Trusta.AI: Bridging the Trust Gap Between Humans and Machines

1. Introduction

The Web3 ecosystem is undergoing significant changes, with AI-driven on-chain agents rapidly becoming the primary mode of interaction. It is expected that within the next 2-3 years, these AI agents with autonomous decision-making capabilities will lead the large-scale adoption of on-chain transactions and interactions, potentially replacing 80% of on-chain human behavior and becoming true "users" on the chain.

The rapid rise of AI agents has also brought unprecedented challenges: how to identify and authenticate the identities of these agents? How to assess the credibility of their actions? In a decentralized and permissionless network, how to ensure that these agents are not misused, manipulated, or used for attacks?

Establishing an on-chain infrastructure that can verify the identity and reputation of AI Agents has become a core proposition for the next phase of Web3 evolution. The design of identity recognition, reputation mechanisms, and trust frameworks will determine whether AI Agents can truly achieve seamless collaboration with humans and platforms, and play a sustainable role in the future ecosystem.

2. Project Analysis

2.1 Project Introduction

Trusta.AI is committed to building Web3 identity and reputation infrastructure through AI.

Trusta.AI has launched the first Web3 user value assessment system - MEDIA Reputation Score, building the largest real-person certification and on-chain reputation protocol in Web3. It provides on-chain data analysis and real-person certification services for multiple top public chains, exchanges, and leading protocols. Over 2.5 million on-chain certifications have been completed on multiple mainstream chains, making it the largest identity protocol in the industry.

Trusta is expanding from Proof of Humanity to Proof of AI Agent, establishing a threefold mechanism of identity establishment, identity quantification, and identity protection to achieve on-chain financial services and on-chain social interactions for AI Agents, thereby building a reliable trust foundation in the era of artificial intelligence.

2.2 Trust Infrastructure - AI Agent DID

In the future Web3 ecosystem, AI Agents will play a crucial role. They can not only complete interactions and transactions on-chain but also perform complex operations off-chain. However, distinguishing between genuine AI Agents and human-intervened operations is central to decentralized trust—without a reliable identity verification mechanism, these agents can easily be manipulated, defrauded, or abused. This is why the multiple application attributes of AI Agents in social, financial, and governance contexts must be built on a solid identity verification foundation.

The application scenarios of AI Agents are becoming increasingly rich, covering multiple fields such as social interaction, financial management, and governance decision-making, with their autonomy and intelligence level continuously improving. It is therefore crucial to ensure that each agent has a unique and trustworthy identity identifier (DID). Without effective identity verification, AI Agents may be impersonated or manipulated, leading to a collapse of trust and security risks.

In the future, in a Web3 ecosystem fully driven by intelligent agents, identity authentication is not only the cornerstone of security but also a necessary defense to maintain the healthy operation of the entire ecosystem.

As a pioneer in the field, Trusta.AI has taken the lead in building a comprehensive AI Agent DID certification mechanism with its advanced technological strength and rigorous credibility system, providing a solid guarantee for the trustworthy operation of intelligent agents, effectively preventing potential risks, and promoting the robust development of the Web3 smart economy.

Project Overview 2.3

2.3.1 Financing Situation

January 2023: Completed a $3 million seed round financing, led by SevenX Ventures and Vision Plus Capital, with other participants including HashKey Capital, Redpoint Ventures, GGV Capital, SNZ Holding, and others.

June 2025: Completion of a new round of financing, with investors including ConsenSys, Starknet, GSR, UFLY Labs, and others.

2.3.2 Team Situation

Peet Chen: Co-founder and CEO, former Vice President of Ant Digital Technology Group, Chief Product Officer of Ant Security Technology, and former General Manager of ZOLOZ Global Digital Identity Platform.

Simon: Co-founder and CTO, former head of AI Security Lab at Ant Group, with fifteen years of experience in applying artificial intelligence technology to security and risk management.

The team has a deep technical accumulation and practical experience in artificial intelligence and security risk control, payment system architecture, and identity verification mechanisms. They have long been committed to the in-depth application of big data and intelligent algorithms in security risk control, as well as security optimization in underlying protocol design and high-concurrency trading environments, demonstrating solid engineering capabilities and the ability to implement innovative solutions.

3. Technical Architecture

3.1 Technical Analysis

3.1.1 Identity Establishment - DID + TEE

Through a dedicated plugin, each AI Agent obtains a unique decentralized identifier (DID) on the chain, and securely stores it in a trusted execution environment (TEE). In this black-box environment, critical data and computational processes are completely hidden, sensitive operations remain private at all times, and external parties cannot peek into the internal operations, effectively building a solid barrier for the information security of AI Agents.

For agents that were generated before the plugin integration, identity identification is based on the comprehensive scoring mechanism on the chain; while agents that are newly integrated with the plugin can directly obtain the "certificate of identity" issued by the DID, thereby establishing an AI Agent identity system that is autonomous, controllable, authentic, and tamper-proof.

3.1.2 Identity Quantification - The First SIGMA Framework

The Trusta team always adheres to the principles of rigorous evaluation and quantitative analysis, committed to building a professional and trustworthy identity authentication system.

The Trusta team first built and validated the effectiveness of the MEDIA Score model in the "human proof" scenario. This model comprehensively quantifies on-chain user profiles from five dimensions: Interaction Amount ( Monetary ), Participation ( Engagement ), Diversity ( Diversity ), Identity ( Identity ), and Age ( Age ).

The MEDIA Score is a fair, objective, and quantifiable on-chain user value assessment system. With its comprehensive evaluation dimensions and rigorous methods, it has been widely adopted by several leading public chains as an important reference standard for airdrop eligibility screening. It not only focuses on interaction amounts but also encompasses multi-dimensional indicators such as activity level, contract diversity, identity characteristics, and account age, helping project teams accurately identify high-value users and improve the efficiency and fairness of incentive distribution, fully reflecting its authority and wide recognition in the industry.

Based on the successful establishment of a human user evaluation system, Trusta has migrated and upgraded the experience of the MEDIA Score to the AI Agent scenario, creating a Sigma evaluation system that is more aligned with the behavioral logic of intelligent agents.

• Specialization Specification: The expertise and level of specialization of the agent.
• Influence: The social and digital influence of agents.
• Engagement: The consistency and reliability of its on-chain and off-chain interactions.
• Monetary: The financial health and stability of the proxy token ecosystem.
• Adoption Rate: The frequency and efficiency of AI agent usage.

The Sigma scoring mechanism constructs a logical closed-loop evaluation system from "capability" to "value" based on five major dimensions. MEDIA focuses on assessing the multifaceted engagement of human users, while Sigma pays more attention to the professionalism and stability of AI agents in specific fields, reflecting a shift from breadth to depth, which better meets the needs of AI agents.

First, based on the professional capability ( Specification ), the degree of engagement ( Engagement ) reflects whether it is consistently and steadily involved in practical interactions, which is a key support for building subsequent trust and effectiveness. Influence ( Influence ) is the reputation feedback generated in the community or network after participation, representing the credibility and dissemination effect of the agent. Monetary ( Monetary ) assesses whether it has the ability to accumulate value and financial stability in the economic system, laying the foundation for a sustainable incentive mechanism. Finally, adoption ( Adoption ) is used as a comprehensive representation, indicating the degree to which the agent is accepted in practical use, serving as the final validation of all preceding capabilities and performances.

This system is layered and structured clearly, capable of comprehensively reflecting the overall quality and ecological value of AI Agents, thus achieving a quantitative assessment of AI performance and value, transforming abstract advantages and disadvantages into a concrete and measurable scoring system.

Currently, the SIGMA framework has advanced cooperation with well-known AI Agent networks such as Virtual, Elisa OS, and Swarm, demonstrating its immense application potential in AI agent identity management and reputation system construction, and is gradually becoming the core engine for promoting trustworthy AI infrastructure development.

3.1.3 Identity Protection - Trust Assessment Mechanism

In a truly resilient and highly trustworthy AI system, the most critical aspect is not only the establishment of identity but also the continuous verification of that identity. Trusta.AI introduces a continuous trust assessment mechanism that allows for real-time monitoring of certified intelligent agents to determine whether they are being illegally controlled, under attack, or subjected to unauthorized human intervention. The system identifies potential deviations during the agent's operation through behavior analysis and machine learning, ensuring that every agent action remains within the established policies and framework. This proactive approach ensures that any deviation from expected behavior is detected immediately, triggering automated protective measures to maintain the integrity of the agent.

Trusta.AI has established a security guard mechanism that is always online, continuously reviewing every interaction process to ensure that all operations comply with system standards and established expectations.

3.2 Product Introduction

3.2.1 AgentGo

Trusta.AI assigns decentralized identity identifiers (DID) to each on-chain AI Agent and rates them based on on-chain behavior data, building a verifiable and traceable trust system for AI Agents. Through this system, users can efficiently identify and filter high-quality agents, enhancing their experience. Currently, Trusta has completed the collection and identification of AI Agents across the network and issued decentralized identifiers to them, establishing a unified summary index platform ---- AgentGo, further promoting the healthy development of the intelligent agent ecosystem.

1. Human users query and verify identity:

Through the Dashboard provided by Trusta.AI, human users can easily retrieve the identity and reputation score of a specific AI Agent to assess its trustworthiness.

• Social group chat scenario: In a project team using an AI Bot to manage the community or speak, community users can verify through the Dashboard whether the AI is a genuine autonomous agent, avoiding being misled or manipulated by "pseudo-AI".

2. AI Agent automatically invokes indexing and verification:

AI can directly read index interfaces between each other to achieve quick confirmation of each other's identity and credibility, ensuring the safety of collaboration and information exchange.

• Financial regulatory scenario: If an AI agent issues a currency autonomously, the system can directly index its DID and rating to determine whether it is a certified AI Agent, and automatically link to the data platform to assist in tracking its asset circulation and issuance compliance.
• Governance Voting Scenario: When introducing AI voting in governance proposals, the system can verify whether the initiator or participant in the voting is a real AI Agent, preventing the voting rights from being controlled and abused by humans.
• DeFi Credit Lending: Lending protocols can grant AI Agents different amounts of credit loans based on the SIGMA scoring system, forming native financial relationships between agents.

The AI Agent DID is no longer just an "identity"; it has become the underlying support for building core functions such as trusted collaboration, financial compliance, and community governance, making it an essential infrastructure for the development of the AI-native ecosystem. With the establishment of this system, all verified secure and trustworthy nodes form a closely interconnected network, enabling efficient collaboration and functional interconnection among AI Agents.

Based on Metcalfe's Law, the value of the network will grow exponentially, thereby driving the construction of a more efficient, trust-based, and collaborative AI Agent ecosystem, achieving resource sharing, capability reuse, and continuous value addition among agents.

AgentGo, as the first trusted identity infrastructure for AI Agents, is providing essential core support for building a highly secure and collaborative intelligent ecosystem.

3.2.2 TrustGo

TrustGo is an on-chain product developed by Trusta.