agentverse enables easy deployment and discovery, while the MeTTa knowledge graphs provide transparency into the agent's decision-making process - a crucial feature for financial applications where users need to understand why certain actions are taken or blocked.