Olas x Isotropic x Shutter: Using Autonomous Agents to Give Blueprint for Fairness on EVM Chains
In July 2024, the Shutter integrated its groundbreaking threshold encryption technology known as Shielded Trading into the Gnosis Chain mempool, bringing front running prevention and censorship resistance to the chain. Today, we’re excited to announce an integration between Olas, Isotropic Solutions, and Shutter to conduct AI-driven experiments using autonomous agents on Shutter’s encrypted mempool on Gnosis Chain. Isotropic Solutions was selected through a community-driven decision, with members of the Shutter DAO 0x36 voting on Snapshot to approve the proposal to build Olas agents.
Why This Integration Matters
1. First-of-its-kind use of Olas autonomous agents to simulate MEV and transaction ordering
This integration marks the first use of autonomous agents to create an open platform for experimentation in Maximal Extractable Value (MEV) prevention and transaction ordering strategies. Moving beyond theoretical research, we’re testing these strategies in near-real-world conditions. By leveraging Olas' cutting-edge AI agents, we’re tackling one of DeFi's biggest challenges—MEV and transaction manipulation—in a tangible way, fast-tracking insights that could otherwise take years to uncover.
2. Building a scalable model for fairness on other chains
This integration will further test how Shutter can effectively prevent front running and reduce censorship, even in more competitive environments like Ethereum. By simulating real-world scenarios, we’re demonstrating how Shutter’s encrypted mempool can serve as a model for fairness on other EVM chains.
The Experiment: A Brief Overview
We’re setting up a live experiment on Gnosis Chain, where some AI agents will operate within Shutter’s encrypted mempool while others won’t. This side-by-side comparison will demonstrate how effectively Shutter protects transactions under real-world conditions.
The goal is to test various transaction ordering strategies and evaluate how much protection the encrypted mempool provides in different scenarios. While Gnosis Chain currently experiences relatively low front running activity, the use of these AI agents will allow us to understand how Shutter's technology performs in more complex, high-activity environments. By gathering real data, we’ll gain deeper insights into the economic benefits of Shutter’s encryption and demonstrate that it’s scalable to other chains, including Ethereum mainnet.
The Experiment: Extended Details
In this experiment, AI-driven autonomous agents from Olas will be deployed to explore transaction ordering strategies and MEV extraction under both encrypted and non-encrypted conditions on the Gnosis Chain. These agents will operate in a carefully designed simulation environment that mirrors real-world blockchain activity, providing a practical setting for understanding how Shutter’s encrypted mempool enhances fairness and reduces manipulation.
The agents are based on advanced algorithms such as Deep Reinforcement Learning (DDPG and PPO), which enable them to adapt and optimize their strategies in real-time. Each agent follows one of two strategies:
- MEV Strategy Agents: These agents are programmed to exploit traditional MEV opportunities such as front running, sandwich attacks, and arbitrage. They operate both inside and outside the encrypted mempool to assess the effectiveness of Shutter's protection mechanisms.
- Normal Trading Strategy Agents: These agents simulate regular trading activities without employing MEV strategies. By comparing the results of these agents with their MEV counterparts, we can measure how much advantage is gained by front runners and how Shutter’s encryption technology mitigates this.
In the Simulation Environment, each agent will interact with a mempool that either uses Shutter’s threshold encryption (for Shielded transactions) or follows the traditional open mempool structure (for unshielded transactions). The simulation environment includes the following key components:
- Simulation Clock: This feature manages the passage of time within the simulation, allowing transactions to be processed block-by-block, as they would be in a real blockchain.
- Data Collection and Analysis Module: This module captures performance metrics, such as the value extracted by MEV agents and the fairness of transaction ordering, providing insights through real-time data analysis powered by Apache Spark.
- Integration Layer: Interfaces with Gnosis Chain and Shutter’s encrypted mempool, ensuring that transactions are submitted and processed correctly while maintaining data security and integrity.
- Dashboard: The simulation's outcomes will be visualized through a customizable dashboard, where stakeholders can monitor agent performance, transaction orderings, and the overall effectiveness of different strategies.
By using a flexible architecture that includes stream processing with Apache Flink for real-time data and game-theory models to simulate agent interactions, this experiment is designed to provide a comprehensive view of MEV dynamics. With Shutter’s encrypted mempool, the AI-driven agents will demonstrate how threshold encryption can reduce MEV extraction, increase fairness, and create more equitable transaction processing.
Key Milestones
This integration follows a structured list of objectives:
- End of October: Isotropic Autonomous Agents Framework setup and MEV ecosystem simulation project plan finalized.
- Mid-November: Initial Minimal Viable Product with basic trading and MEV strategy simulation.
- December: Suite of autonomous agents developed and integrated with the encrypted mempool, along with a basic dashboard for visualizing and monitoring MEV activity.
- Potential Project Extension: Evaluation and refinement of MEV strategies, testing scalability, and initial exploration of cross-chain MEV opportunities.
Each milestone brings us closer to refining this model, proving its viability, and preparing it for broader application across various blockchain ecosystems.
A Glimpse Into the Future
Our goal is to create a versatile system that ensures fairness across multiple EVM chains. This integration is just the beginning, with an initial focus on an open platform for MEV prevention and transaction ordering experimentation on Gnosis Chain. Once this first experiment proves successful, the potential of Olas agents extends far beyond. In the future, they could be used to:
- Simulate behavior and test assumptions in Shielded versus non-shielded governance.
- Optimize Keyper fees and fee distribution.
- Model and analyze collusion behavior among Keypers.
These advancements will not only refine Shutter’s capabilities but also lay the groundwork for fairer and more secure EVM chains.
Links and Resources
Follow us on X-Twitter, visit the Shutter Forum, and join our community on Shutter Discord to learn more about how Shutter works and stay updated on our progress.
