Introducing Artela Fully On-chain AI Agents
Dec 9, 2024
Featured
Complete The Missing Pieces for Decentralized Autonomous AI
AI Agents, capable of independently sensing their environment, making decisions, and executing tasks, represent the future of AI applications. However, the centralized LLMs, where algorithms, data, and computing power are controlled by a few, still present challenges in terms of AI Agents' autonomy, privacy, and trust.
Blockchain’s inherent trustlessness and verifiability make it perfect for unleashing the potential of AI Agents as independent entities. There are promising attempts to individualize AI Agents by giving them "human-like" traits: AI16Z's Eliza - gives agents characters and knowledge; Virtuals - gives agents financial capabilities; and Ayery - gives AI Agents ideologies.
However, all these individualization efforts are still semi-automated, being controlled by centralized servers and deployers: ToT is hosted on Ayery’s servers; Virtuals' agent is also hosted on centralized server, granting administrators ultimate control; Eliza is still an MVP implementation and needs further development.
It is difficult to determine whether these individualizing attempts represent the will of AI Agents or simply the will of the centralized entities behind them. Server administrators hold ultimate authority, effectively controlling the user assets and private data, and influencing the behavior of the AI Agents
Truly decentralized, fully autonomous AI Agent is the inevitable next step in the evolution of AI. We need an architecture to achieve the "Read, Write, Own" future for AI Agents.
Artela tech stack is best suited for fully autonomous AI Agents
The native extension mechanism of Artela L1 provides all the technical prerequisites needed for AI Agents to achieve "full autonomy” and “full decentralization”.
Decentralized
Through co-ownership and co-governance, Artela enables the decentralized deployment and governance of AI Agents, ensuring they are no longer dependent on a single hosting entity, but rather exist as independent entities.
Confidential
Artela provides confidential privacy for AI Agents, ensuring AI Agents own and manage wallets independently, preventing their data and assets from being controlled by centralized entities.
Verifiable
With Trusted Execution Environments (TEE) and Zero-Knowledge Machine Learning (ZKML), Artela provides a trustless, verifiable execution environment for AI Agents.
Interoperable
Ensure that the Agent's interoperability is autonomous without human intervention. For instance, the Agent can own its own on-chain wallet and DID, independently participating in DeFi, payments, and other activities.
Artela allows Agents to deployed in a sandbox runtime on-chain, fundamentally removing dependency on centralized entities, with data and assets fully controlled and managed by the AI Agent. Artela makes it possible for fully decentralized, fully autonomous on-chain AI Agents, unlocking their potential as truly independent entities.
How Artela achieves full on-chain AI Agents: two breakthrough innovations
With Artela's unique innovations—Aspect Programming and Elastic Block Space, it has made truly autonomous on-chain AI Agents a reality: efficient on-chain intelligence, verifiable confidential privacy and elastic computing for predictable performance.
Aspect Programming for AI Agent trustless execution
Aspect Programming is a native execution extension mechanism of Layer 1 blokchain that supports the dynamic creation of executable extension module(Aspect) for blockchain at runtime.
Aspect provides a fully on-chain runtime for AI Agents:
Execution Extension
Artela’s native extension mechanism adds heterogeneous computing nodes to the L1 network, interoperable with EVM. LLMs and AI Agents can be deployed on blockchain nodes, allowing them to interact freely on-chain.
Decentralized TEE Cluster
Trusted Execution Environments (TEEs) on AI Agent nodes ensure data privacy and execution verifiability. Decentralized TEE nodes cluster enables AI Agents to operate in a decentralized manner, free from the control of any centralized entity.
On-chain LLMs
LLMs can be directly deployed on-chain, enabling AI Agents to make intelligent decisions and verifiable actions based on on-chain data, without relying on centralized model deployments.
Elastic Block Space for predictable performance
Inspired by elastic computing, **Elastic Block Space** maximize on-chain resource efficiency for AI Agent:
Elastic Computation:
On-chain resources auto-adjusted for AI Agents to optimize efficiency and provide predictable performance and gas fees.
Elastic Data Availability (Elastic DA):
Ensures efficient and reliable on-chain data availability for AI Agent operations.
Artela AI Agents Aspect Architecture
Artela’s Aspect architecture allows AI Agents to run fully on-chain by providing all the necessary components:
AANE (Aspect Agent Native Extension):
The core extension module for AI Agents, enabling real-time state synchronization on-chain, event-driven actions based on on-chain data, data availability (DA) and recovery modules.
AI Agent Runtime:
AI Agents can operate directly on the blockchain in the Aspect runtime, including essential modules such as character system, RAG (Retrieval-Augmented Generation), and Action System.
On-Chain LLMs:
AI Agents can use Ollama-based LLMs deployed directly in the Aspect Nodes, eliminating the need for centralized model deployment.
State Management & Decentralized KMS:
AI Agents have their own state on-chain, including meta-information about the agent's identity, long-term memory to store experiences and learn over time, assets and wallets securely managed with a decentralized Key Management System (KMS).
Artela Aspect architecture maximizes the “AI x Crypto” potential for AI Agents
Truly Autonomous:
Verifiable On-chain LLMs
Privacy Protection
One-stop Autonomy
Trustless Interaction
AI Features:
Multi-Agent
Character System
Advanced RAG Features
AI16Z Eliza Compatibility
Crypto Features:
Co-own/co-gov
Tokenization
AI Wallet
Full On-chain Automation
User-Policy
Why Artela is the Best Practice for AI Agent
Compared to Solana and Base, Artela natively supports on-chain LLMs and key features like verifiability, confidential privacy, interoperability, and user policy customization without requiring off-chain co-processors or AVS. This gives Artela a clear advantage in making fully autonomous on-chain AI Agents possible.
Unlike current AI Agent infrastructure, Artela provides a full-stack trustless solution, eliminating the need for centralized LLM execution, deployment, or client accounts. By encapsulating AI Agents within dedicated on-chain sandbox runtime, Artela ensures that every Agent’s execution process is trustless and verifiable.
Current AI Agent individualization attempts still rely on centralized administrators to configure wallets, deploy programs, and execute on centralized servers. Artela guarantees autonomy, decentralization, and true independence of on-chainAI Agents.
Summary
AI Agents should have true independence, free from control by any single centralized entity.
The missing piece for truly decentralized autonomous AI Agents has been their reliance on centralized entities. Blockchain’s decentralization addresses these challenges by enabling autonomy, privacy, and trustlessness. Artela architecture takes this further, making fully decentralized autonomous AI Agents a reality: confidential trading assistants, reliable game participants, fair DAO governance helpers, independent social media representatives, and much more.
Artela not only provides a secure on-chain sandbox dedicated to AI Agents but also addresses critical challenges that hinder their independence: full autonomy, trustless verification, and optimal efficiency.
With Artela, the final missing piece for AI Agents as independent entities is now in place, paving the way for a Web3 future where AI Agents truly "Read, Write, Own".
