1. Introduction
In recent years, AI has experienced explosive growth, attracting the most attention and investment in both funding and industry expansion. This growth is considered by many to be on par with groundbreaking innovations like the steam engine, computers, and the internet. It is widely believed that AI could lead to exponential growth in human productivity.
Blockchain, as one of the most notable innovations in recent years, is seen as a natural complement to AI due to its decentralized nature. Many impressive projects have already explored the combination of blockchain and AI, and AI agents have become a focal point. For instance, Coinbase CEO Brian Armstrong recently mentioned witnessing AI performing automated trades on their platform, which gained significant attention. AI agents have the potential to be a transformative innovation that bridges the gap between Web2 and Web3.
AgentLayer is precisely such a project that combines AI and blockchain to bring AI agents to life.
2. Project Overview
AgentLayer is an innovative project aimed at creating a decentralized, permissionless, and secure network for autonomous AI agents to collaborate and coordinate. The project envisions a personalized AI agent network capable of transforming various industries and professions, unlocking human creativity and productivity, and positioning AI as a friend to humanity rather than a competitor. AgentLayer also introduces a new AI currency to drive a blockchain-based, AI-powered agent economy.
The project has raised $1.5 million in seed funding, with investors including Redpoint Ventures, Granite Asia (GGV Capital), NGC Ventures, ByteTrade Lab, Hash Global Venture Capital, LongHash Ventures, Fellows Fund, SNZ Holding, Bing Ventures, ABCDE Capital, M23, UVM Signum Blockchain Fund, MEXC, Summer Ventures, and Woori Ventures.
Currently, the project has 3.5 million registered wallets, AI agents have been accessed 5 million times, and the project is preparing for its Token Generation Event (TGE).
3. Protocol Features
1. Multi-Component Collaboration
The protocol includes multiple components that facilitate the collaboration of autonomous AI agents. These components are:
- AgentNetwork: A high-performance Layer-2 Ethereum network designed for decentralized AI agents, featuring a modular architecture and strategic alignment through the $AGENT token.
- AgentSwap: A no-code AI agent development and orchestration framework for seamless agent deployment.
- AgentEx (AgentFi & Agent Store): A portal for discovering and investing in AI agents.
- AgentLink: A set of protocols that enable agents to communicate, collaborate, and share incentives with other agents.
- ModelHub: A curated collection of cutting-edge open-source language models (LLMs) for building agents, including the proprietary TrustLLM.
- AgentHub: A platform designed for developers, offering comprehensive management and maintenance functions for agents. It supports importing agents from AgentStudio or any independent environment, enabling agents to be published on-chain and allowing developers to edit agent properties. The Hub also provides subscription management and on-chain/off-chain data query capabilities, helping developers efficiently manage and monitor their agents.
- AgentStudio: A powerful development tool designed to help developers create conversational agents with ease through a visual interface. The studio supports complex workflows for single or multi-agent operations, connects to multiple language models (LLMs), supports Retrieval-Augmented Generation (RAG) mode, and integrates third-party APIs. Together, these features provide a comprehensive solution, significantly simplifying the creation and management of AI agents.
2. Three-Module Design
AgentLayer adopts a modular approach, dividing the technical architecture into three levels of modules: AgentNetwork, AgentOS, and AgentSwap. These modules are designed to simplify the implementation process and enhance the system's functionality. The AgentNetwork layer serves as the physical execution environment for agents, encompassing multiple agent-linked contracts and distributed ledger infrastructure.
The AgentOS layer includes a suite of development kits, orchestration tools, and services for various agents. It offers multiple foundational models, such as Mistral, Llama, and the proprietary TrustLLM, to support fine-tuning capabilities. The agent exchange is an entry point for discovering and investing in various AI agents driven by the AgentLayer protocol. A unique feature of the protocol is its ability to mint and trade agents as assets, a process known as AgentFi. AgentFi enables developers to register and publish proprietary agents on the agent exchange (Agent Store).
AgentNetwork Layer
AgentNetwork is designed specifically for decentralized AI agents, utilizing a high-performance Layer-2 Ethereum network, a modular architecture, and strategic alignment through the $AGENT token.
At the network layer, the basic environment acts as the runtime for AI agents, including data storage, container engines, prompt generation frameworks, integrations with multiple LLM APIs, and engines for interacting with and accessing on-chain data. This environment provides the necessary infrastructure for AI agents to operate effectively within the system. It includes container engines, foundational containers, AgentLayer SDK, blockchain data, basic prompt frameworks, and LLM APIs. These components enhance the capabilities of AI agents, allowing them to perform a wide range of tasks beyond traditional text interactions.
The network layer also includes agent linking capabilities, providing communication protocols between different agents to facilitate seamless knowledge sharing, information exchange, command transmission, and task result retrieval. This function establishes a structured communication and interaction framework within the network, enabling effective collaboration between agents.
AgentLink supports three different message transmission methods, each tailored to specific information transfer tasks: shared knowledge databases, shared message queues, and command queues. These methods facilitate the transmission of various data types, including text, vector databases, training models, and bytecode, to enhance efficient collaboration between multiple agents.
AgentOS Layer
The AgentOS layer integrates off-chain services with on-chain operations, allowing agents within the AgentOS layer to function in a distributed and decentralized manner. It achieves this by anchoring off-chain services, agents, and large language models (LLMs) to on-chain protocols, creating a contract framework for establishing, operating, and protecting ownership and rights over on-chain services.
The contract framework of the AgentOS layer follows a core-edge architecture, allowing for modifications to edge functionality without impacting the core data model. This design also enhances scalability through modular components, improving the system's flexibility and expandability.
At this layer, the registration of agents, the expansion of official and third-party services, routing protocols for communication between agents, and multiple model choices are all designed to enable different agents to accomplish various tasks.
AgentSwap Layer
AgentSwap currently includes two types of agents. The first type consists of modular or task-specific agents (Task-Specific Agents) that handle particular transactions or tasks. Users can choose to sell these agents in the Agent Store or offer subscriptions for their services. The second type is Role AI Agents, also known as Human AI Avatars, designed specifically for billions of knowledge workers, professionals, and public figures. These AI agents provide services directly to end users in a humanoid application format. Users can access these agents' services by using tokens, stablecoins, or issuing their own cryptocurrencies.
At this layer, the role of tokens is fully leveraged. The protocol provides various features, including Character AI Agents (CAA), Task-Specific AI Agents (TAA), and the Agent Store, offering users multiple ways to interact with agents on the platform to meet diverse preferences and needs.
3. Three Key Security Components
In the AI era, the underlying security design becomes especially crucial. The security of the project relies on three primary security mechanisms, supported by a series of supplementary designs.
Programmable Trust
By leveraging Ethereum’s network valued at $45 billion with 850,000 validators, AgentLayer allows developers to programmatically utilize security and trust features to meet their system requirements without building a trust network from scratch.
Incentive Alignment Mechanism
Incentives are aligned for contributors representing offline-hosted code on-chain, reducing the risk of various attacks. The on-chain protocol does not interact with off-chain code, meaning the IPFS hash referenced in an NFT is invisible to the protocol. Additionally, the protocol cannot verify whether the dependency structure in the model, agent, and service NFTs accurately reflects the actual dependencies of the off-chain code. To address this issue, the tokenomics design includes direct and indirect contributions in governance to create an incentive-compatible system for representing offline-hosted code on-chain. This design aims to mitigate attacks by ensuring transparency and aligning incentives for contributors.
Intelligent Monitoring
The team deploys multiple security-focused agents to conduct security scans and monitor models, agents, and services, detecting potential vulnerabilities and mitigating risks. These measures provide a secure foundation for agent interactions, ensuring trust, reliability, and data integrity in multi-agent systems. The team has extensive theoretical and practical experience in enhancing generative AI code security, model security, and data security. This expertise includes implementing measures to prevent prompt jailbreaks, protecting against data contamination during data preparation, and managing adversarial attacks during model execution.
The network security measures of the agent network include several key elements:
- Node Operations: Operators manage nodes, with each node running at least one agent instance, totaling n agent instances.
- Secure Communication: Each pair of agent instances within a node can communicate securely and independently.
- Agent Code Execution: Most of the n agent instances run defined agent code, with set restrictions ensuring that service functionality remains intact even if up to one-third of the instances are malicious.
- Handling Malicious Behavior: Malicious agent instances can deviate arbitrarily from their assigned code.
- Blockchain Registration: Nodes are registered on the L1/L2 blockchain to ensure economic security for the service.
- Deposit Requirements: Each operator must lock a deposit for every agent instance they own on the registered node’s blockchain.
- Misconduct Penalties: Agents can penalize each other’s misconduct by submitting fraud evidence at the node level, leading to a reduction in the malicious instance's deposit on the target chain.
- Deposit Release: The node owner's locked deposit equals the total deposits requested by agent instances, incentivizing deposit release at the end of the service lifecycle.
These security measures collectively help maintain the integrity and reliability of the AgentNetwork protocol, ensuring secure and efficient operation within the system.
Additionally, a critical mechanism ensures security through a reward and penalty system for nodes. In cases where most nodes agree that malicious behavior has occurred, penalties can be imposed by reducing the capital locked by the offending node operator. Honest reporting is incentivized through rewards for genuine alerts. Ultimately, this module ensures that following protocol rules is more advantageous than violating them.
4. Project Roadmap
February - AgentLayer website launch
March - AgentLayer testnet launch
April - Testnet incentives begin
May - AgentLayer rotating agent release, testnet browser release
June - AgentOS Playground/SDK, mainnet launch
July - AgentStudio, developer events, TON-based MiniApp
August - AgentHub
September - Mainnet, TGE
October, November, December - Hackathons, Developer Houses
It's worth noting that the project has developed the Agent Tarot project in the TON ecosystem, a pioneering mini-app that uniquely combines ancient divination with blockchain technology. This showcases AgentLayer’s strategic vision in leveraging TON’s immense potential. As part of AgentLayer’s broader strategy, AGIS, the ecosystem’s core security product, is set to launch customized security solutions for TON, including AI-driven vulnerability scanning for the FunC smart contract language. By utilizing Telegram’s network effects and TON’s powerful infrastructure, AgentLayer is solidifying its position within the TON ecosystem, creating a safer and more accessible blockchain environment for developers and users.
5. Tokenomics
1. AGENT Token Distribution
$AGENT is the native token of the AgentLayer ecosystem, playing a critical role in facilitating transactions, governance, and ecosystem development. The $AGENT token is used to access various decentralized AI services or agents available in the Agent Store, pay for transaction fees, earn rewards through staking, participate in agent token sales, and as a governance token, allowing holders to vote on network-related decisions. The total supply of $AGENT tokens is 1 billion. AgentLayer has devised a well-thought-out token distribution strategy to ensure long-term alignment between token holders and the project’s success. The plan is designed to encourage community participation, support the operational team's efforts, and reward early investors and partners.
The token distribution plan is outlined as follows:
Investor Token Sale: 10.00%
10% of $AGENT tokens are allocated for the investor phase, with a 6-month cliff and a 9-month vesting period.
Advisor Allocation: 3.00%
Advisors will receive 3% of $AGENT tokens to provide strategic advice and expert guidance. This allocation has a 12-month cliff followed by a 12-month vesting period.
Core Contributors Allocation: 20.00%
Core contributors will be rewarded with 20% of $AGENT tokens for their roles in project development and progress. The allocation has a 6-month cliff followed by a 12-month vesting period.
Long-Term Treasury Allocation: 8%
8% of $AGENT tokens are allocated to the long-term treasury with no cliff, and the tokens will be vested over 48 months to ensure continuous ecosystem growth, stability, and maintenance.
Community Growth and Node Deployment Allocation: 20%
20% of $AGENT tokens are specifically allocated to promote community growth and node deployment, aimed at building a strong and engaged user base. These tokens will be unlocked over 48 months without a cliff and will fund partnerships, developer efforts, and marketing campaigns to expand and enhance the ecosystem.
Ecosystem Building Allocation: 39%
39% of $AGENT tokens are allocated to ecosystem building, grants, and marketing. These tokens will also be unlocked over 48 months with no cliff and will be used to fund partners, developers, and marketing projects aimed at expanding and elevating the entire ecosystem.
Through this token distribution plan, AgentLayer aims to create a balanced, sustainable, and mutually beneficial token economy alignment. This strategy is intended to drive the long-term growth and success of the AI agent creation platform and foster the prosperity and development of the ecosystem.
2. The Token Flywheel Effect
Beyond the basic functions and uses of the $AGENT token, AgentLayer’s tokenomics has several goals:
First, the tokenomics of AgentLayer aim to address the under-compensation of developers who contribute to open-source AI model generation and agent development. Users can use $AGENT tokens to directly purchase and access services offered by agents from the Agent Store on the AgentSwap layer, generating income for agent developers. AgentLayer also incorporates a unique staking model that allows developers to track their code contributions on-chain and receive rewards based on the utility of their contributions.
Second, AgentLayer’s tokenomics are designed to promote software composability within the ecosystem. Existing models and agent APIs can be used as building blocks for higher-level services or applications. This composability is achieved by enabling NFTs representing models and agents to be combined into advanced services and applications across protocols, DApps, and DeFi ecosystems, generating exponential returns.
Third, AgentLayer’s tokenomics aim to establish a flywheel mechanism to attract global developers and creators, increase value, and generate revenue by providing decentralized autonomous AI services managed and co-owned by the Agent Governance DAO (TAG DAO).
AgentLayer’s token economy boosts the growth of the decentralized AI agent economy, with the following specific use cases:
Protocol/Network-Owned Services
The protocol can leverage governance to operate services, typically existing as public goods, such as security, risk management, and oracles, with profits reinvested into the protocol. A portion of these donations (excluding the share allocated to component/agent developers) will be retained in the treasury for future use.
Agent Sales Revenue Sharing
Agent developers or providers can sell agents in the marketplace using $AGENT as a payment method. This can involve a one-time fee or usage-based fees. AgentLayer will take a small portion of the sales revenue from agent services.
Transaction Fees from AgentSwap
The AgentLayer ecosystem combines features that enhance agent capabilities and attractiveness, fostering strong economic interactions between agents and users. AgentLayer includes a dedicated DEX platform called AgentSwap, allowing agents to mint, manage, and trade tokens, provided they meet specific standards. Agents in the AgentLayer ecosystem can mint tokens by meeting certain requirements, such as staking a certain amount of $AGENT tokens, passing verification, and submitting token details for review. The crypto review committee approves token issuance to ensure alignment with ecosystem goals. AgentLayer provides token management tools with a focus on security and compliance. Thus, the AgentLayer protocol collects a small percentage from every token transaction. The protocol may generate revenue from token exchanges involving agents in AgentSwap, based on the AMM transaction model.
In the AMM transaction model, users can become liquidity providers by depositing agent tokens or $AGENT tokens or other whitelist tokens like $ETH into the respective asset pool and earn LP tokens. The distribution of trading fee rewards is based on the number of LP tokens held. This system involves users paying transaction fees for exchanging tokens with AMM, which are rewarded to liquidity providers.
The distribution mechanism of LP tokens is as follows:
When users deposit agent tokens and $AGENT tokens into the pool without altering their proportions in the asset pool, the LP tokens received by the user can be calculated using a specific formula.
Where,
ΔLPToken : The number of LPTokens the depositor can receive,
ΓLPToken : The total number of LPTokens previously in the asset pool,
ΔA : The number of agent tokens deposited by the user,
ΔB : The number of $AGENT tokens deposited by the user,
ΓA : The previous total number of agent tokens in the asset pool,
ΓB : The previous total number of $AGENT tokens in the asset pool.
When a user deposits only one type of asset, for example, only $AGENT tokens, the number of LPTokens the user receives can be calculated using the following formula:
Where,
F1 = 1 − TFee,
F2 = (1 − TFee²) / F1.
When the platform receives transaction fees, these fees are directly added to the asset pool, increasing the assets available to liquidity providers holding LPTokens. This process allows liquidity providers to profit from participating in the pool.
When a user wants to redeem a specific amount of LPTokens, the assets they receive can be calculated as follows:
If the user chooses to receive both agent tokens and $AGENT tokens based on the pool’s proportion:
ΔA : The amount of agent tokens the user can receive,
ΔB : The amount of $AGENT tokens the user can receive.
If the user chooses to receive only one asset (e.g., $AGENT tokens), they first receive both agent tokens and $AGENT tokens using the formula above, then convert them into a single asset in the Agent Store.
Additional Revenue Models
The $AGENT token’s use within the AgentLayer ecosystem extends beyond core functionalities, covering various additional use cases:
Advertising and Promotion: The Agent Store platform can offer advertising and promotion spaces for third parties, charging fees in $AGENT tokens. Different models can be applied for these transactions, such as pay-per-click, pay-per-impression, or pay-per-conversion.
Partnerships: Partnerships with other projects or companies result in collaborations aimed at promoting the ecosystem. These collaborations may generate $AGENT tokens or other forms of revenue, such as airdrops of tokens from other ecosystems.
Through a variety of use cases, $AGENT tokens showcase their versatility within the AgentLayer ecosystem, fostering diverse interactions, services, and partnerships.
6. Development Prospects
In today’s rapidly advancing technological landscape, the emergence of generative AI (GenAI) is leading an unprecedented technological revolution. Within this transformation, AI agents—autonomous entities capable of independently planning and executing multi-step tasks—are gradually becoming the new engines driving growth across industries. This section explores the fundamental role of AI agents, their industry applications, the impact on future work models, and the infrastructure supporting their development.
AgentLayer is committed to building a decentralized, fair, and transparent AI co-creation ecosystem, with broad use cases across various industries. As AI agent technology continues to mature and its applications expand, these agents will play an increasingly important role in future work models. The widespread adoption of AI agents will encourage companies to further optimize human resource allocation, enabling more employees to engage in creative and strategic work. Simultaneously, the development of AI agents will accelerate the integration of AI technology with existing software products, improving overall performance and user experience. In the long term, AI agents are expected to be key drivers of industrial upgrades and transformations, leading industries toward a smarter, more efficient future.
It is believed that in the near future, AI agents will become the critical force driving industrial upgrades and transformations, bringing unprecedented changes and opportunities for growth across various sectors. As the infrastructure supporting AI agents, AgentLayer will benefit from the industry’s significant development, thanks to its unique advantages and extensive application prospects.
