After much anticipation, Eliza has officially released its technical whitepaper today.

While many AI agents are commonly known to be built on the Eliza open-source framework, there has always been a lack of detailed and formal documentation on how Eliza defines itself from a technical perspective.

This whitepaper provides a comprehensive answer, describing how Eliza enables a deep integration between AI and Web3, its modular system architecture, and the technical implementation details of its role as an open-source framework.

The whitepaper was co-authored by Shaw, several Eliza Labs members, and technical experts from other related organizations. However, due to its extensive technical details and specialized concepts, it may not be very accessible to general readers.

To address this, TechFlow has simplified and distilled the content, aiming to explain the whitepaper in plain language to help readers quickly grasp its essence.

1. Why was Eliza created?

It’s important to set a clear scope when considering this question. Specifically, why develop Eliza within the context of crypto or Web3, rather than comparing this framework to broader AI frameworks?

Following this line of thought, the introduction and background sections of the technical whitepaper provide an excellent answer:

In the intersection of AI and Web3, there has always been a significant gap—a lack of a framework that can seamlessly integrate with Web3 applications.

Specifically, the whitepaper identifies three major challenges facing the Web3 space:

1. Complexity of Decentralized Transactions With the rapid growth of public chains like Ethereum, Solana, and BASE, managing assets and executing transactions across different chains has become increasingly challenging. While there are existing trading platforms, their basic functionalities often fall short for intermediate and advanced users with customization needs.
2. Value Extraction from On-Chain Data Blockchain holds an immense amount of valuable information, ranging from basic metrics like wallet address changes, token prices, and market capitalization to more advanced metrics like whale account ratios and market maker patterns. Effectively transforming this complex data into actionable insights remains a pressing issue.
3. Fragmentation of Social Media Information For the Web3 industry, platforms like Twitter, Discord, and Farcaster are essential channels for information. However, as the number of key opinion leaders (KOLs) grows, information becomes increasingly fragmented. Extracting valuable insights from this flood of information has become a common challenge for traders.

Eliza was created in response to these real-world needs. As the first open-source, Web3-friendly AI agent operating system, Eliza adopts a modular design, allowing developers and users to customize solutions based on their specific requirements.

Eliza aims to lower the barrier for ordinary users to access advanced AI functionalities,

enabling them to build their own AI agents without requiring extensive programming expertise.

Additionally, the whitepaper compares Eliza with several common AI frameworks. A comparative table clearly highlights that Eliza claims to offer the best support for Web3, which is also the key message the entire whitepaper seeks to convey.

2. Eliza’s Design Philosophy and Technical Innovations

Three Core Design Principles: Simplicity Without Oversimplification

Eliza’s success is no coincidence. From the outset, the team established three core principles:

1. Web3 Developer First Recognizing that Web3 development primarily relies on JavaScript/TypeScript, Eliza chose TypeScript as its development language. This decision allows developers to work with familiar tools and seamlessly integrate blockchain functionalities into existing web applications. Simply put, it enables Web3 developers to “use it out of the box.”
2. Modular Plugin Architecture Eliza breaks its system down into a core runtime and four key components:

• Adapter (Data Integration)
• Character (Agent Personality)
• Client (Message Interaction)
• Plugin (General Functionality)

This architecture allows developers to freely add their own plugins, clients, characters, and adapters without worrying about the details of the core runtime. It also enables Eliza to support a wide range of model providers (e.g., OpenAI, Llama, Qwen), platform integrations (e.g., Twitter, Discord, Telegram), and blockchain compatibility (e.g., Solana, Ethereum, Ton).

1. Prioritizing Simplicity Over Complexity

With limited engineering resources, maintaining a simple internal implementation saves time for developing new features, adapting to new scenarios, and keeping up with the rapid evolution of AI and Web3.

Technical Innovations: Strengthening Internally and Expanding Externally

Eliza’s technical innovations can be divided into two dimensions: internal enhancements and external expansions.

1. Internal Enhancements To improve the reasoning capabilities of AI models, Eliza integrates several cutting-edge techniques:

• Chain-of-Thoughts (CoT):
  • Technical Definition: Introduces step-by-step explanations.
  • Simplified Explanation: Similar to solving a math problem by writing out each step, AI explains its thought process step by step instead of jumping directly to the answer. This not only improves accuracy but also makes it easier for humans to understand how conclusions are reached.
• Tree-of-Thoughts (ToT):
  • Technical Definition: Enables branching to explore multiple solutions.
  • Simplified Explanation: Like considering different moves in a chess game, AI explores various possible solutions simultaneously and chooses the best one. It’s akin to navigating a decision tree to find the optimal path.
• Graph-of-Thoughts (GoT):
  • Technical Definition: Connects reasoning pathways.
  • Simplified Explanation: Treats the problem as a network where ideas are interconnected. Similar to how humans create mind maps for complex problems, AI connects various ideas to form a comprehensive understanding.
• Layer-of-Thoughts (LoT):
  • Technical Definition: Connects reasoning pathways.
  • Simplified Explanation: Treats the problem as a network where ideas are interconnected. Similar to how humans create mind maps for complex problems, AI connects various ideas to form a comprehensive understanding.

1. External Expansions To enhance its practical problem-solving capabilities, Eliza integrates various external functionalities:

• Retrieval-Augmented Generation (RAG):
  • Technical Definition: Enhances generative capabilities through retrieval.
  • Simplified Explanation: Like a student consulting a textbook while doing homework, AI accesses its “knowledge base” to ensure more accurate answers.
• Vector Databases:
  • Technical Definition: Stores and retrieves structured data.
  • Simplified Explanation: Functions like a “library” for AI, allowing it to quickly find similar content. For instance, if you ask for a poem about the moon, it can instantly retrieve related works.
• Web Search:
  • Technical Definition: Retrieves real-time information from the internet.
  • Simplified Explanation: Enables AI to browse the web like a human, accessing the latest information rather than relying solely on pre-existing knowledge.
• Text-to-Image/Video/3D Model:
  • Technical Definition: Converts text descriptions into multimedia content.
  • Simplified Explanation: Like an artist drawing based on a description, AI can generate images, videos, or even 3D models based on your input.

Comparison with Other Frameworks in the Web3 Space

Among the current Web3 AI agent frameworks, Eliza demonstrates clear advantages. Based on feedback from over 50 AI researchers and senior blockchain developers, Eliza outperforms other frameworks in the following key metrics:

• Support for model providers
• Chain compatibility
• Feature completeness
• Social media integration

3. Eliza OS: A Well-Crafted Web3 AI Ecosystem

After understanding Eliza’s design philosophy, let’s dive into how this framework operates. Eliza can be envisioned as a meticulously designed LEGO system, where each part fits perfectly while maintaining exceptional flexibility.

Core Components: Five Key Roles

In Eliza’s ecosystem, five core components work together to form a complete intelligent system.

• Agents: The main characters of the system

They function like independent “digital assistants,” responsible for handling various autonomous interactions. Each agent has its own “memory” and “personality” and can engage in coherent conversations and interactions with users through different channels like Discord and Twitter.

• Character Files: The “persona” of the agents

To make these agents more dynamic, Character Files act as their “personal resumes.” These files define the agents’ identity and personality traits, specify which models they can use (e.g., OpenAI, Anthropic), and determine what actions they can perform (e.g., blockchain transactions, NFT minting). Through well-designed character configurations, each agent can exhibit unique professional expertise and behavioral patterns.

• Providers: The “perception system” of the agents

When interacting with the external world, agents rely on Providers as their “perception system.” Just as humans need senses to perceive their surroundings, providers supply agents with real-time information such as market data, wallet details, and sentiment analysis, enabling them to better understand the current environment and context.

• Actions: The “skillset” of the agents

When specific actions are required, Actions serve as the agents’ “skillset.” From simple buy/sell orders to complex NFT generation, each action undergoes rigorous security validation to ensure absolute safety when handling financial tasks. These skills allow agents to truly operate effectively in the Web3 ecosystem.

• Evaluators: The “decision-making system” of the agents

Lastly, Evaluators act as the agents’ “decision-making system,” responsible for evaluating conversation content, extracting key information, and helping agents build long-term memory. They not only track progress toward goals but also ensure the coherence of the entire conversational process.

Intelligent Interaction: More Than Just Simple Conversations

In terms of interaction, Eliza employs a multi-layered understanding system, much like an experienced translator who not only grasps the literal meaning but also understands the context and intent behind the words. This system can accurately comprehend users’ true needs, maintain a consistent experience across various communication platforms, and flexibly adjust responses based on the context.

Plugin System: Unlimited Expansion Potential

Eliza’s plugin system is essentially a toolbox that brings powerful extensibility to the entire framework. This extensibility is reflected in three directions: multimedia generation, Web3 integration, and infrastructure:

• For multimedia generation, it can produce images, videos, and 3D models, support automatic generation of NFT collections, and provide image description and analysis capabilities.
• For Web3 integration, it supports multi-chain operations on networks like Ethereum and Solana, offers a comprehensive suite of trading functionalities, and integrates various DeFi operations.
• For infrastructure, it provides foundational capabilities such as browser services, document processing, and speech-to-text conversion.

Through this modular design, Eliza not only maintains system stability but also offers developers nearly unlimited expansion possibilities. This enables Eliza to adapt to the ever-emerging demands and scenarios in the Web3 ecosystem.

4. How Capable Is Eliza? Insights from Data

When a new technology framework emerges, its actual performance is often the primary concern. In this regard, Eliza provides a candid answer. \
In the GAIA benchmark test (a platform specifically designed to evaluate the ability of AI agents to solve real-world problems), Eliza demonstrated impressive capabilities. This test does not merely assess basic Q&A skills but requires AI agents to possess various competencies, such as logical reasoning, multimodal processing, web browsing, and tool usage. \
Although Eliza’s overall score (19.42%) still lags behind the current top-performing solutions, this is a commendable result, especially considering its focus on the Web3 domain. Notably, in handling foundational tasks (Level 1), Eliza achieved a completion rate of 32.21%, showcasing its solid foundational capabilities.

Web3 Domain: A Pioneering Standard Setter

What’s even more noteworthy is that Eliza plays the role of a “standard setter” in the Web3 field. As Web3-focused AI systems are still in their early stages, Eliza has taken the lead in proposing a comprehensive evaluation framework, setting the direction for the entire industry.

This evaluation framework is divided into three levels, referred to in the white paper as the Web3 AI version of the “Turing Test”:

• Basic Capabilities: Includes fundamental operations such as wallet creation, token transactions, and smart contract interactions.
• Advanced Features: Incorporates the latest AI technologies, such as text-to-video/3D generation and RAG support.
• Premium Traits: Enables autonomous planning and reasoning based on user instructions, achieving truly intelligent decision-making.

Currently, Eliza has successfully implemented all functions at the basic level and is advancing toward the advanced level. The team expressed confidence that they will achieve fully autonomous AI agent systems in the coming years.

5. Practical Applications: The Market Votes with Real Money

The original white paper includes a section showcasing code to demonstrate practical applications achievable with the framework. However, to simplify comprehension and avoid technical details, only the broader applications will be highlighted here.

According to the white paper, as of January 2025, several significant Web3 projects have built their AI agent systems based on Eliza. The combined market capitalization of these partners exceeds $20 billion.

This figure may itself be the best endorsement of Eliza’s technological strength by the market.

More importantly, the Eliza team is confident about the future. They believe that as these “smart agents” continue to evolve, we will witness a new era where multiple AI units work together in harmony. As Anthropic’s CEO Dario Amodei stated in his vision of a “genius data center,” Eliza is paving the way for this future.

6. Existing Limitations and Future Outlook: An Honest Self-Analysis

No technology framework is perfect, and the Eliza team honestly points out the current limitations of their framework in the white paper.

Three major challenges to be solved

• Lack of Workflow System: Just like a skilled assistant requires a standardized workflow, when developers want to implement routine tasks (such as periodically aggregating data from multiple sources), the current Eliza framework does not offer ready-made solutions. For such needs, workflow systems with graphical interfaces, like Dify or Coze, may still be needed.
• Performance Issues in Multi-Agent Systems: As the number of agents increases, the context and memory content the system needs to handle grows exponentially. In particular, balancing computational overhead and operational efficiency while handling large input-output tasks remains a technical challenge to be resolved.
• Expansion of Multi-Language Support: Currently, Eliza is primarily based on TypeScript, but to attract developers from more domains, it will need to expand support for other programming languages like Python and Rust.

Outlook: Pioneering a New Era of Decentralized AI

Despite these limitations, the significance of Eliza has far exceeded that of a mere technology framework.

It represents a pioneering attempt at the deep integration of AI technology and Web3 applications. By designing each functional module as a standard TypeScript program, Eliza ensures that users have complete control over the system. It also offers seamless integration with blockchain data and smart contracts.

This design guarantees both security and scalability. As stated at the end of the white paper, the possibilities of Eliza are only limited by the imagination of its users.

With the continued evolution of AI and Web3 technologies, Eliza will also keep evolving, continuing to lead the development of decentralized AI.

Disclaimer:

1. This article is reproduced from [deep tide TechFlow], the copyright belongs to the original author [ Deep Chao TechFlow], if you have any objection to the reprint, please contact Gate Learn team, the team will handle it as soon as possible according to relevant procedures.
2. Disclaimer: The views and opinions expressed in this article represent only the author’s personal views and do not constitute any investment advice.
3. Other language versions of the article are translated by the Gate Learn team. Unless otherwise stated, the translated article may not be copied, distributed or plagiarized.