AutoTrust PayMesh — MNEE Escrow Settlement System

The Economic Operating System for Autonomous AI Agents. Trustless, programmable financial coordination between autonomous agents using MNEE stablecoin.

Comment

Inspiration

The future of commerce isn't just digital—it's autonomous. As AI agents become increasingly capable of making economic decisions, we identified a critical gap: there's no trustless infrastructure for AI-to-AI financial transactions. Traditional payment systems require human oversight and trust intermediaries. We envisioned AutoTrust PayMesh as the Economic Operating System for autonomous AI agents, enabling programmable money flows with built-in escrow, negotiation, and streaming payment capabilities.

The MNEE hackathon challenged us to build real-world applications for programmable money, and we saw an opportunity to create the foundational layer for an entire ecosystem of AI-driven commerce.

What it does

AutoTrust PayMesh is a production-ready escrow settlement system built on the MNEE ERC-20 stablecoin. It provides:

  • Smart Escrow Contracts: Lock MNEE tokens with deterministic release/refund rules enforced by smart contracts
  • AI-Powered Verification: Multi-agent systems (Compliance, Operations, and Arbiter agents) autonomously verify transaction conditions using Groq's Llama 3.3-70b
  • Negotiation Sandbox: AI-to-AI price negotiation engine where agents haggle and reach consensus before executing transactions
  • Streaming Payments: Real-time, per-second payment streams for ongoing services (e.g., 0.05 MNEE/sec)
  • Complete Audit Trail: Backend indexer captures all on-chain events with an Ops Log API for full transparency
  • Production Deployment: Fully containerized services deployed on Google Cloud Run with live demo URLs

How we built it

Smart Contract Layer (Solidity 0.8.24)

  • Implemented MNEEEscrow.sol with OpenZeppelin's battle-tested libraries
  • Built a rigid state machine (Funded → Locked → Release/Refund) preventing invalid transitions
  • Integrated reentrancy guards and pausable functionality for security
  • Native MNEE token integration with streaming payment logic

Backend Intelligence Layer (Node.js/Express)

  • Integrated Groq API (Llama 3.3-70b-versatile) for ultra-low latency AI inference
  • Built RAG (Retrieval Augmented Generation) system with specialized knowledge base for policy enforcement
  • Created multi-agent orchestration managing Compliance, Operations, and Arbiter agents
  • Implemented event indexer using ethers.js v6 to capture and serve transaction history

Frontend Console (Next.js 14)

  • Built with wagmi v2 and viem for robust Web3 integration
  • Implemented dark/light themes with i18n support (EN/ES)
  • Created real-time balance tracking and escrow status monitoring
  • Designed Negotiation Sandbox with client-side agent simulation

Cloud Infrastructure

  • Containerized all services (frontend, backend, blockchain node) with Docker
  • Deployed to Google Cloud Run for scalable, serverless execution
  • Configured custom networking for inter-service communication
  • Set up CI/CD pipeline for rapid iteration

Challenges we ran into

  1. Smart Contract State Management: Ensuring atomic state transitions while supporting both instant and streaming payments required careful design of the escrow state machine.

  2. AI Agent Coordination: Orchestrating multiple AI agents to reach consensus on transaction verification without introducing single points of failure was complex. We solved this with confidence scoring aggregation.

  3. Real-time Event Indexing: Keeping the backend in sync with on-chain events while handling potential reorgs and missed blocks required robust error handling and recovery mechanisms.

  4. Cloud Deployment Complexity: Containerizing a full blockchain node alongside application services and managing inter-service networking in a serverless environment presented significant DevOps challenges.

  5. Gas Optimization: Balancing feature richness with gas efficiency, especially for streaming payments that could potentially execute many small transactions.

Accomplishments that we're proud of

✅ Full Production Deployment: Live demo with working frontend, backend, and blockchain node on Google Cloud Run

✅ AI-Native Architecture: First-class integration of multi-agent AI systems for autonomous transaction verification

✅ Zero Trust Design: Completely non-custodial system where users maintain full control of their private keys

✅ Comprehensive Documentation: Detailed README with architecture diagrams, workflow explanations, and quick-start guides

✅ Real Innovation: Built novel features like AI-to-AI negotiation and streaming payments that don't exist in traditional escrow systems

✅ Security First: Implemented OpenZeppelin standards, reentrancy guards, and role-based access control

What we learned

  • Smart Contract Design Patterns: Gained deep understanding of escrow mechanics, state machines, and security best practices in Solidity

  • AI Agent Orchestration: Learned how to design multi-agent systems that can make financial decisions with appropriate confidence thresholds

  • Event-Driven Architecture: Mastered patterns for building reliable event indexing systems that bridge on-chain and off-chain worlds

  • Production-Grade Web3: Discovered the complexities of deploying full blockchain infrastructure in cloud environments

  • RAG System Design: Built practical retrieval-augmented generation systems for domain-specific AI applications

What's next for AutoTrust PayMesh

Phase 1: Mainnet & Oracles

  • Deploy to Ethereum mainnet with MNEE token integration
  • Implement Work Validator Oracles for automated service delivery verification
  • Build agent reputation system based on transaction history

Phase 2: Developer Ecosystem

  • Launch mobile app for iOS/Android
  • Create API Marketplace where AI agents can discover and purchase services
  • Release enterprise SDK for white-label solutions

Phase 3: Cross-Chain Expansion

  • Support Arbitrum, Base, and other L2s for lower gas fees
  • Implement cross-chain escrow using bridge protocols
  • Build unified liquidity pools across chains

Phase 4: Advanced AI Features

  • Multi-party escrow for complex supply chains
  • Predictive dispute resolution using historical data
  • Automated contract negotiation with game-theory-optimal strategies

AutoTrust PayMesh represents our vision for the future of autonomous commerce—where AI agents can transact securely, negotiate fairly, and build reputation through verifiable on-chain history. This is just the beginning of the Economic Operating System for AI.

Built With

Share this project:

Updates