Building on-chain Eliza with Move language: A groundbreaking decentralized AI agent

Rooch Network is pioneering new frontiers by exploring the implementation of on-chain Eliza using the Move language. This innovative approach directly stores character data and memories in smart contracts, unlocking the potential for truly decentralized AI agents.

Defining Characters in Move

At the core of this initiative is the definition of characters in Move.

The data structure follows the JSON standard previously used by Eliza, facilitating the 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 leverage on-chain character and memory storage for dynamic interactions with Eliza. Here’s a quick guide to deploying and using it:

1. Deploy the foc_eliza module to Rooch:

   rooch move publish --named-addresses foc_eliza=default

2. Create on-chain characters using the Eliza character JSON file:

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

Why Go On-Chain?

Transferring Eliza's character and memory storage on-chain offers transformative benefits:

1. Dynamic Evolution

On-chain storage allows the character's profile, 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

Through on-chain governance, the community can define and enforce rules for character evolution and memory management. For example:

• Prevent harmful inputs from contaminating memories.
• Audit the behavior of public AI agents to align with community expectations.

1. Decentralized Trustworthiness

Storing characters and memories on-chain enhances transparency and reliability, making Eliza an ideal choice for an open and trustworthy AI ecosystem.

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

How to Get Involved

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

1. Break 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 chat rooms for multi-user interactions.
• Social media utilities or blogging tools for AI agents like Twitter.

1. Explore AI and DeFi/BTCFi Integration

   Combine on-chain AI capabilities with decentralized finance applications to unlock new opportunities for the Bitcoin ecosystem.

Conclusion

The on-chain Eliza project by Rooch Network marks a significant step in the development of decentralized AI agents. By leveraging the Move language and on-chain storage, our goal is to create transparent, community-driven, and reliable AI agents for the future.

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