Fully On-Chain Eliza with Move Language: Pioneering Decentralized AI Agents

Rooch Network is breaking new ground by exploring implementing a fully on-chain version of Eliza using the Move language. This innovative approach stores character data and memory directly within smart contracts, unlocking the potential for truly decentralized AI agents.

Defining Characters in Move

At the core of this initiative lies the character definition in Move.

The data structure adheres to the JSON standard previously used by Eliza, facilitating seamless migration of existing character definitions onto the blockchain. For the complete code, visit the Rooch GitHub repository (opens in a new tab).

Getting Started with On-Chain Eliza

Developers can interact with Eliza dynamically by leveraging on-chain character and memory storage. Here's a quick guide to deploying and using it:

  1. Deploy the foc_eliza modules to Rooch:

    rooch move publish --named-addresses foc_eliza=default
  2. Create an on-chain character using an Eliza character JSON file:

    rooch move run --function default::character::create_character_from_json --args file:../../../eliza/characters/dobby.character.json

Why On-Chain?

Transitioning Eliza’s character and memory storage on-chain offers transformative benefits:

1.Dynamic Evolution

On-chain storage enables characters’ bios, interests, and message templates to evolve dynamically through user interactions. This fosters a transparent, community-driven process, where participants can oversee and shape the development of AI agents.

2. Governance and Auditing

With on-chain governance, communities can define and enforce rules for character evolution and memory management. For example:

  • Prevent memory contamination from harmful inputs.
  • Audit public AI agents’ behavior to align with community expectations.

3.Decentralized Trustworthiness

Storing characters and memory on-chain enhances transparency and reliability, making Eliza an ideal fit for open and trustworthy AI ecosystems.

Future Developments

Our vision for on-chain AI agents includes the following milestones:

  1. Dynamic Character Loader

    Enable Eliza to dynamically load and update character definitions directly from the blockchain.

  2. On-Chain Memory Sync Plugin

    Develop a plugin to synchronize Eliza’s memory with on-chain states in real-time.

  3. AI Oracle Integration

    Integrate AI Oracles for advanced decision-making, enhancing intelligent on-chain interactions.

  4. Standardized On-Chain Components

    Abstract core on-chain components for AI agents and integrate them into the Rooch Move framework to standardize AI agent development.

How to Participate

We invite developers and innovators to push the boundaries of on-chain AI by contributing to the following initiatives:

  1. Push the Boundaries of Smart Contracts

    Collaborate to explore the possibilities of implementing on-chain AI Agents, such as:Supporting automatic state vectorization and vector retrieval directly in smart contracts (on-chain RAG)

  2. Provide More On-Chain Services

    Build services tailored for AI agents, such as:

    • On-chain chatrooms for multi-user interactions.
    • Social media utilities like Twitter or blogging tools for AI agents.
  3. Explore AI and DeFi/BTCFi Integration

    Merge on-chain AI capabilities with decentralized finance applications, unlocking novel opportunities for the Bitcoin ecosystem.

Conclusion

Rooch Network’s fully on-chain Eliza project marks a pivotal step in decentralizing AI agent development. By leveraging Move language and on-chain storage, we aim to create transparent, community-driven, and reliable AI agents for the future.

Join us in building the next frontier of AI agents on-chain!