Autonode

This project is an end-to-end decentralized AI agent system designed to optimize blockchain interactions through automated task processing, leveraging a simple rule-based engine inspired by the classic Eliza model. At its core, the project comprises an AI agent that receives text-based task requests—such as commands to index smart contracts, optimize queries, or batch transactions—and processes these tasks by matching them against a predefined set of patterns to generate appropriate responses. The AI agent is implemented as a lightweight service written in a language like JavaScript (using Node.js) and is initially hosted on Replit, where it operates as a REST API that accepts HTTP POST requests containing task details, processes them using its Eliza-based logic, and returns a response indicating the action taken.

The system integrates with an existing smart contract that serves as the on-chain interface for users. This smart contract allows users to register as agents, submit tasks with associated payments, and track task statuses. When a user submits a task via the smart contract, an off-chain service—acting as a bridge—listens for task-related events (such as TaskSubmitted events) and then forwards the task data to the Replit-hosted AI agent via an API call. Once the AI agent processes the task and produces a response, this off-chain service updates the smart contract with the result, thereby ensuring that the task lifecycle is fully traceable and that payments are appropriately settled.

In addition to the core functionality, the project is designed with future scalability in mind by planning for integration with EigenLayer—a decentralized compute network that will allow the AI agent to be containerized (using Docker) and deployed in a trustless, verifiable manner. EigenLayer integration aims to enhance the system by providing decentralized, secure execution of tasks, possibly incorporating Trusted Execution Environments (TEEs) to ensure that computations are tamper-resistant and fully verifiable on-chain. For the minimum viable product (MVP), the system operates via Replit and a centralized off-chain service, but the project roadmap includes migrating the AI agent to EigenLayer once the initial functionality is proven.

Overall, this project bridges decentralized blockchain infrastructure with automated AI-powered task processing, offering a streamlined and scalable approach to handling complex operations such as smart contract indexing and query optimization. It is a comprehensive solution that not only provides immediate efficiency improvements for blockchain interactions but also lays the groundwork for a fully decentralized and verifiable compute environment in the future.

This project is built around a lightweight, Eliza-based AI agent that processes text commands for blockchain-related tasks such as smart contract indexing, query optimization, and transaction batching.

We developed the AI agent using Node.js and Express, writing a simple rule-based function that matches incoming task commands against predefined patterns and returns corresponding responses.

The agent is hosted on Replit, which allowed us to rapidly iterate and deploy the service as a REST API, making it easy to test and refine our logic in a live environment.

Our existing smart contract—already implemented in Solidity—handles key functions like agent registration, task submission, and payment settlement; this contract emits events whenever a task is submitted.

To bridge the on-chain and off-chain worlds, we built an off-chain service in Node.js using ethers.js to listen for these smart contract events and forward the task data to our Replit-hosted API.

Once the AI agent processes a task, the off-chain service receives the response and interacts with the smart contract to update the task status and release payments, thereby closing the loop between the blockchain and our centralized compute layer.

For future scalability and enhanced decentralization, we plan to containerize the AI agent using Docker and integrate it with EigenLayer—a decentralized compute network—which will enable us to run the agent’s computations in a trustless and verifiable manner.

This integration involves creating a Docker image of our Node.js service and following EigenLayer’s deployment guidelines, including configuring environment variables, resource limits, and potentially integrating Trusted Execution Environments (TEEs) for added security.

One hacky but effective workaround we implemented was using a centralized off-chain service as an intermediary between the smart contract and our AI agent; while not fully decentralized, this approach allowed us to demonstrate end-to-end functionality within the limited timeframe of the hackathon.

Overall, our solution leverages partner technologies like Replit for fast prototyping, ethers.js for blockchain interactions, and Docker/EigenLayer for future decentralized deployment, piecing together a streamlined system that bridges automated AI processing with decentralized finance.