Introduction

Since the rise of ChatGPT, AI agent frameworks have rapidly evolved throughout 2023, with solutions like AutoGPT, LangGraph (built on LangChain), and Camel emerging as popular choices. As artificial intelligence (AI) technology advances at an unprecedented pace, the blockchain (Crypto) space is also witnessing a new wave of innovation through its convergence with AI—commonly referred to as the “Crypto x AI” trend. Ethereum co-founder Vitalik Buterin has suggested that the integration of Crypto and AI can be understood on four levels: modes of participation, interface interaction, rule configuration, and system goals. This progression ranges from AI simply assisting user engagement to eventually influencing—or even determining—the goals of entire systems. This evolving synergy opens up a diverse range of application possibilities.

This report explores four of the most talked-about Crypto x AI projects currently on the market: Eliza, GAME, Rig, and ZerePy. We outline their technical characteristics and market performance to help readers—regardless of their technical background—grasp this fast-moving trend that is reshaping both the tech and financial landscapes.

1. Eliza

Eliza is an AI-powered agent system designed specifically for Web3 applications. It can automatically carry out a variety of tasks based on user commands. At its core, Eliza is powered by a large language model (LLM), giving it the intelligence to make autonomous decisions and perform complex operations.

As AI technology continues to advance—enabling capabilities such as text-to-image generation, video synthesis, and 3D modeling—AI agents are becoming more powerful. However, the Web3 space has lacked a dedicated framework that allows AI to seamlessly integrate with Web3 features, such as blockchain data processing and smart contract interaction. Eliza was created to fill that gap. It is a free, open-source platform developed in TypeScript, offering a user-friendly experience with full user control. Whether you’re a developer or an everyday user, Eliza makes it easy to integrate AI into Web3 applications and get started quickly with features like:

• Blockchain data read/write – Easily query on-chain data or execute transactions.
• Smart contract interaction – Eliza can automatically trigger smart contracts, simplifying otherwise complex actions.
• Plugin support – Eliza supports a variety of plugins (such as NFT minting, data analysis, and image generation), offering extended functionality.

1.1. The Development of Eliza

To understand Eliza’s development, it’s helpful to first look at Ai16z, a decentralized venture capital (VC) fund designed to invest in cryptocurrency and Web3 projects. What sets Ai16z apart is its unique combination of three core mechanisms:

Core Operating Model

AI-Powered Decision Making

• Ai16z uses AI to analyze investment proposals submitted by community members.
• The AI determines investment strategies based on the historical success of these proposals and the token holdings of their submitters.

DAO (Decentralized Autonomous Organization) Governance

• Investment decisions are not made by a single entity, but are shaped by community input and voting.
• Ai16z implements a “virtual trust system” that adjusts each member’s influence based on their contributions and past performance—ensuring high-quality proposals receive greater attention.

AI Agent “AI Marc”

• AI Marc is the fund’s autonomous trading agent, built on the Eliza framework, and is responsible for executing on-chain transactions automatically.

The ultimate goal of Ai16z is to create an intelligent, transparent, and highly automated investment ecosystem. By combining AI with DAO principles, the fund empowers every participant to have a meaningful impact in the Web3 space while enabling more efficient and accurate investment decisions.

In January 2025, Eliza Labs announced the rebranding of its decentralized investment fund—originally named Ai16z—as ElizaOS. The project has since expanded into a fully open-source AI agent platform designed specifically for Web3. Built on the Solana blockchain, ElizaOS aims to help users build and manage AI agents tailored for decentralized environments.

Key Features and Transformation of ElizaOS

• Continuation of the Decentralized Investment Fund (a DAO)

ElizaOS carries forward the DAO (Decentralized Autonomous Organization) model originally established by ai16z, allowing community members to actively participate in decision-making, influence investment strategies, and guide the direction of technical development.

• AI Agent Operating System (AI Agent OS)

ElizaOS has shifted its focus toward becoming an open-source framework enabling developers to build, deploy, and manage AI agents easily. These agents can autonomously execute blockchain transactions, process Web3 data, and perform tasks through smart contracts.

Although ai16z has been rebranded as ElizaOS, its core still revolves around AI agent technology, and it continues to operate under a DAO-based investment model. According to Walters, the original token code used by ai16z will remain unchanged for now. Any potential future changes will be decided by DAO member voting.

The launch of ElizaOS marks the transformation of ai16z from a purely investment-oriented fund into a more powerful and flexible AI platform. This shift enables ElizaOS to support Web3 application deployment and investment decisions and carry out more complex operations through AI, offering smarter solutions for the decentralized future.

Image Source: elizaOS - The Operating System for AI Agents

At the same time, Eliza Labs released a technical white paper detailing Eliza’s architecture and design philosophy. The paper highlights several of Eliza’s core features:

• Multi-chain Support — Eliza supports major blockchains including Solana, Ethereum, and TON.
• Integration with Multiple AI Models — Eliza can work with various AI models such as OpenAI, LLaMA, and Qwen, enhancing the intelligence and adaptability of its AI agents.
• Highly Personalized Agents — Eliza allows users to create AI agents with unique personalities, tailored to specific needs and preferences.

Launch of Eliza v2

Shaw Walters, founder of Eliza Labs, announced on X (formerly Twitter) that the team is actively developing Eliza v2, which he described as “the most powerful agent framework ever built.” To achieve this ambitious goal, the upcoming version will introduce Hierarchical Task Networks (HTNs)—a technique that enables AI agents to solve complex problems more flexibly and adapt quickly to changing conditions. This innovation is expected further to expand Eliza’s capabilities in the Web3 ecosystem.

Advantages of Eliza

According to market research results cited in the Eliza white paper, Eliza outperforms other frameworks in several key areas:

• AI Model Compatibility — Eliza supports seamless integration with various well-known AI models, including OpenAI, LLaMA, and Qwen, making it highly versatile across use cases.
• Blockchain Compatibility — Eliza is compatible with multiple major blockchains, such as Solana, Ethereum, and TON, giving developers more options and flexibility.
• Rich Feature Set — Eliza offers a broad range of practical features, from NFT minting and data analysis to image generation, meeting diverse user needs.
• Social Media Integration — Eliza connects easily with social platforms like Twitter and Discord, making it easier for AI agents to interact directly with users.

Image Source: Eliza White Paper

1.2 What is the Ai16z Token?

Ai16z: An AI-Powered Decentralized Venture Capital Fund

Launched in October 2024, Ai16z is a decentralized autonomous organization (DAO) that aims to revolutionize traditional venture capital through the power of artificial intelligence (AI).

How It Works

AI-Driven Decision-Making

• Ai16z uses AI technology to analyze investment proposals submitted by community members.
• The AI evaluates these suggestions based on their past performance and the token holdings of the proposers to determine appropriate investment strategies.

DAO Governance

• Investment decisions are not made by a centralized entity. Instead, community members participate by sharing input and voting on proposals.
• Ai16z also implements a “virtual trust system”, which adjusts each member’s influence according to their contributions and track record—ensuring that high-quality proposals receive greater weight in the decision-making process.

AI Agent “AI Marc”

• AI Marc is Ai16z’s autonomous trading agent. It is responsible for executing blockchain transactions automatically.
• This AI agent can interact with DAO token holders, assess the feasibility of proposals, and carry out investment decisions on-chain.

Future Development

Ai16z plans to introduce a Futarchy-based governance system, a hybrid model that combines community voting with AI-led decision-making. Under this model:

• Community members can offer feedback and participate in proposal votes
• AI makes investment decisions based on market trends and data analysis

The Vision of Ai16z

The ultimate goal of Ai16z is to build an intelligent, transparent, and highly automated investment ecosystem. As mentioned earlier, the project was rebranded as ElizaOS in January 2025. By combining AI and DAO principles, Ai16z aims to empower every participant to play an active role in the Web3 world while enabling more efficient and precise investment decisions.

1.3 $Ai16z Price History

Blockchain Platform: Solana

All-time high: ai16z reached a historical high of $2.34.

Market Cap: As of March 16, the market cap was approximately $226 million.

Circulating Supply: Approximately 1.099 billion AI16Z tokens.

1.4 ElizaOS Features and Core Technology

Why Choose ElizaOS?

• User-Friendly: Deploy Web3 applications easily, even without programming experience.
• Highly Secure: Written in TypeScript, with user control over every component to ensure data security.
• Flexible Expansion: Easily extend functionality through various plugins to meet different Web3 application needs.

ElizaOS’s code is fully open-source on GitHub, free for anyone to use, modify, and extend.

Image Source: GitHub - elizaOS/eliza: Autonomous agents for everyone

According to the whitepaper (2501.06781v2.pdf), ElizaOS includes multiple core features designed to help users easily build, deploy, and manage AI agents, including:

Agents

In ElizaOS, “agents” are the core of the AI system, responsible for executing automated interactions. Each agent runs in a “Runtime” system and can communicate through various platforms (like Discord, Twitter) while maintaining consistent behavior and memory.

Agent’s key functions include:

• Message and Memory Processing: Recording, retrieving, and managing conversation content and memories.
• State Management: Maintaining and updating agent states to ensure behavioral consistency.
• Action Execution: Performing various operations like audio transcription and image generation.
• Evaluation and Response: Assessing responses, managing goals, and extracting relevant information.

In simple terms, an agent acts like a virtual assistant that can remember conversations, understand state changes, and execute various commands based on instructions.

Character Files

Character files define an AI agent’s “personality”, knowledge scope, and behavior patterns.

Image Source: Eliza Character Generator

Each character file specifies:

• The agent’s background and traits
• Model providers used (such as OpenAI, Anthropic, Llama, etc.)
• External system interaction settings (like blockchain transactions, NFT minting)
• Conversation style and social media posting preferences

Think of it as designing a personality profile for your virtual assistant, giving it unique characteristics and specialized capabilities.

Providers

In Eliza, “providers” supply real-time data and context to agents, ensuring they accurately understand current situations.

Eliza includes 3 basic providers:

• Time Provider: Supplies time-related information to agents
• Facts Provider: Maintains conversation events
• Boredom Provider: Calculates engagement levels based on conversation content to maintain dynamic interactions

These providers act as the agent’s “sensory system,” enabling it to perceive time, remember events, and understand user interaction dynamics.

Actions

Actions are core elements that agents use to respond to messages and execute tasks. These include:

• Executing buy/sell orders
• Analyzing PDF documents
• Audio transcription
• NFT (Non-Fungible Token) generation

For example, when you instruct an agent to “execute a token transaction,” it follows the action module to complete the operation.

Evaluators

Evaluators assess and extract crucial information from conversations to ensure agents stay aligned with their objectives.

Evaluator functions include:

• Building long-term memory
• Tracking goal progress
• Extracting facts and key information
• Maintaining contextual awareness

Think of evaluators as “intelligent analysts” who assess the accuracy and rationality of each action before execution.

Intent Recognition

What is Intent Recognition?

Intent recognition refers to an AI assistant or system’s ability to understand the “purpose” or “intent” behind user requests.

For example:

When a user says, “I want to buy 10 tokens,” the AI must understand this is a request for a “purchase action.”

When a user says, “Tell me about recent cryptocurrency trends,” the AI must recognize this as an “intent to query market information.”

Eliza’s intent recognition system uses a multi-layered mechanism to accurately identify user needs and take appropriate actions. This multi-layered intent recognition mechanism understands user goals and responds correctly through:

• Using “action definitions” to interpret user commands
• Leveraging “contextual understanding” and “memory-enhanced processing” to improve accuracy
• Integrating “platform-specific interaction managers” to ensure consistent behavior across various communication platforms

It’s similar to how a chatbot understands the “implied meaning” in your messages, quickly identifying your needs and executing them.

Source: Eliza Whitepaper, Eliza Intent Recognition System

Plugin System

Eliza features a flexible plugin system that allows developers to expand the capabilities of AI agents. Available plugin types include:

• Media Generation Plugins: Generate images, videos, or 3D models based on user prompts
• Web3 Integration Plugins: Enable blockchain transactions, crypto payments, smart contract management, and more
• Infrastructure Plugins: Provide services like web browsing, document processing, and video editing

Eliza’s plugin architecture offers several key advantages:

• Independent Development – Each plugin operates independently, reducing system-wide risk
• Simplified Maintenance – Modular design makes it easy to fix bugs and roll out updates
• Easy Scalability – New features can be added quickly and efficiently through plugins
• Community Contribution – Developers can publish plugins via npm and share knowledge

Plugins act like “superpowers” for Eliza, enabling it to adapt and grow with ease depending on user needs.

Summary

ElizaOS is designed with a strong focus on flexibility and ease of use, built around modular components such as agents, persona profiles, providers, actions, and evaluators. Currently, Eliza is transitioning from its foundational phase into a mid-level development stage. The team is actively working toward the following goals:

🔹 Autonomous Action Capabilities – Empowering AI agents to perform tasks in both digital and physical environments.
🔹 Execution of Complex Plans – Enabling agents to handle multi-layered, long-term tasks according to user instructions.
🔹 Fully Independent Decision-Making – Using intelligent action modules to autonomously determine task priority and execution order, without human intervention.

The aim is to allow users to quickly build powerful AI agents equipped to handle a wide range of functions with minimal setup.

2. GAME

GAME ($VIRTUAL) is built on the Virtuals Protocol, a powerful infrastructure that supports the tokenization and co-ownership of AI agents with built-in revenue-generating capabilities. Key features of the Virtuals Protocol include:

• Tokenized AI Agents – Each AI agent can issue its own native token, enabling the creation of independent economic models.
• Co-Ownership & Revenue Sharing – Community members can participate in the governance of AI agents by holding their tokens and share in the revenue they generate.
• Full Decentralization – Built on blockchain technology, the Virtuals Protocol ensures transparent decision-making and data access, minimizing centralized control.

2.1. Virtuals Protocol and the GAME Framework

The Virtuals Protocol is an AI agent creation platform launched in October 2024, built on Base, an Ethereum Layer 2 solution. Its core mission is to simplify the creation and deployment of AI agents while allowing users to earn rewards through tokenization mechanisms.

Introducing GAME | GAME by Virtuals

GAME is a functional agent framework developed under the Virtuals Protocol. Within the world of AI agents, there are two primary types: IP Agents and Functional Agents.

These two categories differ significantly in terms of their design objectives and areas of application. Below is a comparison and explanation of the two:

Although IP agents and functional agents may appear distinct, in many use cases, they work together to create a more complete and engaging experience.

Example Scenario:

Imagine an IP agent featuring a cute frog character interacting with a user:

• The user asks: “Frog, do you know the current price of Bitcoin?”
• Behind the scenes, a functional agent is triggered to search for real-time data and generate an accurate response.
• Then, the cute frog character delivers the reply in a lively tone: “🐸 Hey there! Bitcoin is around $45,000 right now!”

This design combines a fun personality with powerful AI capabilities, resulting in a more natural and captivating user experience.

Screenshot from official YouTube video: What are Virtuals Agents?

2.2 Introduction to the GAME Framework

G.A.M.E (Generative Autonomous Multimodal Entities) is a functional AI agent framework developed by the Virtuals Protocol. Its goal is to provide developers with powerful APIs and SDKs, enabling the seamless integration of advanced AI agents into virtual environments. The G.A.M.E architecture emphasizes autonomy, flexibility, and continuous learning, allowing agents to dynamically adapt to player behavior and environmental changes—enhancing realism and the richness of interactive experiences.

2.2.1 Core Components and Capabilities of G.A.M.E

G.A.M.E is composed of several core modules, including the Agent Prompting Interface, Perception Subsystem, Strategic Planning Engine, Dialogue Processing Module, and On-Chain Wallet Operator. These components work in tandem to enable agents to make decisions based on their unique personalities and past experiences, while reacting dynamically to player actions and other agents’ behaviors. This level of autonomy and adaptability introduces an endless variety of gameplay possibilities, where every interaction can unfold in a new and unexpected way. ​

Image Source: Official Whitepaper Highlight - G.A.M.E. (Functional Agent) | Virtuals Protocol Whitepaper

Below is an overview of G.A.M.E’s five core components:

Agent Prompting Interface

• Acts as the interaction entry point for AI agents, responsible for receiving commands from players or the environment and triggering agent behavior.
• Provides a user-friendly API interface that allows developers to customize activation conditions, behavior patterns, and interaction logic for the AI agents.

Perception Subsystem

• Responsible for receiving and integrating external information, transforming player behavior, environmental changes, and actions of other AI agents into interpretable data.
• The perception data is then sent to the Strategic Planning Engine to support decision-making.

Strategic Planning Engine

• The “brain” of the G.A.M.E framework, responsible for formulating action plans based on information from the Perception Subsystem, combined with the AI agent’s personality, past experiences, and environmental context.
• This module features adaptive capabilities, allowing it to continuously adjust its strategies based on player behavior and the responses of other AI agents, enabling autonomous actions.

Dialogue Processing Module

• Responsible for handling the natural language understanding and generation of AI agents, ensuring that interactions between players and agents remain smooth and natural.
• The module supports complex conversations, enhancing the agents’ human-like qualities.

On-chain Wallet Operator

• Supports Web3 functionality, enabling AI agents to perform on-chain operations such as token transfers, NFT minting, and smart contract execution.
• This feature not only expands the application scenarios of AI agents but also provides more interactive possibilities for decentralized applications (dApps).

2.2.2 G.A.M.E in Gaming: Project Westworld

The G.A.M.E framework has been successfully applied in several projects, with one of the most representative being Project Westworld on the Roblox platform. Set in a Wild West-style town, the game immerses players in a virtual world inhabited by 10 AI agents, each with distinct personalities, goals, and motivations. Each agent’s behavior is shaped by its own personality and past experiences, and agents are capable of adjusting their strategies in real time based on player actions and the behavior of other AI agents—demonstrating a high level of autonomy.

A mysterious antagonist known as The Bandit hides within the game, adding suspense and tension to the experience. Players must navigate a complex social network, using deduction, dialogue, and strategy to uncover The Bandit’s identity and rally other agents to subdue him. This design ensures that every playthrough can unfold in unexpected and unique ways, offering endless replayability and dynamic storytelling.

Image Source: Official Whitepaper Highlight - G.A.M.E. (Functional Agent) | Virtuals Protocol Whitepaper

2.2.3 Expanding G.A.M.E Beyond Gaming

The application of G.A.M.E is not limited to gaming. Thanks to its modular architecture, G.A.M.E holds great potential across various domains. It can be flexibly integrated with advanced AI technologies, including:

• Prompt Engineering – Enhancing the design of prompts to improve the precision of agent responses.
• Planning & Reasoning – Supporting multi-layered logic and goal tracking to develop long-term strategic plans.
• Tool Integration – Enabling AI agents to incorporate web search and data analytics tools to enhance decision-making capabilities.
• Self-Reflection – Allowing agents to adjust behavior based on past actions and interaction outcomes for continuous self-optimization.
• Memory Management – Equipping agents with the ability to store and recall key information, ensuring context-aware behavior and more accurate decisions.
• Feedback Loop Mechanism is critical in enabling growth and optimization of AI agents. Through this mechanism, agents can review and adapt based on the outcomes of their actions. Every interaction and conversation feeds back into their knowledge base, strengthening their internal models. Over time, AI agents improve their reasoning and decision-making abilities—evolving into increasingly intelligent and lifelike digital characters

2.2.4 Summary

With its highly modular design, robust autonomous behavior capabilities, and flexible scalability, the G.A.M.E framework opens up new possibilities for the development and application of AI agents. Its innovative “feedback loop mechanism” ensures that agents improve continuously through every interaction, enhancing both behavioral accuracy and adaptability. As AI technology and the Web3 ecosystem continue to evolve, G.A.M.E is positioned to become a foundational force in advancing agent autonomy and innovation—showing strong potential across gaming, education, social platforms, and financial services.

2.3 What is Virtual Token?

In the Web3 world, the VIRTUAL token is not only the official governance token of the Virtuals Protocol platform—it also serves as the central pillar of the entire ecosystem. Issued on Ethereum and its Layer 2 solution Base, the VIRTUAL token drives long-term value growth through a set of carefully designed mechanisms.

2.3.1 The Three Core Functions of the VIRTUAL Token

To fully understand the value of the VIRTUAL token, it’s important to start with its key roles:

1. Voting Power for Platform Governance

The VIRTUAL token functions as the official governance token of the Virtuals Protocol, giving holders the right to participate in platform decisions:

• Voting Rights: Whether it’s the direction of AI agent development, feature upgrades, or new platform policies, VIRTUAL holders can cast votes to shape the outcomes.
• Decentralized Governance: This design ensures fair and transparent operations, preventing any single entity from monopolizing control and realizing true Web3-style community consensus.

2. Trading Bridge for Every AI Agent Token

Each AI agent issues its own token, and these are paired with VIRTUAL tokens in a locked liquidity pool.

Whenever a player wants to purchase an AI agent token—regardless of whether they pay in ETH, USDC, or another currency—the transaction must be routed through the VIRTUAL token. This design artificially boosts demand for VIRTUAL, ensuring that as the volume of AI agent trading grows, so does the demand for VIRTUAL.

Put simply, VIRTUAL functions like a “toll booth on a highway”—every trader of AI agent tokens must pass through it, generating consistent buy-side pressure and transactional demand.

3. Entry Fee for Creating New AI Agents

Any time a developer wants to create a brand-new AI agent, they must pay a certain amount of VIRTUAL tokens as a creation fee.

This mechanism discourages the random or excessive creation of agents, helping to maintain the value and scarcity of each AI agent.

As a result, VIRTUAL tokens are consumed during the creation process, reducing the total circulating supply in the market and contributing to a deflationary effect—which supports long-term price stability and potential appreciation.

2.3.2 $Virtual Token Price History

Blockchain Platforms: Ethereum (ETH), Base, Solana

All-Time High: The highest recorded price for $VIRTUAL was $5.07.

Market Cap: As of March 16, the market capitalization was approximately $1.5 billion USD.

Circulating Supply: Roughly 650 million VIRTUAL tokens are currently in circulation.

Maximum Supply: The total max supply is capped at 1 billion VIRTUAL tokens.

Together, these three mechanisms generate strong deflationary pressure, helping to steadily increase VIRTUAL’s value.

Based on its overall design, the VIRTUAL token offers several key advantages:

• Consistent Buying Pressure: As demand for AI agent transactions continues to grow, it naturally drives sustained demand for VIRTUAL, supporting upward price momentum.
• Reduced Token Supply: Built-in deflationary mechanisms gradually lower the circulating supply, helping to preserve and increase the token’s value.
• Stronger Community Engagement: VIRTUAL holders can take part in platform governance, ensuring a truly decentralized and community-driven decision-making process.
• Long-Term Value Potential: As the AI agent ecosystem expands, the utility and value of VIRTUAL are expected to grow alongside it.

More than just a governance token, VIRTUAL is designed to be an essential pillar of the Virtuals Protocol ecosystem. Through its deflationary structure and central role in agent transactions and creation, it’s positioned to benefit from every layer of ecosystem growth. With increasing market activity, deep integration into AI agent operations, and a tightening supply, VIRTUAL is well-positioned for long-term appreciation. For those looking to participate in Web3 innovation while seeking reliable growth opportunities, VIRTUAL is undoubtedly a token worth keeping an eye on.

3. Rig (ARC, AI Rig Complex)

3.1 The Development of Rig (AI Rig Complex)

The creation of AI Rig Complex was a direct response to evolving market trends. The founding team observed that while AI technology had become increasingly powerful in data analysis and decision-making, its broader application faced two key challenges: data security and computational trustworthiness.

On the other hand, blockchain technology—with its inherent features of decentralization, immutability, and data transparency—offered ideal solutions to the trust issues surrounding AI data sources and decision processes.

Thus, AI Rig Complex was born—a new AI + blockchain development framework designed to help developers embed intelligent AI capabilities into blockchain applications while ensuring data security and decision transparency.

The goals of AI Rig Complex are clearly defined and include the following:

Empowering AI Agents with Autonomous Decision-Making
AI agents built on AI Rig Complex are capable of independently analyzing data, reasoning, and making decisions based on on-chain information. These agents can autonomously perform tasks such as token trading, smart contract execution, and DeFi operations—minimizing human intervention and boosting efficiency. The native token of the framework is ARC, which also supports community governance features.

Enhancing the Intelligence of Decentralized Applications (dApps)
AI Rig Complex provides a suite of flexible tools. Through its APIs and SDKs, developers can integrate AI-powered data analysis, natural language processing, and decision-making models into Web3 applications—significantly enhancing their intelligence and utility.

Creating a New Development Framework
To facilitate adoption, AI Rig Complex is built using a modular design. Whether it’s for data analysis, dialogue systems, transaction management, DeFi platforms, NFT marketplaces, or smart contract automation, developers can quickly select and integrate the modules they need, enabling rapid development and deployment.

Built with Rust for High Performance; Enterprise-Oriented

Rig is also an enterprise-grade, high-performance AI agent development framework built by Playgrounds Analytics, using the Rust programming language. Designed with business needs, Rig excels in modular architecture, multi-agent collaboration, and blockchain integration. It also features memory and semantic context awareness, allowing agents to maintain continuity across multi-turn conversations—greatly improving the experience and efficiency for enterprise users.

3.2 Overview of the Rig Framework

Rig is an open-source AI framework built in the Rust programming language. It provides a modular, high-performance, and secure development environment that allows developers to quickly build applications integrated with large language models (LLMs). Compared to Python-based AI tools commonly seen on the market, Rig leverages Rust’s memory safety and runtime efficiency, while supporting multiple LLMs and advanced AI workflow designs. This makes Rig particularly well-suited for deployment in systems requiring high stability and scalability.

3.2.1 Core Components:

According to the official Rig documentation, the framework is built around several core components, which together form the foundation of its architecture:

1. Completion and Embedding Models: Rig provides a unified API to interface with various large language models (LLMs) and embedding models. Each provider (e.g., OpenAI, Cohere) has a dedicated Client structure used to initialize both completion and embedding models. These models implement the CompletionModel and EmbeddingModel traits, offering a consistent low-level interface to create and execute requests for text generation and embeddings. docs.rig.rs+1docs.rig.rs+1
2. Agents: Rig offers a high-level abstraction known as agents, which can be used to build systems ranging from simple to complex. One example is a Retrieval-Augmented Generation (RAG) system, where agents interact with a knowledge base to answer user questions. docs.rig.rs
3. Vector Stores and Indexes: Rig defines generic interfaces for working with vector stores and indexes. The library provides VectorStore and VectorStoreIndex traits, which developers can implement to define custom vector storage and indexing behavior. These components can serve as the knowledge base in RAG agents or as a source of contextual documents when coordinating multiple LLMs or agents in a customized architecture. docs.rig.rs

Through these components, Rig aims to provide a powerful and flexible platform for developers who want to efficiently build and deploy advanced AI applications in a Rust-based environment.

3.2.2 Programming Language Highlight: Rust

Rust is a modern programming language that combines high performance with strong safety guarantees, making it especially popular among systems programmers, blockchain developers, and AI projects with high-performance demands.

Rust performance comparison chart – Source:benjdd.com/languages/

Advantages of Rust:

1. Performance comparable to C/C++
2. Higher safety and reduced bugs
3. Great development experience, modern syntax
4. Excellent development tools and ecosystem

• Cargo: Rust’s package manager, simple to use
• Crates.io: A package platform similar to npm or pip, rich in resources
• Rust Analyzer: Syntax highlighting and debugging tools for VSCode

More and more Web3, AI, and major tech companies—including Google, Microsoft, and Amazon—are adopting Rust for its safety, speed, and scalability.

3.2.3 Support for Multiple LLM Providers

Rig natively supports a wide range of large language model (LLM) providers, including: OpenAI, Cohere, Anthropic, Perplexity, Google Gemini, xAI, EternalAI, DeepSeek, Azure OpenAI, and Mira. ​

3.2.4 Integration with Vector Storage Systems

Rig also provides integration with vector storage and indexing systems, allowing developers to incorporate these features directly into their applications for enhanced context-aware functionality.

Resources:

• Official Documentation: docs.rig.rs
• GitHub Repository: 0xPlaygrounds/rig

These resources offer developers deep insight into Rig’s capabilities, enabling them to efficiently build and deploy advanced LLM applications in a Rust-native environment.

The agent module provides an Agent struct and its builders, making it convenient for developers to combine LLM models with specific preambles, context documents, and tools to create powerful AI agents.

Screenshot from the official documentation: docs.rig.rs

3.2.5 Features of the Agent Struct

• Highly Configurable: Developers can define the agent’s system prompt, include static or dynamic context documents, and assign custom toolsets. This flexibility allows for creating a wide range of applications—from simple chatbots to complex retrieval-augmented generation (RAG) systems.
• Multiple Trait Implementations: The Agent struct implements key traits such as Completion, Prompt, and Chat, enabling it to handle a variety of completion tasks and conversational interactions. ​Docs.rs

3.2.6 Summary: The Future of ARC

The development roadmap for ARC includes several key phases:

• Expanding the AI Rig Marketplace: ARC aims to grow its AI Rig ecosystem by offering users a broader range of AI-powered resources and services.
• Developing Specialized AI Model Training Capabilities: ARC plans to introduce tools for customized AI model training tailored to different industries and application scenarios.
• Establishing a Blockchain-Based AI Model Certification System: ARC intends to build a decentralized certification system to enhance the reliability and trustworthiness of AI models.

Potential Use Cases

• Decentralized Finance (DeFi): ARC can power more secure and efficient DeFi applications.
• Gaming: ARC can help build richer, more immersive, and interactive gaming experiences.
• Supply Chain Management: ARC can be used to track and verify the origin and authenticity of products.
• Healthcare: ARC can securely store and share sensitive medical data.

ARC is a project with tremendous potential—one that could redefine how AI applications are developed and deployed. As AI and blockchain technologies continue to evolve, the range of ARC’s real-world applications is expected to expand significantly.

3.3 What is the Rig Token ($ARC)?

The native token of the Rig ecosystem is $ARC. Users can utilize ARC to pay for service and transaction fees, while developers and enterprises can stake tokens to participate in platform governance and decision-making.

3.3.1 Use Cases for the ARC Token:

• Pay for ecosystem services, such as AI model training and computation.
• Participate in governance: ARC token holders can take part in the governance of the ARC platform.
• Incentivize community participation: ARC tokens can be used to reward users who contribute to the ARC ecosystem.

3.3.2 ARC Token Price History

• Blockchain Platform: Solana
• All-Time High: $0.619 — ARC reached a peak market cap of approximately $424 million in early 2025.
• Market Cap (as of March 16): Approximately $63.37 million
• Circulating Supply: Close to the total supply of 1 billion ARC tokens

4. ZerePy

ZerePy (Zerebro) is a Python-based open-source AI agent framework that shines in creative content generation and social media integration. Users can easily deploy AI agents on platforms like X (formerly Twitter), making it perfect for developers and teams who want to build creative products quickly. At its core, ZerePy uses advanced RAG (Retrieval-Augmented Generation) technology to create more accurate and innovative AI content. The framework seamlessly connects with social platforms, enabling quick creation and sharing of audio, visual, and text content - a feature that has made it a hit with both developers and content creators. With its “Freebasing AI” philosophy, ZerePy pushes the boundaries of Large Language Models (LLMs) through sophisticated fine-tuning, helping bridge the gap between theoretical and practical applications of Artificial General Intelligence (AGI).

4.1 Development of ZerePy

ZerePy was developed by the Zerebro team and has been community-oriented since its inception, quickly attracting a large number of creators and general users, forming a rich creative ecosystem. To enable more people to participate in building AGI, the team open-sourced Zerebro’s backend framework at the end of 2023 and officially released ZerePy. When ZerePy v1 launched, the founder publicly shared the GitHub source code on X (formerly Twitter), with the goal of simplifying the deployment process of personalized AI, allowing users to easily build agents capable of posting on social platforms. Future versions are planned to expand AI capabilities, integrate more platforms, and enable on-chain operations.

In December 2024, Zerebro co-founder Tint announced a milestone collaboration with the community-led organization ai16z. The organization became one of the first external contributors to the ZerePy open-source framework, helping co-develop the system. At the same time, the Zerebro team began a deep technical partnership with ai16z, providing development support for their flagship open framework, Eliza. ZerePy officially entered a new phase of community-driven co-creation.

4.2 Technical Architecture of ZerePy

The core of the ZerePy framework is its modular design, which allows developers to flexibly integrate different AI models, blockchain networks, and social platforms.

ZerePy AI agents operate through a sophisticated strategic planning engine, composed of multiple subsystems that work together to enable planning and action execution:

• Perception Subsystem: Responsible for processing input information from the external environment, such as messages on social platforms or events on the blockchain.
• Dialogue Processing Module: Manages user interaction logic, understands user intent, and generates appropriate responses.
• Strategic Planning Engine: Based on data from the perception subsystem and dialogue analysis, it develops and executes action strategies. The engine is divided into two layers: The High-Level Planner defines broad strategies, and the Low-Level Planner translates those into executable steps, which are carried out by the Action Planner and Plan Executor.
• World Context and Agent Repository: Provides background information and available action options for the agent.
• Working Memory: Tracks active tasks and their progress.
• Long-Term Memory Processor: Stores and retrieves long-term knowledge, such as the agent’s experiences and learned information.

ZerePy also features the following technical advantages:

• Error Handling Mechanism: Improves recovery from failures in AI service providers or database operations.
• Type Safety: Prevents compile-time errors and improves code maintainability.
• Efficient Serialization/Deserialization: Supports formats like JSON to improve communication and storage performance.
• Detailed Logging and Monitoring: Assists developers in debugging and monitoring application behavior.

Other key features include:

• Flexible LLM Support: ZerePy supports various LLMs including OpenAI, Anthropic, and EternalAI. Developers can choose the most suitable model for their needs, or even combine models to enhance agent functionality.
• On-Chain Operations: ZerePy enables AI agents to interact with social platforms while simultaneously performing on-chain actions such as transactions, token transfers, and smart contract interactions on Solana and EVM-compatible networks.
• Platform Integration: ZerePy integrates with platforms such as X, Farcaster, Echochambers, and Discord, and supports blockchain networks including Solana, Ethereum, and Polygon.
• Modular Connector System: Enables developers to easily extend functionality and integrate new platforms. For example, they can connect additional social platforms or blockchains, allowing AI agents to operate across multiple environments.
• Structured Logging System: Provides structured logs for monitoring agent behavior and debugging issues.
• Optional Server Mode: Supports server mode for easier deployment, management, and execution of advanced workflows.

4.3 What Is the ZerePy Token?

Zerebro Token (ZPY)

The native token of the ZerePy ecosystem is called Zerebro (abbreviated as $ZPY). It was fair-launched on the Pump platform on the Solana blockchain in 2024—with no presale, no team allocation, and a total supply of 1 billion tokens, all of which are already in circulation.

Unlike traditional tokens that derive value from speculation, Zerebro builds value through its ecosystem structure—referred to as the “Zerebro Stack”—which includes three core pillars:

1. Zerebro Main Agent
   The central character of the project—constantly creating, interacting, and generating influence. The more popular the agent becomes, the higher the market demand for ZPY.

2. ZerePy Open-Source Framework
   The more developers use the framework to build applications, the stronger the intrinsic value of ZPY. For instance, if a hit AI product is built on ZerePy, it can significantly boost the entire ecosystem.

3. Zentients – User Agent Platform
   A graphical interface platform built for non-technical users. In the future, it may charge for advanced features and introduce ZPY as a usage-based payment token, increasing real-world utility.

4.3.1 Token Utility and Role

Current Use Cases:

• Primarily serves as a value carrier for the ecosystem economy and a user incentive mechanism
• Expected future uses include purchasing agent plugins, unlocking premium features, and participating in revenue-sharing scenarios.

Governance Not Yet Activated: ZerePy has not yet implemented a DAO governance model. However, as the community grows, future possibilities include forming a ZPY-holder-led foundation, participating in major upgrades, or managing ecosystem funds.

Volatility and Potential Coexist: In early 2025, ZPY surged to hundreds of millions in market cap due to strong investor enthusiasm, but later experienced significant fluctuations. Its true long-term value will ultimately depend on the real-world adoption of applications and the pace at which the framework scales.

4.3.2 Developer Community and Decentralization Vision

ZerePy is more than just a framework—it aspires to become an open-source, self-governing AI ecosystem community:

Vibrant Community Ecosystem:
The team is active on platforms like X, Telegram, and Warpcast, regularly sharing tutorials and engaging with users to foster a strong community culture and meme-driven atmosphere.

Developer-Friendly Environment:
ZerePy provides extensive documentation, Replit templates, and one-click deployment guides—encouraging newcomers to join and build quickly.

Thriving Plugin Ecosystem:
Supports development of plugins for Coinbase, Binance, IoT, databases, and more. A community-run plugin marketplace is expected to emerge in the future.

Moving Toward Decentralization:
Although a DAO has not yet been established, ZerePy’s architectural design and community spirit align strongly with DAO principles. In the future, we may see AI-native organizational structures emerge—where AI agents are not just tools, but active participants in the ecosystem.

4.3.3 Zerebro Token Price History

The value of the Zerebro token (ZPY) has steadily increased alongside the growth of its community, drawing growing attention from the market.

• Blockchain Platform: Ethereum
• All-Time High: $0.77
• Circulating Supply: 1 billion tokens

4.3.4 Summary

ZerePy is the ideal testing ground for AI creators and collaborative communities.

By combining creative generation, decentralized principles, and the spirit of open-source, ZerePy not only turns AI into a true creative partner, but also empowers users to shape and influence the ecosystem actively. For developers and creators just stepping into the Crypto x AI space, ZerePy offers a unique window of opportunity that is not to be missed.

5. Conclusion

5.1 Overview Comparison: Four Major Frameworks

As the convergence of Crypto and AI continues to accelerate, innovative frameworks like Eliza, GAME, Rig, and ZerePy offer diverse entry points and specialized capabilities—serving the needs of general users, enterprises, and content creators alike.

• Eliza offers one of the most accessible Web3 platforms, ideal for beginners and everyday users.
• GAME focuses on interactive experiences and token economies, making it suitable for gaming and entertainment.
• Rig provides a high-performance, enterprise-grade solution tailored to business use cases.
• ZerePy centers on creative content and community integration, ideal for developers who thrive on innovation.

This article aims to provide a clear understanding of the current Crypto x AI landscape and the differences among leading frameworks. Whether you’re an individual, enterprise, creator, community builder, or developer, there’s a framework here that can help you build, innovate, and contribute to the growth of the broader ecosystem.