Emerging Trends in the Web3 Space: Analysis of AI Agent Projects

In the first half of 2024, the total market value of the crypto market grew by 37.3%. Among several popular narratives, Memecoin, AI, and RWA performed the strongest. The price performance of AI-related project tokens was only second to Memecoin.

The Web3 x AI track has gained favor from venture capital, with older projects like Bittensor also receiving attention. This track is expected to continuously innovate alongside traditional AI development. AI Agents are the latest trend introduced into Web3, with many new projects like Spectral and Olas Network emerging, while older projects such as Fetch.AI and Phala are also following closely. At the ETHCC conference, Web3 developers and top VCs showed strong interest in the AI Agent narrative.

AI Agent Overview

AI Agent is an "agent" based on large language models that can autonomously perceive, think, decide, and act. Its core functions include three steps: perception, planning, and action.

Unlike chatbots, AI Agents are designed to autonomously complete tasks without continuous human guidance. For example, when a user is ill, the AI Agent can automatically analyze symptoms, issue a sick leave note, order medication, and more.

AI Agent products in the Web2 domain, such as Perplexity, CrewAI, AutoGPT, and MultiOn, can be used for batch document processing, information integration, and more. MultiOn aims to simplify user interaction with the internet, such as automatically completing search tasks for specific videos on YouTube.

Innovation of Web3 x AI Agent

Web3 brings the following advantages for AI Agents:

1. Anti-censorship: Decentralized LLM can avoid the biases of centralized models.
2. Decentralization/Owership: Data is no longer centralized with AI providers.
3. Monetization: Providing new avenues for developers and investors
4. Composability: Interoperability between Agents

AI Agent also brings transformation to Web3:

1. Conversational AI Agent on the chain, simplifying on-chain information retrieval.
2. Change the way users interact with Web3, such as automating airdrops, building trading bots, etc.

Web3 x AI Agent projects can be divided into two categories:

• Web2-style on-chain dialogue AI Agent: for learning and research, does not include on-chain operations
• Web3 Native AI Agent: Capable of on-chain interaction

Spectral Project Analysis

Spectral has transformed from the Ethereum credit scoring protocol into a machine intelligence network, allowing users to build on-chain AI Agents.

Main products:

• Spectral Syntax: Officially developed collection of Agents
• Spectral Nova: Decentralized Machine Learning Inference Platform
• Agent Wallets: Wallets integrated with AI Agent
• Inferchain: Layer 1 powered by AI Agents

Development History:

• Completed financing in 2021-2022
• Syntax will be launched in March 2024.
• 2024.05 TGE and Q1 Airdrop
• Collaborating with Turnkey in June 2024, launching Agent Wallet in Q3

Spectral aims to achieve transparency, decentralization, and verifiability of AI applications in the Web3 space.

Token Economics:

• $SPEC is used for payment and governance
• Circulating Market Cap 85 million USD, FDV 800 million USD
• Current circulation accounts for 10.3% of the total.
• Core contributors and investors will unlock starting from 2025.05

Financing background: including well-known institutions such as Galaxy, General Catalyst, and Social Capital.

Use data:

• Registered users 65,362
• Number of contracts generated 1,055,568
• Average interactions 25/user

GaiaNet Project Analysis

GaiaNet is a distributed AI infrastructure dedicated to becoming a decentralized AI Agent ecosystem.

Core components:

• GaiaNet Node: Software stack for quick deployment of AI Agents
• GaiaNet Domain: A collection of nodes in a specific domain
• GaiaNet DAO: Manage Ecosystem

GaiaNet moves the Web2 AI Agent ecosystem on-chain, simplifying the Agent deployment process.

Team Background:

• Matt Wright: Former Community Lead at Consensys
• Shashank Sripada: Venture Capital Background
• Sydney Lai: Co-founder and CTO of EVM Capital

Financing situation:

• May 28, 2024 Seed Round 10 million USD
• Investors include Mantle Network, ByteTrade, EVM Capital, etc.

Use data:

• 35 available conversational AI agents
• Total number of nodes 18,594

Other Early Projects

• Zotto: Users can create trading intention AI Agent
• AgentLayer: AI Agent Layer2 based on OP Stack
• Olas Network: Off-chain AI Agent Ecosystem
• Theoriq: Modular AI Agent Foundation Layer
• AgentCoin: Universal Web3 AI Agent
• Giza: Web3 AI Agent framework, using ZKML