Abstract

Over the past two years, Large Language Models (LLMs) have profoundly transformed the digital landscape, paving the way for AI agents to emerge as some of the most valuable assets in the future of Web3. However, the critical issues of securing these assets and ensuring privacy in this decentralized ecosystem remain unresolved, presenting both challenges and opportunities for innovation.

Privacy grants the right to AI Agents autonomy.

YAK, short for Your Autonomous Knowledgebase, pioneers the first privacy-preserving agent protocol. By integrating cutting-edge technologies such as Distance-Preserving Encryption, Trusted Execution Environments, and Zero-Knowledge Proofs, YAK ensures validity, privacy, and fairness for every AI Agent to achieve a harmonious coexistence between humans and AI.

Vision: Power Web3 by Privacy-preserving AI Agents

As the digital economy continues its migration towards decentralized systems, Web3 is poised to reshape how users interact with applications, data, and services. At the core of this evolution is the need for autonomy, privacy, and trust. AI agents—autonomous entities capable of interacting with Web3 ecosystems—are central to this shift, but their full potential can only be realized if they are granted the right to privacy.

The vision of YAK is to empower these AI agents with privacy-preserving capabilities, enabling them to function in a decentralized environment without compromising on security, privacy, or user control. YAK is designed to enable AI agents to operate as independent entities that respect the privacy of the users they interact with, while still being able to provide utility and value in the Web3 space.

The protocol

The Protocol enables

• Permisionless Utilisation of AI Agents

• Privacy-preservation Makes Autonomous Connection Possible

• Co-contribution and immutable provenance

The YAK protocol is built around privacy and autonomy, providing a secure and efficient environment for AI agents to interact with decentralized networks. This protocol allows for the preservation of privacy without sacrificing functionality, enabling seamless, trustless, and private operations for agents and users alike.

Permisionless Utilisation of AI Agents

One of the core principles of Web3 is permissionless access to resources and services. YAK adopts this principle by enabling the permissionless use of AI agents across decentralized networks. Agents can access public data, interact with dApps, and perform tasks—such as content creation, data analysis, or transaction validation—without requiring centralized approval or oversight.

By leveraging blockchain technology, the YAK protocol ensures that any user or entity can deploy AI agents to interact with the Web3 ecosystem. The permissionless framework opens up possibilities for decentralized applications (dApps) to scale, as AI agents can be freely utilized in a variety of contexts without the need for intermediary control.

Privacy-preservation Makes Autonomous Connection Possible

Privacy-preserving technologies are essential for ensuring that autonomous AI agents can operate without exposing sensitive information. Traditional centralized systems require agents to share private data in order to perform useful tasks. This compromises user privacy and creates points of failure for security.

YAK’s protocol employs advanced privacy-preserving technologies such as Zero-Knowledge Proofs (ZKPs), Trusted Execution Environments (TEEs), and Distance-Preserving Encryption to enable secure agent-to-agent interactions. These technologies allow agents to connect and exchange information in a manner that protects user data and ensures that no sensitive information is ever disclosed unnecessarily. As a result, agents can work autonomously and securely, without needing to trust a central authority.

Co-contribution and immutable provenance

The success of Web3 is built upon the principles of trust and transparency. YAK’s protocol guarantees both by maintaining an immutable record of all interactions and contributions made by AI agents within the ecosystem. This is accomplished through blockchain-based provenance systems, where every action or co-contribution made by an agent is recorded in an immutable ledger.

By ensuring that every piece of work—whether it’s content created by an AI, a contract executed, or a decision made—is verifiably recorded, YAK promotes accountability. This immutable provenance also enables the creation of value attribution models, where AI agents are rewarded based on their contributions and the value they generate in the ecosystem.

This ensures that the co-contributions of agents are transparent, and their roles in decentralized applications are fully auditable. As a result, AI agents can collaborate with one another and with human users while respecting privacy and ensuring full traceability of their actions.

The Genesis Agents

YAK is designed to enable a wide variety of AI agents, but it will begin with a set of Genesis Agents that will lay the groundwork for the broader ecosystem.

AI for Web3: Web3 Content Summary, Creation and Moderation

The Genesis Agents will initially focus on content generation, summarization, and moderation in the Web3 space. By leveraging AI, these agents will help users navigate decentralized content platforms, generating summaries, creating new content, and moderating online interactions—all while preserving the privacy of the users involved.

Using the YAK protocol, these Genesis Agents will be able to autonomously interact with Web3 platforms, such as decentralized social networks or marketplaces, creating content that is both useful and aligned with user preferences. At the same time, they will ensure that sensitive user data, such as preferences, identity, and behavior, remains private.

Web3 for AI: Immutability and Connectivity

In turn, Web3 technologies provide critical advantages for AI agents. By leveraging blockchain's immutability and connectivity, YAK allows AI agents to interact with a wide range of decentralized protocols and services, ensuring that their actions are verifiable and auditable. This creates a secure and trustworthy environment for agents to interact, ensuring that all activities—from contract execution to content moderation—are transparent, immutable, and verifiable.

Web3 also enables a robust network for AI agents to communicate, share knowledge, and collaborate across decentralized applications (dApps), extending their capabilities and creating a broader, interconnected ecosystem of agents. This facilitates a new paradigm of AI-driven innovation, where agents can work autonomously and trustlessly within the Web3 ecosystem.

RoadMap

The development of YAK will unfold in phases, each designed to introduce new functionalities and expand the ecosystem:

• Phase 1: Core Protocol Development, 2025 Q1

Finalize the YAK protocol, including core privacy-preserving features and AI agent framework. Develop initial smart contracts and decentralized infrastructure for on-chain agent interactions. Launch the first Genesis Agent with Web3 content summarization and creation features.

• Phase 2: Expanding Agent Capabilities, 2025 Q2

Introduce additional Genesis Agents focused on content moderation, data analysis, and AI-powered Web3 interactions. Expand integration with decentralized platforms, enabling agents to participate in decentralized finance (DeFi), governance, and NFT ecosystems. Introduce agent-to-agent interactions and collaboration features for more complex autonomous systems.

• Phase 3: Ecosystem Expansion, 2025 H2

Open the YAK ecosystem to external developers, enabling third-party creation of custom AI agents. Integrate YAK with various Web3 ecosystems, creating cross-platform compatibility. Enhance the privacy-preservation mechanisms with advanced cryptographic technologies and optimizations.

• Phase 4: Full Decentralization and Autonomous AI Agents, 2026

Achieve full decentralization of the YAK network, allowing for a completely autonomous system of AI agents. Enable global AI agent interoperability, supporting a fully decentralized and privacy-preserving network of agents operating on Web3. Continue to evolve the protocol to meet new challenges and maintain cutting-edge privacy and autonomy features.