Deconstructing AI Frameworks: From Intelligent Agents to Decentralization Exploration

Introduction

The development speed of the AI Agent track is astounding. In just two months since the "Truth Terminal" opened the Agent Summer, the narrative combining AI and cryptocurrency has seen new changes almost every week. Recently, the market's attention has focused on "framework-type" projects dominated by technical narratives, which have already produced multiple projects with market capitalizations exceeding hundreds of millions, even billions, in the past few weeks. These projects have also spawned a new asset issuance model, where projects issue tokens based on their Github repositories, and Agents developed based on the framework can issue tokens again. With the framework as the foundation and Agents on top, a model similar to an asset issuance platform is forming, which is actually an infrastructure model adapting to the AI era is emerging. This article will start with an introduction to the framework and interpret the significance of AI frameworks in the cryptocurrency field, combined with personal thoughts.

1. What is a framework?

AI frameworks are underlying development tools or platforms that integrate pre-built modules, libraries, and tools, simplifying the process of building complex AI models. These frameworks typically include functionalities for processing data, training models, and making predictions. Simply put, a framework can be understood as the operating system of the AI era, similar to Windows and Linux in desktop operating systems, or iOS and Android in mobile platforms. Each framework has its own advantages and disadvantages, allowing developers to choose based on specific needs.

Although the "AI framework" is a new concept in the cryptocurrency field, the development of AI frameworks has been close to 14 years since Theano was born in 2010. In the traditional AI field, both academia and industry have mature frameworks to choose from, such as Google's TensorFlow, Meta's Pytorch, Baidu's PaddlePaddle, and Byte's MagicAnimate, each with its own advantages for different scenarios.

The emerging framework projects in cryptocurrency are built on the massive demand for Agents sparked by the AI boom, and then expanded into other tracks, ultimately forming AI frameworks in different niche areas. Here are introductions to several mainstream frameworks:

1.1 Eliza

Eliza is a multi-Agent simulation framework specifically designed for creating, deploying, and managing autonomous AI Agents. Developed in TypeScript, it has good compatibility and is easy to integrate with APIs.

Eliza is mainly aimed at social media scenarios, supporting multi-platform integration such as Discord, X/Twitter, Telegram, etc. In terms of media content processing, it supports PDF document reading and analysis, link content extraction and summarization, audio transcription, video content processing, image analysis and description, and dialogue summarization.

Eliza currently supports use cases mainly including AI assistant applications, social media personas, knowledge workers, and interactive roles. Supported models include local inference with open-source models, cloud inference using OpenAI's API, default configuration as Nous Hermes Llama 3.1B, and integration with Claude for complex queries.

1.2 G.A.M.E

G.A.M.E(Generative Autonomous Multimodal Entities Framework) is an automatically generated and managed multimodal AI framework, primarily designed for intelligent NPCs in games. The framework is characterized by its accessibility to users with low-code or even no-code backgrounds, allowing them to participate in Agent design simply by modifying parameters.

The core design of G.A.M.E is a modular design that works through the collaboration of multiple subsystems, including the Agent Prompt Interface, Perception Subsystem, Strategic Planning Engine, World Context, Dialogue Processing Module, On-chain Wallet Operator, Learning Module, Working Memory, Long-term Memory Processor, Agent Repository, Action Planner, and Plan Executor.

The framework mainly focuses on the decision-making, feedback, perception, and personality of agents in virtual environments, suitable for gaming and metaverse scenarios.

1.3 Rig

Rig is an open-source tool written in Rust, designed to simplify the development of applications using large language models (LLMs). It provides a unified operating interface that allows developers to easily interact with multiple LLM service providers and vector databases.

The core features of Rig include a unified interface, modular architecture, type safety, and efficient performance. Its workflow involves user requests passing through the provider abstraction layer, then in the core layer, smart agents call various tools or query vector storage to obtain information, and finally generate responses through mechanisms such as retrieval-augmented generation.

Rig is suitable for building question-answering systems, document search tools, context-aware chatbots or virtual assistants, and supporting content creation.

1.4 ZerePy

ZerePy is an open-source framework based on Python, designed to simplify the process of deploying and managing AI Agents on the X platform. It inherits the core functionalities of the Zerebro project but adopts a more modular and extensible design.

ZerePy provides a command-line interface that supports large language models from OpenAI and Anthropic, directly integrates with the X platform API, features a modular connection system, and plans to integrate a memory system in the future.

Compared to Eliza, ZerePy focuses more on simplifying the process of deploying AI Agents on specific social platforms, leaning more towards practical applications.

2. The replica of the BTC ecosystem

The development path of AI Agents has many similarities with the recent BTC ecosystem. The development of the BTC ecosystem can be summarized as: BRC20 - multi-protocol competition - BTC L2 - BTCFi. AI Agents are developing more rapidly based on a mature traditional AI technology stack, and their path can be summarized as: GOAT/ACT - Social type Agents / Analytical AI Agent framework competition. In the future, infrastructure projects focusing on Agent Decentralization and security may become the main theme of the next stage.

This track is unlikely to become homogenized or bubble-like like the BTC ecosystem. AI framework projects provide a new idea for infrastructure development. Compared to Memecoin Launchpad and inscription protocols, the AI framework resembles a future public chain more, while Agents are more like future Dapps.

In the future AI era of Crypto, the debate may shift from the discussion of EVM and heterogeneous chains to a framework contest. The current key issue is how to achieve Decentralization or chainization, and what the significance of doing this on the blockchain is.

3. What is the significance of going on-chain?

The combination of blockchain with anything must face the question of its meaning. Consider the success factors of DeFi: higher accessibility, better efficiency, and lower costs, as well as the security of decentralization without the need for trust. Based on these ideas, the reasons supporting Agent chainization may include:

1. Achieve lower usage costs, improve accessibility and choice, allowing ordinary users to participate in AI "rental rights";

2. Provide blockchain-based security solutions to meet the needs of Agents interacting with the virtual or real world;

3. Develop unique blockchain financial gameplay, such as Agent-related computing power, data tagging investment, etc;

4. Achieve more attractive interoperability than the agent browsers provided by traditional internet giants through transparent and traceable reasoning.

4. Creative Economy

Framework projects may provide entrepreneurial opportunities similar to the GPT Store in the future. Simplifying the Agent building process and offering a framework for complex function combinations may have advantages in the future, creating a more interesting Web3 creative economy than the GPT Store.

Compared to the GPT Store, the Agent creative economy of Web3 may be fairer and more open, allowing ordinary people to participate. Future AI memes may be smarter and more interesting than existing Agents, providing new opportunities for the creative economy.