AiVS

AiVS (short for AI Agents + AVS) is a framework that leverages the power of AI-driven decision-making in conjunction with decentralized Autonomous/Actively Validation Services (such as EigenLayer) to securely distribute tasks and verify outcomes on-chain. By combining a diverse panel of AI “operators” with a public and tamper-proof ledger, AiVS ensures that any type of task—whether in law, finance, content moderation, or beyond—can be evaluated fairly, transparently, and efficiently.

For this hackathon we created the use case of a fair distributed justice system, which illustrates how AiVS can revolutionize the legal intake and adjudication process, offering a glimpse of its potential across countless domains. Here's how we describe this project:

Imagine a world where filing and adjudicating legal cases is fast, transparent, and secure—where decisions are formed by a committee of diverse AI agents from different jurisdictions and ideologies, and the process is recorded immutably on-chain. That is exactly what this new system aims to achieve by harnessing the power of crypto (via EigenLayer AVS) and AI-driven “juries.”

We've used typescript and solidity for creating the core AiVS (EigenLayer AVS) and the operators (jury, judge) including the aggregation, we have designed the backend modules (operator code) to be as micro services to use RestFul APIs to communicate with each other. As for the user facing UI (Clerk Intake) we aimed to use Autonome custom UI for a chat interface, however up to the last 24 hours we were unable to get it to work as it would stay in "deploying" state, so we developed a simple chat interface (HTML, python, flask) and Gaia API, which uses CDP AgentKit and IPFS to file the case on chain.