Hangzhou, China — September 29, 2025 — Endless Protocol proudly extends its congratulations on the establishment of the Global Alliance for AI and Intangible Valuation (GAIIV), an international initiative co-founded by Professor Yu Xiong, President of Endless Protocol, Professor at the University of Surrey, and Fellow of the UK Academy of Social Sciences.

The Alliance, officially launched in Hangzhou, brings together over 20 leading universities and institutions — including the University of Oxford, University of Surrey, Tsinghua University, Fudan University, Zhejiang University, and the University of Macau — alongside major enterprises and experts in innovation, finance, and digital economy. Its mission is to create a global framework for AI-driven valuation of intangible assets, such as data, algorithms, intellectual property, and digital identities.

Redefining Value in the Age of AI

In a rapidly evolving digital economy, intangible assets — from data and cultural IP to technological know-how and brand capital — have become the driving forces of global innovation. Yet, their valuation remains one of the most complex challenges facing modern economies.

Speaking at the Alliance’s founding symposium, Professor Yu Xiong emphasized that the intersection of artificial intelligence, blockchain, and metaverse technologies will fundamentally redefine how societies perceive and measure value.

“We are entering an era where AI not only powers innovation but transforms how value itself is created, represented, and exchanged,” said Professor Xiong.
“Many of tomorrow’s assets will exist entirely within virtual environments. Their valuation will depend on intelligent, transparent systems that can recognize intangible worth across digital ecosystems.”

Endless Protocol’s Aligned Vision

As President of Endless Protocol, Professor Xiong brings to the Alliance a deep commitment to building technological foundations for the next generation of intangible assets.

Endless Protocol is a next-generation Web3 infrastructure platform that enables the generation, tokenization, and valuation of intangible assets using AI and blockchain technologies. The protocol is designed to empower creators, institutions, and AI systems to participate in a trusted, borderless digital economy.

“At Endless Protocol, we are developing an open ecosystem where ideas, algorithms, and digital identities can evolve into measurable and tradable forms of value,” Professor Xiong added.
“Our vision naturally complements the work of the Global Alliance for AI and Intangible Valuation — both aim to define how humanity recognizes and manages value in the intelligent era.”

A Global Collaborative Effort

The Alliance’s founding meeting gathered distinguished academics and leaders from across the world.
Speakers included Professor Xiaolan Fu (University of Oxford), Professor Zhang Jun (Dean, School of Economics, Fudan University), Professor Wu Xiaobo (Zhejiang University), Professor Qian Jun (Dean, Fanhai International School of Finance, Fudan University), and Professor Chas Bountra (Pro-Vice-Chancellor for Innovation, University of Oxford).

The establishment of GAIIV marks a milestone in global collaboration around AI, valuation science, and digital transformation, with a shared mission to shape international standards for the future of intangible assets.

About Endless Protocol

Endless Protocol is a pioneering Web3 infrastructure platform that integrates artificial intelligence, blockchain, and metaverse technologies to enable the creation, tokenization, and valuation of intangible assets. Led by President Professor Yu Xiong, the company is building a transparent, intelligent economy where innovation and intangible value can thrive beyond borders.

Associate Vice President, University of Surrey

The battle between the White House and Harvard University over a $2.2 billion federal funding freeze and demands to ban international students is no isolated attack. It’s part of a broader war on liberal higher education — and a harbinger of a wider global struggle.

A federal court ruling may have temporarily blocked the student ban, but the message is clear: these attacks are ideological, deliberate, and dangerous.

The 24 universities backing Harvard’s lawsuit know this is bigger than campus politics. Undermining academia weakens one of the last independent institutions shaping AI’s impact on society.

By weakening the institutions that embed human knowledge and ethical reasoning into AI, we risk creating a vacuum where technological power advances without meaningful checks, shaped by those with the fastest resources, not necessarily the best intentions.

The language used in discussions about ethical AI — terms like “procedural justice,” “informed consent,” and “structural bias” — originates not from engineering labs, but from the humanities and social sciences. In the 1970s, philosopher Tom Beauchamp helped author the Belmont Report, the basis for modern medical ethics. Legal scholar Alan Westin’s work at Columbia shaped the 1974 Federal Privacy Act and the very notion that individuals should control their own data.

This intellectual infrastructure now underpins the world’s most important AI governance frameworks. Liberal arts scholars helped shape the EU’s Trustworthy AI initiative and the OECD’s 2019 AI Principles — global standards for rule of law, transparency, and accountability. U.S. universities have briefed lawmakers, scored AI companies on ethics, and championed democratized access to datasets through the bipartisan CREATE AI Act.

But American universities face an onslaught. Since his inauguration, Trump has banned international students, slashed humanities and human rights programs, and frozen more than $5 billion in federal funding to leading universities like Harvard.

These policies are driving us into a future shaped by those who move fastest and break the most.

Left to their own devices, private AI companies give lip service to ethical safeguards, but tend not to implement them. And several, like Google, Meta, and Amazon, are covertly lobbying against government regulation.

This is already creating real-world harm. Facial recognition software routinely discriminates against women and people of color. Denmark’s AI-powered welfare system discriminates against the most vulnerable. In Florida, a 14-year-old boy died by suicide after bonding with a chatbot that reportedly included sexual content.

The risks compound when AI intersects with disinformation, militarization, or ideological extremism. Around the world, state and non-state actors are exploring how AI can be harnessed for influence and control, sometimes beyond public scrutiny. The Muslim World League (MWL) has also warned that groups like ISIS are using AI to recruit a new generation of terrorists. Just last month, the FBI warned of scammers using AI-generated voice clones to impersonate senior U.S. officials.

What’s needed is a broader, more inclusive AI ecosystem — one that fuses technical knowledge with ethical reasoning, diverse cultural voices, and global cooperation.

Such models already exist. The Vatican’s Rome Call for AI Ethics unites tech leaders and faith groups around shared values. In Latin America and Africa, grassroots coalitions like the Mozilla Foundation have helped embed community voices into national AI strategies.

For instance, MWL Secretary-General Mohammad Al-Issa recently signed a landmark long-term memorandum of understanding with the president of Duke University, aimed at strengthening interfaith academic cooperation around shared global challenges. During the visit, Al-Issa also delivered a keynote speech on education, warning of the risks posed by extremists exploiting AI. Drawing on his work confronting digital radicalization by groups like ISIS, he has emerged as one of the few global religious figures urging faith leaders to be directly involved in shaping the ethical development of AI.

The United States has long been a global AI leader because it draws on diverse intellectual and cultural resources. But that edge is fading. China has tripled its universities since 1998 and poured billions into state-led AI research. The EU’s newly passed AI Act is already reshaping the global regulatory landscape.

The world needs not just engineers, but ethicists; not just coders, but critics. The tech industry may have the tools to build AI, but it is academia that holds the moral compass to guide it.

If America continues undermining its universities, it won’t just lose the tech race. It will forfeit its ability to lead the future of AI.

Professor Yu Xiong is Associate Vice President at the University of Surrey and founder of the Surrey Academy for Blockchain and Metaverse Applications. He chaired the UK All-Party Parliamentary Group on Metaverse and Web 3.0 advisory board.

July 13, 2025 — London — At the Brunel Hack 25 — Blockchain Hackathon & Festival, Endless Web3 Genesis Cloud (Endless Protocol) announced the successful conclusion of its Developer Grant Program — First Stage. After two rounds of expert evaluations and project interviews, StarAI, Funmeta, and VDEP have emerged as Tier 1 winners, each receiving $200,000 in funding.

Held over July 12–13, the hackathon was partnering with 10 leading UK universities, including Oxford, Cambridge, Imperial, Westminster, and Southampton, representing top communities in blockchain, fintech, and decentralized technologies. Endless Protocol, in collaboration with its ecosystem’s decentralized social app Luffa, launched two competition bounties: building cross-chain tools or AI Agents for Web3 social, with a total prize pool of $10,000. Endless AI scientist Joe also demonstrated relevant DEMOs to developers at the on-site Workshop. Enthusiastic participants actively engaged, giving rise to outstanding projects like the AI Booking Assistant and the on-chain game Arena.

Ned, Chief Token Economist of Endless, delivered a keynote speech, introducing developers to the technical features, ecosystem advantages, and developer incentive policies of Endless and Luffa, while announcing the winners of Endless Developer Grant Program — First Stage.

These projects will collaborate with Endless and its Luffa ecosystem to develop innovative applications across multiple tracks, including AI and gaming. Their efforts will focus on deep technical, product, and market integration with the Endless economic system, driving the advancement of AI-powered Web3 applications and seamless Web2-to-Web3 transitions.

Ned also revealed that Endless and Luffa are planning the next phase of the Developer Incentive Program, inviting more AI Agent and mini-program projects to stay tuned for updates and registration details on the official website.

Endless Web3 Genesis Cloud, a decentralized intelligent protocol bridging Web2 and Web3 ecosystems, has established strategic partnerships with renowned institutions such as the University of Surrey, Alibaba Cloud, Stability AI, and Foresight Ventures. The Endless mainnet, launched on March 6, 2025, leverages a Move-based public chain integrated with various intelligent components to simplify the transition from Web2 applications and developers to Web3. Its mission is to lower the barriers to Web3 entry for developers while delivering a Web2-like experience for users, harnessing the full potential of decentralization and artificial intelligence.

The success of the first-stage Developer Grant Program marks a significant milestone in building the Endless ecosystem and advancing Web3 innovation.

Two Strategic Directions Attract Over 50 Projects

Launched on April 1, 2025, the first-stage Developer Grant Program invited global developers to build decentralized applications (DApps) on the Move-based Endless public chain, encouraging exploration of two key directions: AI Agent + Web3 and Web2-to-Web3 migration. The initiative attracted participation from over 50 projects.

The Endless Foundation evaluated projects based on originality, the fusion of AI and decentralization, depth of Endless utilization, technical execution, user experience, and commercial viability. Based on these criteria and the strategic directions, StarAI, Funmeta, and VDEP were selected as Tier 1 winners, each exceptionally awarded $200,000.

The applicant pool included projects already listed on top exchanges like Binance and those previously funded in ecosystems such as Sol, SUI, Aptos, and BNB Chain. Endless extends its gratitude to all participants for their support and recognition of the ecosystem’s vision, with many expressing interest in exploring further collaborations.

Winning Projects to Deeply Integrate with Endless and Luffa

StarAI, one of the Tier 1 projects, is a leading AI + Web3 content creation platform, empowering over 8 million creators. With powerful AI tools and an open ecosystem, StarAI enables creators to monetize their work and earn sustainable income through Create-to-Mine, Pay-to-Earn, and Trade-to-Earn incentives.

StarAI will deploy its proprietary large language model and AIGC system on Endless’ infrastructure, integrating with the Luffa privacy-focused social ecosystem based on the Endless protocol. Key developments include NFT minting, trading marketplaces, and AI-driven task systems, establishing the world’s first AI-powered on-chain social experience.

Funmeta, a comprehensive DePin + game + AI service provider, offers game centers, guild management systems, and blockchain gaming infrastructure for players, guilds, and third-party developers, while pioneering entertainment innovations with trendy smart hardware as a new entry point. Its past project, ForthBox, minted over 60,000 Hams NFTs and recorded millions of on-chain interactions, ranking among the top 10 games on BNB Chain according to DappRadar historical data. Funmeta will leverage Endless’ SDK and APIs to enhance its gaming ecosystem, integrating $EDS as a core payment token and creating exclusive game content and benefits for $EDS holders to encourage long-term retention and build a sustainable in-game economy.

VDEP is a decentralized digital entertainment platform built on the Endless, focusing on creating a “fair, transparent, entertainment-driven, and profitable” Web3 game ecosystem holding an official license from the Vanuatu government. It integrates multiple on-chain games with a pre-disclosed result mechanism, ensuring transparent, verifiable gameplay and immutable data.

Currently integrated with the Luffa platform, VDEP allows users to access games with a single-click via the Luffa wallet, eliminating complex registration, and delivers a Web2-like experience through deep integration with Endless and Luffa social components. VDEP will expand its GameFi offerings, leveraging Endless’ technical support and integrating with upcoming ecosystem projects such as the “Taoist application” and “Bmiss live-streaming platform” to further boost on-chain activity.

A Milestone for Endless and Web3 Innovation

The first-stage Developer Grant Program underscores Endless’ commitment to nurturing a vibrant developer community. The winning projects will receive ongoing support, including cloud computing resources, technical mentorship, and introductions to investment partners.

By providing funding, technical resources, and incubation support, Endless will keep empowering projects like StarAI, Funmeta, and VDEP to push the boundaries of AI and Web3 convergence.

About Endless Web3 Genesis Cloud

Endless Web3 Genesis Cloud is a decentralized intelligent protocol designed to connect Web2 and Web3 ecosystems, offering developers a one-stop platform for Web3 application development and delivering a Web2-grade experience for users. Built on a Move-based public chain, Endless integrates advanced AI capabilities and plugins, aiming to be the premier connector between AI and Web3. It enables developers to build Crypto AI applications swiftly and modularly, paving the way for the advent of AI Agentic Super Intelligent Systems.

Website: https://www.endless.link/

Twitter: https://x.com/EndlessProtocol

Telegram: https://t.me/EndlessProtocol

About Luffa

Luffa is a social platform with distributed and encryption technology , robustly protecting user privacy and data security. All user data is protected by E2EE and is only stored on the user’s own device, with no centralized backup. User accounts are not linked to real user information, effectively preventing personal information leaks. You can also easily create your own channel, mini program and AI bot here.

Website: https://www.luffa.im/
Twitter: https://x.com/LuffaMessage
Telegram ：https://t.me/LuffaMessage

In the context of the continuous evolution of blockchain and decentralized technologies, the traditional storage model is undergoing a profound transformation. As a project aiming to build an efficient, secure, and auditable data storage system, Endless leverages a verifiable Proof of Storage mechanism based on cryptographic commitment and the Challenge-Response protocol to provide a promising technical solution for the Web3 ecosystem. This article offers a detailed explanation of Endless’s storage consensus model, striving to help cryptocurrency enthusiasts and blockchain industry practitioners better understand the project’s innovations in decentralized storage from both a technical and procedural perspective.

As blockchain technology continues to expand into various application scenarios, user demands for diverse data storage and heightened security are also on the rise. In traditional centralized storage architectures, users must trust third-party service providers, leaving data vulnerable to tampering, loss, and privacy leaks. The immutability of blockchain and the characteristics of peer-to-peer networks bring new possibilities to distributed storage systems. Through a storage resource-based consensus model built on the blockchain network, Endless allows each node to earn rewards by providing genuine storage services. This enables users to store data with greater peace of mind in an immutable and transparent environment.

Compared to a typical consensus mechanism, Endless focuses on “storage evidence” itself — namely, how to ensure that data is truly stored without fully trusting the storage nodes, and can be audited and verified anytime, anywhere. This further guarantees data retrievability and integrity.

From a user’s perspective, data sharding and encryption safeguard privacy. Additionally, a random challenge mechanism encourages nodes to continuously provide storage services, preventing data loss. From a developer’s perspective, only hashes and commitment information are written on-chain, saving project costs and network resources. Meanwhile, higher throughput can be achieved through sharding and distributed nodes in a decentralized storage network, ensuring stable performance for dApps. From a project standpoint, risks are dispersed to ensure transparency and compliance. Enterprises, too, can better control operating costs in a decentralized network, while benefiting from high stability and data redundancy offered by quality nodes to guarantee service quality.

Overview of the Storage Resource-Based Consensus Model

The storage resource-based consensus model adopted by Endless is driven by ensuring data integrity, reliability, and non-repudiation. In this mechanism, storage providers are not simply writing user data onto a hard drive; they must present verifiable proofs of storage, proving to the network or the user that they properly maintain the data at all times.

This model primarily hinges on the following key factors:

• Decentralized Storage Network and Cryptographic Measures: Under this model, each storage provider is a node within the network. Users employ encryption to deliver files securely to these nodes, greatly reducing the risk of information leaks or tampering.
• Cryptographic Commitment and Challenge-Response Protocol: This element forms the technical core of the entire model. It serves to periodically verify whether nodes still possess the complete data, preventing nodes from “gaming the system” or forging stored data to claim rewards.
• Auditability and Immutability: Each time the storage provider generates a storage proof, it is linked with the blockchain, so the historical record of the entire process cannot be easily altered. This guarantees transparent and fair auditing.
• Incentive Mechanism: Storage providers earn monetary rewards by honestly storing user data and submitting valid storage proofs as required, thus forming a healthy and sustainable ecosystem.

By incorporating these aspects, Endless ensures the authenticity and verifiability of data storage in a decentralized environment, providing the technical foundation for further expansion into additional fields.

User Case

Background: Xiao Wang is an ordinary blockchain enthusiast who wants to back up personal data, photos, and documents securely to a decentralized storage network. However, he lacks extensive understanding of the complex blockchain ecosystem and cumbersome key management processes. He worries about mishandling these processes or having his data stolen by untrustworthy nodes.

Pain Points: Xiao Wang wants functionality similar to traditional cloud drives for uploading and managing files, yet he refuses to tolerate centralized risks. He needs assurances that the data he uploads is encrypted and verifiable, safe from arbitrary tampering. Lacking trust in storage nodes, he is unsure whether nodes can store the data long-term or whether they might secretly delete it.

Solution: With the verifiable Proof of Storage mechanism adopted by Endless, Xiao Wang can first encrypt and shard his data prior to uploading it. Thus, even though the nodes cannot decrypt his data, they can still store it in full. Using a front-end interface or a mobile app provided by the project, Xiao Wang can upload data with one click. The system will automatically perform encryption, generate metadata, and record the digital signature on-chain. The network periodically initiates random challenges, requiring storage providers to submit proofs, ensuring that nodes continue to store the data honestly.

User Value:

• Privacy and Security: Sharding and encryption prevent any accidental data leakage, while the blockchain’s verifiability thwarts malicious tampering.

• Stability and Sustainability: The random challenge mechanism motivates nodes to continually provide storage services to avoid data loss.

• Simplified Operation: The front-end or mobile-based applications lower the technical barrier, allowing regular users to easily enjoy the security and trustworthiness of decentralized storage.

Core Process Analysis

Endless’s storage resource-based consensus model consists of six core steps. Below is a more detailed description from a technical and procedural perspective.

File Upload (Files to Provider)

First, the user splits and encrypts the files to be stored. During the sharding process, the file is broken into multiple small fragments, each of which can be independently identified and verified. Encryption ensures that even if these fragments are transmitted or stored on the network, they will not expose the file’s content. This approach enhances reliability and security during the data storage process.

After sharding and encryption, the user sends all fragments to the storage provider. Based on a decentralized storage architecture or the configuration of cloud storage providers, different fragments are distributed among various nodes. As a result, even if a particular node fails or acts maliciously, other nodes still guarantee the file’s integrity.

Metadata Generation (Metadatas to User)

Once the storage provider receives the file fragments, it must generate metadata for each file. Such metadata typically includes:

• Hash Value: A summary obtained by using a hashing algorithm (e.g., SHA-256) on the file or each fragment.

• Storage Location: Indicates the specific node or physical location where the fragment is stored.

• Timestamp: Records the time of file upload and fragment storage.

• Other Necessary Information, etc.

This metadata is returned to the user so that the user can monitor and audit, ensuring the storage provider cannot later deny its obligation to maintain the data.

Digital Signature Binding (Signatures)

Based on the metadata above, the storage provider will use a digital signature algorithm (commonly ECDSA) to sign the metadata. The purpose of the signature is to:

• Prove Data Ownership: It confirms that the file has been handed over to the designated storage node at a specific point in time.

• Guarantee Data Integrity: It prevents any third party or even the node itself from tampering with the stored fragment data in the future.

• Clarify Storage Responsibility: In the event of a data loss or malicious tampering, the signature can be used to trace and hold the storage provider accountable.

This digital signature binding mechanism also highlights the storage provider’s trustworthiness in the entire process. Once a signature is generated, any subsequent data tampering will be uncovered during future audits.

Cryptographic Commitment Generation (Commitment)

Besides simply signing the metadata, the storage provider must also generate a storage commitment using advanced cryptographic algorithms (e.g., KZG polynomial commitment). The essence of the commitment is that the storage provider “declares” to the blockchain network its commitment to continually hold and maintain a certain piece of data, while also recording a corresponding commitment value on the blockchain.

Thanks to the blockchain’s immutability, once the storage provider submits this commitment to the network, it can neither arbitrarily change nor deny the event. This has two significant implications:

• Auditability: Subsequent Challenge-Response processes can compare the commitment made earlier. If the commitment value does not match, it suggests the storage provider may not have been honest.

• Non-repudiation: After submitting the commitment, the storage provider must maintain this data throughout the entire storage contract period. Otherwise, it will fail subsequent challenges and might face penalties or lose rewards.

Challenge Mechanism (Challenge)

To ensure that the “storage commitment” does not remain mere words, the network or users can, on a random and periodic basis, issue challenges (Challenges) to the storage provider. During the challenge process, the provider is required to quickly verify the required file fragments (or the entire file) and submit a corresponding proof.

The challenge mechanism is valuable in two respects:

• Real-Time Constraint on Storage Providers: The randomness makes it impossible for any node to predict exactly when a challenge will occur, forcing them to maintain data integrity over most of the time.

• Guarding Against Data Loss or Malicious Deletion: If a storage provider fails to submit a correct proof during a challenge, this indicates dishonest behavior, leading to loss of economic incentives or even penalties.

Storage Proof Generation (Proof)

Upon receiving a challenge, the storage provider immediately calculates a storage proof (Proof) based on the data it holds and the prior commitment recorded on the blockchain. This proof is submitted to the blockchain for verification. By comparing the hash value, commitment value, and digital signature, the network can determine whether the storage provider still retains and maintains the complete data.

If verification succeeds, the storage provider continues to earn rewards; if it fails, it is deemed negligent or dishonest, resulting in loss of income or punitive actions. Consequently, a highly effective and viable trust and supervision framework for storage providers is formed throughout the network.

Technical Value

• Honesty and Credibility: By employing a verifiable Proof of Storage mechanism, Endless’s decentralized storage consensus framework removes the need for users to place excessive trust in any single storage provider. The network uses periodic challenges and rigorous verification to ensure data security and availability.
• Economic Incentives and Sustainability: Storage providers receive rewards for delivering genuine, reliable storage services. This incentive strategy effectively drives nodes to invest in storage resources, fostering a healthy and continuously growing ecosystem.
• Efficient Web3 Data Infrastructure: As the Web3 landscape grows, numerous applications require decentralized storage as a core underpinning. Endless’s model provides these applications with a comprehensive infrastructure, enabling decentralized applications (dApps) to enjoy secure and auditable data, coupled with high read/write efficiency.
• Auditability and Compliance: The open and immutable nature of the blockchain, combined with the commitments and proofs submitted by storage providers, offers an objective basis for regulators or audit bodies. This helps meet further regulatory and auditing requirements in diverse industries.

Developer Case

Background: Li Ming is a developer specializing in decentralized applications (dApps). He has been searching for a storage solution that balances efficiency and security for his DeFi project. Given the project’s large user base and high demands for data search and real-time performance, Li Ming needs a file storage system that is both fast in read/write operations and verifiable on-chain.

Pain Points:

• The DeFi project might involve large volumes of data. Storing everything on-chain hugely increases on-chain storage costs.

• It requires fast and low-latency read and write services to meet the demands of real-time applications.

• It must ensure auditability and verification of data writing, storage, and subsequent calls, in order to address potential user doubts or compliance checks.

Solution: Developers can use Endless’s storage SDK to encrypt and store large files or critical data on decentralized nodes. Only small hashes or commitment data need to be written on-chain. The SDK includes a built-in Challenge-Response protocol that interacts with the blockchain to periodically conduct random verifications of the storage nodes, while also providing a traceable log function to help developers monitor data calls and updates. By using advanced cryptographic commitments (such as KZG polynomial commitments), one can quickly validate data authenticity and integrity, reducing the performance overhead arising from redundant operations or excessive verification.

User Value:

• Reduced On-Chain Storage Costs: Only hashes and commitment information are recorded on-chain, saving project funds and network resources.

• Improved Read/Write Efficiency: A decentralized storage network can achieve high throughput through sharding and distributed nodes, ensuring stable performance for dApps.

• Compliance and Transparency: Verifiable Proof of Storage and on-chain records meet compliance and audit needs, enabling developers to confidently face regulatory requirements and user concerns.

Industry Case

Background: A large content distribution and cloud computing company wishes to integrate blockchain technology to provide decentralized storage solutions for its massive collection of documents (e.g., e-tickets, e-contracts, business records). The company still has many doubts about efficiently completing data migration and integration, and it seeks to strike a balance between economic incentives and operational cost controls.

Pain Points:

• In traditional centralized storage solutions, infrastructure costs keep climbing as data volume grows over time, and there is a lack of transparent, strict audit mechanisms.

• The company needs to share data across multiple departments and multiple business lines while strictly managing permissions for sensitive information.

• It worries about the stability of decentralized networks and the quality of node services, fearing potential data losses or access delays if nodes fail or go offline.

Solution: By leveraging Endless’s consensus model, the enterprise can shard vast documents, encrypt them, and distribute them among different nodes, then record the storage commitments on the blockchain. The enterprise can sign long-term storage contracts with providers and rely on the Challenge-Response protocol to monitor the storage status of each node. If a node does not fulfill its duties, it is easily identified and replaced after a challenge. When it comes to integration, the enterprise can connect its existing cloud architecture to the decentralized storage network using Endless’s APIs or specialized modules, while strengthening internal auditing and compliance reporting via digital signature binding.

User Value:

• Risk Dispersion: Since document shards are stored in different locations, even if one node fails, business continuity remains largely unaffected.

• Transparency and Compliance: The verifiable blockchain record allows regulators or auditing departments to trace document flows at any time, lowering compliance risks.

• Cost and Incentive Balance: By using a decentralized network, enterprises can better manage operating costs while enjoying high stability and data redundancy from quality nodes to ensure service quality.

Conclusion: Unlocking Value with Decentralized Storage

By innovatively applying cryptographic commitments and the Challenge-Response protocol, Endless has built a Proof of Storage solution fit for the blockchain era. Across file uploads, metadata generation, digital signature binding, commitment creation, challenge mechanisms, and storage proof submission, the framework reflects a strong emphasis on data integrity and auditability:

• Multiple verification measures for sharded and encrypted data, ensuring credibility in storage behaviors.
• Digital signatures and polynomial commitments record information on the blockchain, guaranteeing objectivity in subsequent audits.
• Random challenges prompt storage providers to maintain data continuously, preventing short-term speculation or cost-cutting measures that might harm user interests over the long term.

On these grounds, Endless’s consensus model offers a feasible and promising path in the decentralized storage domain, delivering a more stable, secure, and sustainable storage solution for the Web3 ecosystem.

For technology professionals concerned with the blockchain’s future, this model represents a pioneering trend in the convergence of next-generation storage and computing. In increasingly complex decentralized application scenarios, “storage proofs” can more solidly support the secure interaction of digital assets and data, allowing developers to delegate heavy data tasks to a dispersed community of nodes. As technology advances and more service providers join, Endless is poised to play an increasingly significant role in decentralized storage systems.

Looking forward, there will be more functions and optimization opportunities derived from the verifiable Proof of Storage mechanism — such as improving challenge efficiency, reducing storage node loads, and enhancing scalability. These will likely remain the core pursuits of Endless’s iterative developments. For the blockchain sector as a whole, these efforts not only promise users a better storage experience but also lay a solid foundation for the flourishing of decentralized applications.

London, UK — June 20, 2025 — A seminar titled Green AI: Can Emerging Technologies Save the Planet? , focusing on sustainability was successfully held at the University of Surrey, hosted by Endless Protocol and its decentralized social platform, Luffa. Endless Web3 Genesis Cloud, a pioneering Web3 infrastructure, and Luffa, built upon it, united to explore the future of sustainable technology.

The session featured Satya S. Tripathi, former United Nations Assistant Secretary-General and current Secretary-General of the Global Alliance for a Sustainable Planet (GASP), in a keynote that sparked urgent debate on the role of next-gen technologies in advancing climate action.

Chaired by Professor Yu Xiong, Co-founder and Co-President of Endless Protocol and Associate Vice-President of the University of Surrey, the session was moderated by Oliver Venables, Luffa’s Product Director. Joining the panel was Jay Shen, Founder of Transreport, a tech firm renowned for enhancing transport accessibility, fostering a rich dialogue spanning technology, governance, and societal impact.

Key Themes Explored:

• The Dual Nature of Tech: How AI, blockchain, and the metaverse serve as breakthroughs for sustainability while posing risks of new emissions and inequalities.
• Digital Climate Finance: Innovative applications of DeFi and tokenized carbon credits to bolster climate resilience and investment in underserved regions.
• Global Governance and Ethical Leadership: The need for systemic change and multilateral collaboration to steer technology toward shared planetary goals.

The forum underscored Luffa’s ambition to build the AI-social layer of Web3, positioning it as more than a social platform — a critical infrastructure for global sustainability. “Luffa, born from AI-native infrastructure, is about connecting users to a higher purpose,” said Oliver Venables. “Today’s dialogue highlighted the transformative potential of decentralized systems in advancing climate action, equity, and global collaboration.”

Endless Protocol has long been a champion of sustainable development. In November 2024, at COP 29 in Baku, Azerbaijan, Professor Xiong delivered a keynote at the “Empowering Sustainable Innovation” forum, showcasing how Endless leverages blockchain and AI to support net-zero goals. By tokenizing carbon credits and using blockchain for transparent tracking, Endless enhances carbon market efficiency. Its AI-driven tools enable precise energy management, real-time consumption monitoring, and predictive algorithms to minimize waste, significantly reducing the protocol’s carbon footprint.

Professor Xiong emphasized, “Endless is not just a technological innovation — it’s a commitment to a green future, offering a sustainable pathway to global net-zero targets.”

“We are honored to collaborate with thought leaders like Satya and institutions like the University of Surrey,” added Professor Xiong. “Endless provides a decentralized infrastructure for a fair, green future, while Luffa serves as a core platform for a purpose-driven, co-creation economy — empowering users, developers, and communities alike.”

This event marks a key milestone in Endless Protocol and Luffa’s mission to foster a co-creation economy, enabling collective participation, governance, and benefit.

About Luffa

Luffa is a decentralized, AI-native social platform built on distributed and encryption technologies, prioritizing user privacy and data security. All data is protected by end-to-end encryption (E2EE) and stored solely on users’ devices, with no centralized backups. Accounts are detached from real-world identities to prevent data leaks, while users can effortlessly create channels, mini-programs, and AI bots.

About Endless Web3 Genesis Cloud

Endless connects Web2 and Web3 ecosystems, offering developers a one-stop platform for building Web3 applications with Web2-level user experiences. Powered by the Move language-based public chain, Endless integrates advanced AI capabilities and plugins, positioning itself as the premier connector of AI and Web3. It enables modular, rapid development of encrypted AI applications, paving the way for super-intelligent AI agent systems.

In today’s booming blockchain industry, how to realize flexible management and well-reasoned distribution of tokens across various application scenarios remains a crucial topic for both project teams and communities. Whether a new project aims to incentivize more users through token rewards, or a mature project tries to further optimize its tokenomics, the locking and unlocking of tokens plays an indispensable role. As blockchain applications diversify, the need for a standardized solution to efficiently, transparently, and fairly regulate token circulation becomes increasingly urgent.

For users, tokens in the Endless ecosystem maintain price stability and ensure long-term returns. For developers, the out-of-the-box locking and unlocking functionality significantly reduces development time while avoiding potential security loopholes from reinventing the wheel. For project teams, token locking and unlocking helps maintain relative stability of token prices, bolstering investor confidence. Moreover, it enables rapid completion of audits or reconciliations, mitigating information asymmetry and operational risks.

This article introduces the latest token-locking standard introduced in the Endless system contract and its core smart contract, “locking_coin_ex.move.” Designed to allow project teams and users to lock and unlock tokens and perform related queries in a secure, controlled environment, this contract provides a framework for token release at different time intervals to ease the impact of concentrated release, stabilize market liquidity, and offer sound tokenomics tools for more blockchain projects.

Background and Significance

Typically, blockchain projects in early development reserve a certain percentage of tokens for team incentives, partner allocations, community operations, and more. If these tokens are all released onto the market in a short period, it can cause volatility in investor and user expectations, while also placing significant market pressure on the project. By leveraging “locking” and “unlocking” mechanisms, a project can meticulously plan token releases at the design stage, satisfying both team growth and community incentive needs while maintaining stable market supply, thereby hedging against the price swings resulting from excessive speculation or selling.

Therefore, introducing a token locking standard has several essential benefits:

• Helps maintain token price stability and mitigate market volatility
• Enables project teams to more prudently allocate tokens, ensuring long-term development
• Enhances community trust in the project and increases user engagement
• Provides a reliable functional component for constructing decentralized finance (DeFi) and other application ecosystems

Introduction to the Endless System Contract and the New Contract

Within the Endless system contracts, the team has freshly released the smart contract “locking_coin_ex.move” to manage token locking and distribution. Developed in Move and built on top of Endless’s core financial contracts, this contract performs a series of on-chain security checks to ensure that the relevant tokens are released in batches over predetermined time intervals. Put differently, it offers an automated and auditable “time lock” function, helping token issuers adjust the token release pace more flexibly while transparently disclosing the unlocking rules.

Throughout the process, all participants on the chain can easily track each address’s locked amounts, unlocking rules, and timelines via calls to the smart contract’s query interfaces or by viewing contract events. This type of “openness and transparency” is one of the greatest advantages of blockchain applications: even if users and project teams do not have sufficient offline trust, they can still cooperate under the contract’s security and autonomy, thus avoiding the financial management risks posed by centralization.

Main Functions

Based on the underlying technical principles and contract features, this contract primarily provides the following key capabilities:

• Token Locking: The contract allows administrators to lock tokens to a specific address and set an unlocking schedule. Locking means that within a designated time period, the locked tokens cannot move; they will only be released in stages once the contract’s unlocking rules or triggers come into effect. In scenarios such as project operations and community incentives, this feature can be used to lock tokens in bulk to specific addresses, enabling beneficiaries to receive token support over the medium to long term.
• Token Unlocking: Under a predetermined unlocking plan, the contract automatically releases a certain number of tokens at the end of each unlocking cycle. This is akin to configuring an “unlocking calendar” for tokens, allowing tokens allocated to different cycles to enter circulation in an orderly fashion while enhancing stability. For project teams, this not only allows precise management of the amounts allocated to teams or communities at different stages, but also prevents the market shock from a one-time massive release.
• Query Functionality: The Endless team has provided the contract with rich query interfaces so that ecosystem participants can stay informed about the status of locked and unlocked tokens in real time. For example, users can check the total locked quantity in the system, view information about all lockers (or those locked), or find details about the locked amount, unlocking schedule, and remaining locked tokens for a specific locker address. As a result, community members or investors do not have to rely on manual disclosures by the project team; they can self-check how tokens are being released and unlocked.
• Event Recording: During token unlocking and claiming, the contract logs relevant event information. Because blockchain lends itself well to audit and tracking processes, these event logs not only help the community build trust in the project, they also serve as publicly verifiable evidence in the event of compliance checks or disputes. Consequently, when the project team conducts financial audits, allocates tokens, or reconciles with other partners, they can swiftly locate and retrieve specific unlocking events, creating a robust framework for supervision and auditing.

User Case

Background: Xiao Wang is a retail investor enthusiastic about new blockchain projects and is also active in community engagements. He often finds that the tokens he purchases are prone to speculation, and excessive market volatility affects his long-term confidence in these projects. Moreover, while he hopes to gain more project incentives, he also doubts quick, arbitrary airdrops, fearing that mass sell-offs could cause price swings.

Pain Points:

• Rampant speculation and short-term volatility can generate extreme token price fluctuations
• Lack of transparent insight into the project team’s distribution plans, making it hard to track token release details in time
• Desires to balance long-term holding and short-term gains but worries about excessive selling leading to losses

Solution: The community in which Xiao Wang participates adopted the “locking_coin_ex.move” contract from Endless, employing a lock-up and step-by-step unlocking strategy for both airdrops and rewards in line with the project plan. The tokens Xiao Wang receives are released in multiple phases, ensuring that a large number of tokens do not flood the market at once. Community members can also view Xiao Wang’s and other users’ unlocking progress through query interfaces, keeping tabs on the transparency and timing of token releases.

User Value:

• Price Stability: Gradual unlocking and token lock-up control help avert severe market turbulence caused by concentrated sell-offs
• Increased Confidence: Xiao Wang can check his own unlocking schedule and contract details in real time, boosting trust
• Long-Term Returns: Phased token unlocking helps Xiao Wang balance short-term cash-outs and long-term holdings, thereby ensuring improved prospects for the project’s long-term potential

Core Design: Locking and Gradual Unlocking

The contract’s core design lies in the “locking” and “gradual unlocking” mechanism, enabling comprehensive management of token release speed. Concretely, when the administrator initiates a “lock” operation in the contract, the corresponding tokens are frozen under a specific module address within the contract until the unlocking plan takes effect and enters its respective unlocking phase.

This design offers the following advantages:

• Flexibility: The project team can configure custom lock-up rules, unlocking intervals (for instance, weekly, monthly, or quarterly), and amounts per release, meeting diverse business requirements.
• Security: Once locking takes effect, tokens can only be released when the contract’s preset unlocking conditions are met, preventing the administrator or other roles from arbitrarily altering token ownership midway.
• Transparency: All locking and unlocking activities are traceable on-chain. The community or investors can feel confident participating since no suspicious transaction can escape scrutiny from contract events and blockchain explorers.

By offering these three significant benefits, projects can effectively control the liquidity of tokens over the short term, sparing the market from roller-coaster fluctuations and, from a macro perspective, maintaining the stable operation of the entire project ecosystem.

Developer Case

Background: Li Ming is a programmer passionate about developing blockchain smart contracts and provides technical support for multiple decentralized applications (DApps). While writing contracts for a next-generation social application project, he needs a solution that helps the team manage token lock-ups, phased unlocking, and data traceability, to meet the project roadmap and community expectations.

Pain Points:

• Traditional token distribution methods are not easily customizable and lack comprehensive query interfaces
• Absence of standardized, reusable contract modules leads to extensive custom validation logic
• High security and stability demands that do not allow any smart contract vulnerabilities to derail the project’s progress

Solution: Li Ming studied the “locking_coin_ex.move” contract from the Endless project and found it to provide comprehensive locking and unlocking logic along with detailed event logging and query interfaces. He integrated the contract module directly into his project and configured unlocking strategies, operational permissions, and event listeners according to his specific business requirements. Thus, even if adjustments are needed after launch, it can be done through on-chain parameters without broadly rewriting the contract logic.

User Value:

• Rapid Development: Out-of-the-box lock-up and unlocking functionalities save substantial development time
• Extensibility: The smart contract’s open interfaces accommodate various business scenarios, making custom features easy to add
• Security Assurance: The contract itself has undergone extensive functional and security audits, avoiding vulnerabilities typically introduced by rewriting code from scratch

Industry Case

Background: A gaming company aims to partner with a decentralized platform to launch a blockchain game and plans to issue a platform token as the vehicle for value both in and outside the game. The company intends to bring in outside investors and gaming guilds but worries that releasing too many tokens initially might cause market volatility and undermine mutual trust among investors. Simultaneously, the company also worries that complicated approvals and reconciliation processes could slow down collaboration.

Pain Points:

• Industry partners require transparent and compliant handling of funds and token allocations to avoid regulatory risks
• All parties need mutual trust, but are concerned that an excessively large short-term token release could lead to sharp price fluctuations
• Offline approvals and reconciliations are cumbersome and prone to time lags and information asymmetry

Solution: The gaming company uses the Endless ecosystem’s “locking_coin_ex.move” to handle token allocation and lock-ups for its partnership network. A lock-in period is established in advance, ensuring that the core team’s and guilds’ token entitlements are released in batches within specified time frames. External investors and gaming guilds can utilize the contract’s query interfaces and event logs to check token allocation and unlocking progress at any time. All collaboration processes are recorded on-chain, greatly reducing the reconciliation overhead typical in traditional paper-based or centralized systems.

User Value:

• Mutual Trust Mechanism: Transparent and queryable smart contracts offer all partners reliable assurance
• Market Stability: Phased token unlocking helps maintain relative token price stability and instills investor confidence
• Convenient Auditing: On-chain event logs can be retrieved with a single click to obtain relevant data, swiftly completing reviews or reconciliations while mitigating information asymmetry and operational risks

Conclusion: Endless Ensures Ecological Stability through Token Locking and Unlocking

By integrating “locking_coin_ex.move” into the Endless system contracts, the project delivers a standardized, universal, fully traceable solution for token locking and unlocking. For project teams, this mechanism helps them design more flexible token distribution strategies in various blockchain application scenarios. For communities and users, such a locking and unlocking mechanism boosts trust in the project, enabling all parties to engage and share rewards under a single transparent set of rules.

As blockchain technology continues to evolve, this standard can also be applied to a much wider range of scenarios, such as the token release processes of decentralized autonomous organizations (DAOs) when managing community funds, or the phased distribution of collateral or mining rewards in more sophisticated DeFi protocols. In doing so, not only does it give the market clearer and more predictable insights into the state of token circulation, it also lays a solid foundation for the roll-out of more large-scale applications in the future.

If you have further questions or interest regarding this locking contract or the Endless system contracts, feel free to explore Endless’s technical documentation, community forums, or developer discussion groups. Under fair and transparent rules, the blockchain ecosystem is poised to become more professional, trustworthy, and efficient.

In the blockchain world, how to securely and flexibly manage accounts has always been a core concern for numerous applications and asset owners. From personal wallets to institution-level asset custody, and even the treasury management of decentralized autonomous organizations (DAOs), multi-signature (“multi-sig”) is an effective means to mitigate the risk of private key leaks or single points of failure. However, on current mainstream blockchain platforms (such as Aptos and Sui), on-chain multi-sig solutions still exhibit clear limitations in economic efficiency, interaction flow, and rigid account-type restrictions.

Endless addresses these pain points by deeply re-engineering its multi-signature mechanism. Leveraging a dynamic authentication key (auth_key) architecture substantially augments the flexibility and usability of multi-sig accounts, providing a more forward-looking account security solution for the Web3 ecosystem.

For Users: Endless multi-sig accounts are intuitive and easy to use, eliminating the need to deploy a brand-new multi-sig smart contract. They can be set up quickly while saving costs. Additionally, the risk of private key theft is significantly reduced, boosting confidence among less experienced users in controlling their digital assets.

For Developers: A decentralized permission management approach ensures that every team member has clearly defined responsibilities and rights, enhancing transparency and security. As a DApp iterates, multi-sig participants or threshold values can be swiftly added or removed to meet changing project needs.

For Enterprises: Multi-sig strategies can integrate with existing approval processes in their finance and compliance departments, inheriting internal risk-control principles. It also lowers operational expenses and makes audits easier for internal or third-party auditors to conduct post-transaction examinations or reviews.

Potential Application Scenarios for Multi-Sig Account Security

As blockchain applications evolve, the demand for multi-sig accounts is growing. Below are a few potential use cases:

• DeFi: In various decentralized finance applications, high-value fund movements often require approval from multiple parties. Multi-sig can serve as an insurance mechanism to reduce risk in liquidity pools. At the same time, flexible key configuration can more quickly address contract loopholes or emergencies, furnishing real-time security strategies for projects.
• DAO Governance: The core of a DAO is decentralized administration and democratic decision-making. Multi-sig capabilities give DAOs a convenient option for asset management and voting. By enabling multi-party signatures among several council members or management addresses, a DAO can achieve consensus and ensure that key decisions are implemented openly, transparently, and securely.
• Enterprise Blockchain Applications: Conventional businesses evaluating blockchain are especially cautious about asset custody and permission management. A multi-sig framework provides checks and balances across departments or job positions, so that critical transactions require the signatures of multiple responsible individuals — this aligns with corporate security and compliance needs.

By introducing innovative dynamic authentication keys at the protocol layer, Endless has set a new benchmark for account security management on blockchains. Going forward, Endless will continue refining its protocol-level identity management to meet rising demands for security, compliance, and user experience in multi-sig accounts, and will further promote multi-sig adoption in DeFi, DAO governance, enterprise blockchain solutions, and more.

Limitations of Traditional On-Chain Multi-Sig Solutions

In next-generation public chains like Aptos and Sui, multi-sig implementations generally depend on smart contracts. Taking Aptos as an example, when users create an account, they generate a 32-byte authentication key (auth_key) and employ a specific multi-sig contract module (such as 0x1::multisig_account) to administer the multi-signature process. While this approach partially satisfies the need for multiple-identity signatures, three main drawbacks remain:

1. Poor Economic Efficiency: Multi-sig transactions require frequent smart contract calls, incurring significantly higher gas fees than standard transactions. In scenarios requiring high-frequency operations, the cumulative impact of multi-sig transaction costs becomes increasingly burdensome, detracting from user experience and adding to economic strain.
2. Complex Interaction Flow: Under a contract-driven multi-sig model, typically multiple rounds of on-chain confirmations are necessary. After initiating a transaction, multiple signers must submit or authorize signatures through the contract, followed by final contract verification and execution of the transaction. These repeated steps greatly prolong the time to completion and increase complexity.
3. Rigid Account Types: Such multi-sig accounts are bound to fixed contract logic at the time of creation, tying accounts to a particular verification rule. Once deployed, single-signature accounts cannot easily be converted to multi-sig, and multi-sig accounts cannot revert to single-signature mode. This lack of flexible, dynamic adjustment is a core limitation.

These constraints pose challenges for meeting the security and efficiency requirements in diverse blockchain applications. Projects or individuals frequently grapple with higher gas costs and cumbersome confirmation procedures, and cannot smoothly switch account types if their needs evolve. Such a system struggles to fulfill the extended requirements of highly scalable, multi-use scenarios.

Endless Reconstructs the Multi-Sig Mechanism

To raise the security, flexibility, and cost-effectiveness of multi-sig accounts, Endless has chosen to rebuild the multi-sig mechanism directly at the protocol layer. The project’s dynamic authentication key architecture eliminates reliance on contract-driven multi-sig logic by embedding multi-sig transaction verification directly into the base protocol, achieving multiple technical breakthroughs.

In Endless, a single account’s auth_key is no longer limited to a single 32-byte validation key. Instead, it can be configured as a collection of 1 to 32 addresses:

• When the collection contains only one address, the account functions as a single-signature account.
• When the collection contains multiple addresses, the account automatically switches to multi-sig mode.

This architecture decouples account creation from binding with a multi-sig contract. Users can freely shift between single-sig or multi-sig mode under the same account without redeploying complex multi-sig contracts or managing extra contract accounts.

1. K-of-N Threshold Signature Strategy

To accommodate diverse permission requirements, the Endless dynamic authentication key supports standard K-of-N multi-signature. Users can set a “threshold” K and expand the auth_key’s address collection to up to 32 distinct key addresses. Only when at least K addresses collectively sign does the base protocol validate and execute the transaction. Such threshold signature logic benefits various organizations, roles, and cross-organizational collaborations. Using command-line tools or dedicated DApps, account holders can dynamically add or remove addresses, or adjust threshold values, to easily manage multi-sig logic.

User Case: A New Investor’s Digital Asset Security

Background: Xiao Bai (pseudonym) is new to blockchain, only recently starting to invest in crypto and use DeFi services. She is aware of multi-sig accounts but feels existing contract-driven solutions are “too complicated to deploy,” “come with excessive gas fees,” and “involve steps too difficult to grasp.”

Pain Points: She fears that a private key leak or accidental loss could wipe out all her digital assets. She also finds the learning curve high, since contract-based multi-sig demands a cumbersome creation process. It is hard to maximize both asset security and ease of operation; she must either sacrifice security with single-signature or pay high gas fees and spend considerable time deploying a multi-sig contract.

Solution: By leveraging Endless’s native multi-sig mechanism, Xiao Bai first creates a standard single-signature account with no extra gas cost compared to normal transactions. Via the Endless Multisig DApp, she then adds a “backup authority key” to her account, setting a 2-of-2 signature threshold. By storing keys on two separate devices (phone, hardware wallet), every transfer requires two signatures, substantially lowering the risk of a single compromised private key.

User Value:

• Straightforward Setup: No need to deploy a new multi-sig smart contract from scratch, quickly establishing an account.
• Cost Effective: The gas consumption for multi-sig transactions is almost the same as for single-sig, saving money for investors.
• Upgraded Security: The likelihood of private key theft leading to catastrophic asset loss is much lower, giving Xiao Bai greater control and confidence in her digital holdings.

Enterprise Case: Institutional Asset Custody and Auditing

Background: A traditional financial institution (such as an asset management firm or crypto exchange) is exploring using the Endless blockchain to issue and hold digital assets. While it has robust risk control and compliance systems in traditional finance, transitioning to blockchain means new technical and security risks. It typically has multiple departments (finance, legal, tech support, compliance) that must jointly manage on-chain transactions and asset flows.

Pain Points:

• Complex Authorization Processes: In traditional finance, all fund operations undergo multiple levels of approvals. Mapping this compliance mindset onto a blockchain multi-sig framework is tricky.
• Cost and Performance: Common contract-based multi-sig solutions can result in high gas expenses and complex on-chain calls, clashing with the company’s desire for a lean process.
• Operational Security: Large-scale assets need a higher threshold (e.g., 5-of-8 or 6-of-10) plus oversight to prevent internal fraud. It is unclear if typical multi-sig contracts can meet enterprise requirements. • Solution: By harnessing Endless’s native multi-sig features, an enterprise-grade account can be set up with multiple addresses, designating signatory rights for department heads or automated compliance bots. Internally, the company can assign sign-off authority to finance, legal, and compliance staff, requiring, say, “at least 3-of-5” signatures to execute large-value transactions. If staff changes or responsibilities shift, keys or thresholds can be dynamically reconfigured at the protocol layer without rewriting or redeploying new accounts.

User Value:

• Integration with Traditional Approval Processes: The multi-sig framework aligns with established corporate risk-control principles.
• Reduced Operational Costs: Since multi-sig validation is embedded at the protocol level, excessive repeated signatures and contract calls are unnecessary, lowering the company’s blockchain overhead.
• Audit-Friendliness: Multi-party signatures are clearly recorded on-chain, aiding internal or third-party auditing of each transaction and enhancing compliance and transparency.

2. Instant Switching Between Single-Sig and Multi-Sig Modes

A chief innovation of Endless lies in its ability to seamlessly shift from single-sig to multi-sig at the account level. If the user initially only needs a personal signature, they can set the address collection to just one address. Whenever they require stronger security or multiple-party management, they can add more keys and update the signature threshold. Conversely, if the account was configured for multi-sig but later wants to downgrade to single-sig due to different operational needs, removing additional addresses makes reverting quick and painless.

Developer Case: Multi-Party Collaboration in Contract Management

Background: Alice is the technical lead in a decentralized application (DApp) development team, planning to deploy new features on the Endless blockchain. The team includes front-end developers, back-end developers, and security auditors who need to collectively manage contract permissions and a funds pool.

Pain Points:

• Centralized Contract Authority: If only one person manages the private key, a hack or internal mismanagement puts all contracts and funds at risk.
• Complex Upgrade Process: Traditional multi-sig solutions require additional contract deployments and repeated contract calls. Team members must confirm multiple times on-chain, increasing uncertainties during iterative development.
• Limited Flexibility in Fund Management: The development team may occasionally need to allocate or freeze funds, but threshold signatures on typical multi-sig contracts are mostly rigid. Once deployed, quick adjustments are cumbersome. • Solution: On Endless, the team configures the project’s account with multiple core team addresses, applying a 3-of-5 multi-sig threshold through the dynamic authentication key approach. They use dedicated DApp tools to dynamically manage each address’s permissions: for example, removing an address if a member leaves, or adding a new partner’s address. Because multi-sig transaction validation is built into the Endless protocol, the team avoids frequent external contract calls, greatly lowering gas and cutting out multiple phases of interactions.

User Value:

• Decentralized Permission Management: Each member has clear duties and rights regarding contract management, building transparency and security.
• Flexible Iteration: As the DApp evolves, the team can swiftly add or remove signers, or adjust threshold values to meet new project requirements.
• Cost Savings: With native multi-sig at the protocol layer, the development team can stay focused on the product itself without spending excessive time or funds on deploying and maintaining external multi-sig contracts.

Three Core Advantages of Endless Native Multi-Sig

Compared to contract-driven multi-sig approaches, Endless’s native protocol support delivers these significant benefits:

1. Reduced Gas Fees: Because the multi-sig signature verification is embedded directly in the base protocol, multi-sig transactions no longer require extra smart contract calls, making gas costs for multi-sig almost on par with single-sig. For individuals or institutions frequently using multi-sig, these savings can be substantial, cutting expenses without sacrificing security.
2. Improved User Interaction: Endless provides a user-friendly, visual interface via specialized DApps (e.g., Endless Multisig DApp). Creating a multi-sig account, adding or removing key addresses, and modifying thresholds can be done in just a few clicks. Repeated executions within smart contracts are no longer necessary, drastically reducing steps and enhancing operational efficiency.
3. Enhanced Security: For accounts with frequent value transfers, multi-sig has become critical for mitigating the risk of private key theft and strengthening asset safety. By distributing authorization among multiple addresses in Endless’s dynamic authentication key model, the risk of single-point takeover is minimized, as a transaction must meet the threshold signature requirement before execution. Even if one address is compromised, the attacker cannot independently transfer assets, thus greatly reducing possible attack vectors.

Conclusion: Endless Multi-Sig Balances Security and Convenience

Amid the expansion of the Web3 ecosystem, multi-sig accounts are becoming ever more vital — not just for safeguarding assets but also for facilitating multi-party collaboration, from Ethereum to other emerging public chains. With ongoing technical updates and the onboarding of more developers and organizations, Endless’s native multi-sig approach is expected to inspire or be integrated by additional projects, fueling decentralization and improving user security.

Endless’s native multi-sig account security system, centered on dynamic authentication keys, directly resolves the efficiency and flexibility shortcomings of traditional contract-based multi-sig. Without compromising security, it cuts multi-sig transaction fees and refines the user experience, enabling both developers and users to seamlessly switch between single-sig and multi-sig accounts. This not only perfects multi-sig usage in blockchain scenarios but also offers a new line of thinking for the industry at large.

In a rapidly expanding Web3 domain with increasingly diverse decentralized applications, Endless’s technology demonstrates the far-reaching implications of protocol-layer innovation for blockchain projects: deeper infrastructure optimizations build a more robust, efficient, and secure environment for higher-level applications. Looking ahead, as multi-sig usage grows across varied contexts — commercial organizations, private users, and DAO communities — Endless’s native solution will safeguard assets and governance, establishing a solid foundation for a more open, collaborative, and transparent blockchain ecosystem.

Author: Dr. Amit Kumar Jaiswal, Endless

Model Context Protocol (MCP) is an open-source communication framework introduced by Anthropic, which is designed to standardize interactions between AI models, applications, and decentralized systems. It serves as a middleware layer and functions as a “universal adapter” that enables seamless data exchange, persistent connections, context preservation, and interoperability across heterogeneous AI agents, APIs, and blockchain networks, moving away from the traditional one-off API call approach.

Why MCP?

Existing LLM-driven systems have static input/output structures that make real-time collaboration and dynamic information use difficult. They are inefficient because they have to pass context every time, and integration with various tools and data sources is complex and non-standard. MCP solves these problems, allowing LLM to act as a conversational agent that accesses real-time information, actively calls external tools, and maintains context.

Key Features of MCP

● Standardized Communication: MCP defines a structured protocol for AI models to exchange contextual data without losing semantic meaning.

● Decentralized Interoperability: Facilitates interactions between AI agents in distributed environments, including Web3 and blockchain ecosystems.

● Context Preservation: Maintains continuity in multi-turn conversations and complex workflows involving multiple AI models.

● Security: Authority control, authentication, and data integrity

MCP is particularly relevant in scenarios where AI models need to collaborate dynamically such as in autonomous agents, decentralized AI marketplaces, and cross-chain smart contracts.

Core Components of MCP

MCP is based on a client-server architecture enabling seamless communication between AI models and external data sources. Its main components are:

1. Local Machine

The local environment where the MCP system operates (user’s device or server).

● Role:

○ Hosts the MCP Client and Host Application.

○ Executes AI workflows by coordinating between local and remote MCP Servers.

2. Host

The central orchestrator managing MCP Clients and AI integrations.

● Functions:

○ MCP Client Management:

• Spawns and terminates MCP Clients (e.g., one per MCP Server).
• Enforces authentication and session policies.

○ Context Aggregation:

• Combines outputs from multiple MCP Clients (e.g., merging LLM responses with database queries).

○ Security & Compliance:

• Applies user consent rules (e.g., GDPR) and data access controls.

○ AI Coordination:

• Routes requests to appropriate AI models (e.g., Claude, GPT-4) via MCP Clients.

● Examples:

○ Claude chatbot app, VSCode AI plugin, or a decentralized AI agent platform.

3. MCP Clients

Lightweight processes created by the Host to interface with individual MCP Servers.

● Functions:

○ Stateful Sessions: Maintains persistent connections to MCP Servers (e.g., WebSocket).

○ Protocol Negotiation:

• Handles versioning, encryption (TLS), and data serialization (JSON/Protobuf).

○ Message Routing:

• Bidirectional communication between Host and MCP Servers.

○ Pub/Sub Management:

• Subscribes to event-driven updates (e.g., real-time AI inference results).

● Key Features:

○ One client per MCP Server for isolation.

○ Implements failover mechanisms for reliability.

4. MCP Servers

Specialized servers exposing AI models, tools, or data sources.

● Roles:

○ Resource Exposure:

• Provides access to LLMs (e.g., GPT-4), databases, or APIs (e.g., weather data).

○ Decentralized Execution:

• Runs independently with focused responsibilities (e.g., “text summarization only”).

○ Sampling & Inference:

• Processes MCP Client requests (e.g., generates AI responses).

○ Security:

• Validates requests against access policies (e.g., rate limiting).

○ Message Broker:

• Uses pub/sub mechanisms (e.g., MQTT, Redis) for real-time AI agent communication.

○ Blockchain Integration Layer:

• Smart contracts can trigger MCP requests and AI responses can be recorded on-chain.

● Types:

○ Local Servers: Run on the same machine (e.g., Local Data Source A).

○ Remote Servers: Cloud-based or blockchain-linked (e.g., External Service C).

5. MCP Protocol

The standardized communication layer between components.

● Technical Stack:

○ Transport Layer: Uses HTTP/2, WebSockets, or libp2p (for P2P networks).

○ Security: End-to-end encryption (e.g., TLS, Noise Protocol).

6. Data Flow Example

1. User Request: A query is sent via the Host (e.g., “Fetch news and summarize”).
2. MCP Client Routing:

○ Host spawns two MCP Clients:

• Client 1: Connects to MCP Server A (news API).
• Client 2: Connects to MCP Server B (LLM summarization).

3. Execution:

○ MCP Server A retrieves news data → Sent to MCP Server B via the Host.

○ MCP Server B generates a summary → Returned to the user.

4. Context Preservation:

○ The context_id ensures follow-up queries (e.g., “Explain the 2nd point”) retain history.

MCP vs. Traditional APIs

MCP distinguishes itself from traditional application programming interfaces primarily by offering a persistent connection session rather than discrete, single request-response interactions. Unlike APIs, which require context to be transmitted with every request, MCP maintains and updates context in real-time. This approach enhances scalability as it uses a single standard protocol, alleviating the need for separate connections for each tool, a common requirement in traditional API systems.

MCP is a foundational technology enabling LLMs to transition from mere text generators to real-time interactive systems. It facilitates direct collaboration between AI and various external data sources, software tools, and user interfaces, thereby allowing for the execution of more complex tasks. MCP is envisioned to become the foundational protocol for AI agent ecosystems and the standard communication protocol for “Tool-Using AI”, paving the way for more sophisticated and integrated AI applications.

Integration with LLMs & AI Agents

● Dynamic Model Switching: MCP allows swapping LLMs (e.g., GPT-4 → Claude 3) mid-conversation without breaking context.

● Multi-Agent Workflows: AI agents can collaborate via MCP (e.g., one handles image generation, another processes text).

● Fine-Grained Access Control: MCP can manage permissions for AI model usage in decentralized networks.

MCP in Blockchain & Web3

MCP enhances decentralized AI applications by acting as a bridge between on-chain logic and off-chain AI processing.

Use Cases:

1. Decentralized AI Oracles

○ Example: A smart contract needs sentiment analysis of Twitter data before executing a trade.

○ MCP Workflow:

• Smart contract sends a request to an MCP node.
• MCP routes the query to an NLP model (e.g., GPT-4).
• The AI response is formatted into MCP and returned on-chain.

2. AI-Powered DAOs

○ Example: A DAO uses an AI agent to analyze governance proposals.

○ MCP Workflow:

• Proposal text is sent via MCP to an LLM for summarization.
• The summarized output is voted on by DAO members.

3. DeFi + AI Risk Assessment

○ Example: A lending protocol uses MCP to fetch AI-generated credit scores.

○ Technical Implementation:

• MCP fetches on-chain transaction history + off-chain credit data.
• An AI model processes this and returns a risk score via MCP.

Enabling Decentralized AI Agent Coordination in Endless AI’s Ecosystem

Endless AI utilizes the MCP to provide a unified and standardized approach for integrating AI within the Endless ecosystem. By combining MCP with Endless’s privacy-preserving mechanisms, such as Zero-Knowledge Proofs (ZKPs) and KZG commitments the protocol ensures that interactions where AI agents access and leverage external context remain both verifiable and compliant with privacy standards. This means that while AI agents can utilize external data to enhance their performance, the entire process is auditable on-chain, and user privacy is safeguarded through cryptographic proofs and data isolation.

● AI Models as Microservices: Each model is an MCP node, allowing seamless composition (e.g., GPT-4 + Stable Diffusion).

● Blockchain-Based Monetization:

○ Users pay in crypto to access AI services via MCP.

○ Smart contracts audit MCP interactions for fairness.

● Cross-Platform AI Agents:

○ Endless AI’s agents use MCP to interact with both Web2 and Web3 apps.

Luffa (SocialFi Platform):

○ MCP connects AI curators to personalize feeds while preserving privacy via zero-knowledge proofs.

AI-NFT Minting:

○ Creators use MCP to verify originality before minting NFTs.

DeFi Agent Swarms:

○ Autonomous traders use MCP to analyze cross-chain liquidity in real time.

MCP standardizes context-aware interactions, ensuring AI agents retain memory, adapt to real-time data, and collaborate across domains.

A. Agentic AI & Autonomous AI Agents

● Context-Aware AI Swarms:

○ MCP allows AI agents to share context (e.g., user preferences, transaction history) across sessions.

○ Example: A DeFi trading agent uses MCP to remember a user’s risk tolerance and adjust strategies dynamically.

● Multi-Agent Collaboration:

○ Agents negotiate via MCP-formatted messages (e.g., one agent handles NLP, another executes trades).

○ Use Case: AI moderators in Luffa (Endless’s SocialFi platform) collaborate to detect harmful content.

B. DeFi & Smart Contracts

● AI-Powered Oracles:

○ MCP fetches off-chain data (e.g., market sentiment) for smart contract triggers.

○ Example: A lending protocol uses MCP to pull AI-generated credit scores.

● Autonomous DeFi Agents:

○ Agents interact with liquidity pools via MCP, optimizing yields based on real-time analytics.

C. Large Language Models & AIGC

● Dynamic Model Switching:

○ MCP routes queries to the best-suited LLM (e.g., GPT-4 for text, Claude for code).

○ Use Case: An artist in Endless’s NFT marketplace generates AI art via MCP-connected Stable Diffusion.

● On-Chain IP Attribution:

○ MCP timestamps AI-generated content, minting it as royalty-bearing NFTs.

D. Decentralized AI Training (Proof-of-Useful-Work (PoUW))

● Federated Learning via MCP:

○ AI models train on user-owned data without central collection (privacy-preserving).

○ MCP aggregates model updates from distributed nodes.

Scenario: Endless AI’s MCP-Powered Chatbot

1. User query → MCP router → Best-suited LLM (based on cost/speed).
2. Response is formatted via MCP and returned to the user.
3. Payment is settled on-chain via crypto.

MCP is a game-changer for AI interoperability, especially in blockchain and Web3 ecosystems. By standardizing AI communication, enabling decentralized workflows, and enhancing LLM collaboration, MCP lays the foundation for the next generation of autonomous, composable AI systems.

Endless AI plans to expand MCP for:

● AI-to-AI Smart Contracts: Agents autonomously negotiate deals (e.g., licensing AI art).

● Quantum-Resistant MCP: Secure against future cryptographic threats.

● Self-Improving AI Ecosystems: MCP will enable AI swarms to self-optimize via decentralized learning.

London, June 2, 2025 — Endless Web3 Genesis Cloud (Endless) aims to hasten the widespread adoption of decentralized artificial intelligence (AI) by integrating Stability AI’s cutting-edge image generation capabilities with Endless’ robust Web3 infrastructure.

Supported by academic resources from the University of Surrey, Endless is set to drive innovation in sectors such as creative arts and finance, paving the way for a decentralized intelligent society.

Ignite the Potential of Web3 and AI Integration

Endless is a decentralized intelligent protocol designed to bridge Web2 and Web3 ecosystems, offering developers a one-stop platform for building Web3 applications with a Web2-level user experience. Built on the Move programming language, Endless’ public blockchain integrates numerous AI functionalities and plugins, positioning itself as the optimal connector between AI and Web3. Its modular framework enables developers to efficiently create Web3 AI applications, facilitating the rise of hyper-intelligent AI agent systems.

Leveraging Stable Diffusion 3.5, Stability AI’s most advanced image model, and a custom Sketch-to-Image workflow developed for Endless, we aim to lower barriers to AI application development, accelerating the real-world deployment of decentralized AI solutions.

Initial Key Collaboration:

Endless leveraged Stability AI’s research expertise to develop a custom Sketch-to-Image workflow to enhance the depth and differentiation of its product offering. Initially launching on the Luffa application, this workflow allows Luffa’s 200,000 strong userbase to easily create engaging and fun visual content to enhance their communication experience. Users, including professionals and hobbyists, can quickly draw a simple sketch and transform it into a compelling image, sticker or the like to send to friends.

Practical Applications: Empowering Creativity and Asset Ownership

Utilizing Stability AI’s technology will initially target high-demand, pain-point-intensive use cases, showcasing the enhanced user experiences enabled by technical synergies. Key applications include:

● Streamlined Content Creation: By integrating Stability AI’s “sketch-to-image” functionality with Endless’ on-chain AI infrastructure, users can transform simple sketches into high-quality, stylistically consistent images. This lowers the creative barrier, encouraging broader participation while boosting efficiency for professional creators through AI-driven style learning and automated complex draft generation.

● On-Chain Assetization and Trading: Creators can leverage Stability AI’s AI-generated content (AIGC) tools to produce works efficiently, then use Endless’ to mint these creations as on-chain assets, such as non-fungible tokens (NFTs), ensuring verifiable ownership and seamless trading. Endless’ community-driven token incentives further motivate creators to contribute high-value content to the platform.

Scott Trowbridge, Stability AI’s Vice President of Business Development & Partnerships said: “It’s exciting to see Stability AI’s powerful generative media tools enabling developers on the Endless platform to scale creative expression across the Web3 landscape. We’re eager to see the innovative applications Endless’ ecosystem unlocks with this cutting-edge technology.”

Yu Xiong, Co-President of Endless and Associate Vice-President at the University of Surrey, as well as Director of its Academy for Blockchain and Metaverse Applications, added: “This collaboration marks a pivotal step toward Endless’ vision of a decentralized intelligent society. By combining cutting-edge AI with Web3’s trustless infrastructure, we are poised to deliver transformative innovation in finance, creative arts, and beyond.”

About Endless Web3 Genesis Cloud

Endless is a decentralized intelligent protocol connecting Web2 and Web3 ecosystems, providing developers with a streamlined platform for building secure, efficient Web3 applications while ensuring user privacy, asset security, and data autonomy. Its integrated AI capabilities enable developers to seamlessly incorporate AI into dApps or create on-chain AI agents, solidifying its role as a bridge between AI and Web3.

Website: https://www.endless.link/

Twitter: https://x.com/EndlessProtocol

Telegram: https://t.me/EndlessProtocol

Media Inquiries

Endless Foundation

Email: Contact@endless.link

Contact: Victor Lau

Author: Dr. Tao Tang, Endless AI Researcher

1. Overview of Small-Parameter Language Models

On April 29, 2025, Alibaba open-sourced the Qwen3 series of language models and simultaneously released the model weights on GitHub and Hugging Face. The smallest model in the series has only 0.6 billion parameters and features a switch to toggle reasoning at the interface level. This “single-weight, dual-mode” design reduces deployment thresholds for terminal devices and has drawn interest from the edge and privacy computing communities for localized inference solutions.

In the chain-edge collaborative framework of the Endless Web3 Genesis Cloud (referred to as the Endless protocol or Endless), small models are positioned as key components for intent recognition and tool routing on the terminal side. Therefore, Qwen3–0.6B’s compact size and switchable reasoning mechanism are inherently compatible with the Endless tech stack, offering a viable option for lightweight deployment on PoUW (Proof-of-Useful-Work) nodes and user devices.

This paper explores the technical convergence of the two, discussing the potential for deep integration and coordination between compact models and the Endless decentralized intelligence network, while identifying promising application scenarios.

Overall, the author argues that due to the low computational requirements of small models, they can perform local inference on the terminal side. This aligns with the privacy protection demands of decentralized environments and allows proof of work to be uploaded via cryptographic hashes without exposing private or sensitive data. These characteristics present opportunities for meaningful collaboration with decentralized networks and Web3 systems.

2. A Closer Look at Qwen3–0.6B: Balancing Compactness and Depth

Beyond obvious metrics such as parameter count, number of layers, and context window size, Qwen3–0.6B introduces several noteworthy changes in usage.

Firstly, it enables switching between “reasoning mode” and “conversation mode” through a simple toggle: when reasoning is enabled, the model first outputs a thought process tagged with , followed by a conclusion; when disabled, it directly returns an answer. Tests show this switch adds only a few hundred milliseconds of latency on common edge devices, yet significantly improves performance on mathematical tasks and code generation, while maintaining near real-time interaction speeds in conversation mode.

Secondly, the 0.6B version extends the maximum context window to 32k tokens, enabling it to process entire technical documents or lengthy contracts in one pass. This reduces the engineering cost of context slicing and cloud callbacks in scenarios such as compliance audits and cross-chain transaction tracing.

Regarding inference parameters, the temperature and top-p settings provided by the official release have been thoroughly validated across community frameworks like vLLM, Ollama, and Llama.cpp. These settings mitigate common issues such as hallucinations or infinite loops, thereby lowering deployment and fine-tuning complexity.

In summary, Qwen3–0.6B addresses the industry’s key question: “How small can a model be while still being practical?” It maintains terminal-deployable compute requirements while delivering inference capabilities comparable to mid-sized models, occupying a balanced position just above the intelligence threshold without hitting hardware limits.

3. Endless: Making On-Chain Intelligence Verifiable

Endless AI aims to place the reasoning and training of generative AI within a verifiable, incentivized, and data-sovereignty-friendly Web3 runtime. The Endless AI white paper breaks the architecture down into four interlinked layers.

The foundational AI infrastructure layer includes a modified AI-Optimized Move VM, globally distributed GPU/TPU nodes, and a cross-chain AI oracle. Results are verified using cryptographic tools such as BLS12–381 signatures, while external data is securely injected on-chain via cross-chain bridges.

Above this is the framework layer, where the Agentic AI framework provides unified model invocation and permission management. Its core, the Model Context Protocol (MCP), standardizes access to smart contracts, external APIs, and off-chain tools, allowing agents to interact with real-time market data, perform DeFi operations, or generate content without exposing private keys.

The framework also supports multi-agent collaboration: autonomous agents can reach consensus through on-chain messaging to perform complex tasks such as portfolio rebalancing and content moderation.

The third layer is the data and privacy layer. Endless employs zero-knowledge proofs (ZKPs) to verify inference correctness, KZG polynomial commitments to ensure the integrity of training data, and decentralized storage to preserve encrypted data replicas. This enables models to access required context without leaking original sensitive data.

The outermost incentive layer uses the EDS token and optional NUSD stablecoin to pay for inference costs, distribute revenues, and maintain node integrity through staking and penalties. The Proof-of-Useful-Work (PoUW) consensus recognizes valuable AI tasks like gradient computation and output validation as accounting work. Nodes earn EDS based on contributions and are penalized for errors or malicious output.

In this layered structure, Qwen3–0.6B can serve as a lightweight inference core. With its compact size, it can reside on end-user devices for low-latency local inference or participate in decentralized training and verification on PoUW nodes. MCP provides a standard interface for accessing external tools and contracts, on-chain logs record every call and contribution, and the incentive layer rewards or penalizes nodes based on output quality. Thus, model behavior, data provenance, and economic incentives are tightly linked and verifiable on-chain.

4. Three Typical Convergence Scenarios

Endless AI outlines several practical use cases demonstrating how small models complement the decentralized framework on the product side. Three such examples highlight Qwen3–0.6B’s roles in terminal inference, on-chain verification, and decentralized training.

Luffa SocialFi: Local Generation, On-Chain Ownership

On the decentralized social platform Luffa, creators may first use small-parameter models locally to generate summaries, sentiment labels, and copyright risk alerts for their posts. After drafting, users can package the content and metadata through MCP and mint it as NFTs on-chain along with ZK proofs. This process secures content originality, timestamps, and ownership while preserving terminal-side privacy. Heavier tasks such as formatting or cross-chain retrieval are handled by larger cloud models. According to the Endless AI white paper, Luffa’s “terminal inference + on-chain notarization” model provides a full loop from content generation to profit sharing.

DeFi Agent Swarm: Market Recognition and On-Chain Auditing

In financial subnetworks, Endless plans to deploy AI agent clusters for trading wallets. Qwen3–0.6B runs on terminal devices to monitor DEX depth and price movements in real-time. When potential arbitrage is detected, it triggers multi-step scripts executed by 8–30B models and Move smart contracts. After trades, the small model encodes transaction summaries into ZKPs for blockchain upload, enabling third-party auditability without exposing strategies. The white paper refers to this mechanism as “Decentralized Finance Swarms,” core to improving on-chain financial transparency and responsiveness.

PoUW Fine-Tuning: Decentralized Model Evolution

In the training layer, PoUW treats small models like Qwen3–0.6B as the minimum distribution unit. Participating nodes download official weights and apply local fine-tuning (e.g., via LoRA) on private data. Instead of uploading raw data, nodes submit gradient hashes and KZG commitments to earn EDS through on-chain contracts. This design enhances model knowledge via terminal data while preserving data sovereignty, with weekly OTA updates sustaining long-tail capabilities. The white paper highlights this privacy-preserving federated learning as key to sustainable model improvement.

Across these scenarios, small models serve as general-purpose intelligent cores in the Endless ecosystem. They enable local real-time inference, integrate with MCP tools, leave verifiable on-chain records, and evolve continuously through community training and data contributions under PoUW. This role bridges users, on-chain protocols, and decentralized training networks through traceable, incentivized collaboration.

5. Looking Forward: From Component Markets to Agent DAO

The Endless AI roadmap spans approximately 24 months across three phases, with longer-term goals outlined.

Phase one involves launching a decentralized AI component marketplace, where developers can package pre-trained or fine-tuned models (e.g., Qwen3–0.6B) into callable smart contract components. Ownership and licensing are recorded via NFTs. Payment occurs through smart contracts using EDS or NUSD, with micropayments distributed automatically. The marketplace emphasizes standardized SDKs and MCP interfaces, enabling plug-and-play integration across dApps.

Phase two focuses on deploying a decentralized training mainnet. PoUW consensus becomes fully operational, and nodes earn EDS by contributing compute to training or gradient verification. Training supports federated learning and secure aggregation, requiring only hash and commitment uploads, not raw data. This “community gradient governance” allows faster model iteration and broader data coverage without centralized teams.

Phase three envisions an Agent Swarm DAO: lightweight models handle local sensing and decisions, mid-sized models process logic, and MoE-scale models manage long-horizon reasoning. These agents coordinate on-chain using consensus tools like wVRF, invoking external tools or contracts as needed. They are expected to drive automation and governance across DeFi, gaming, and content platforms.

In the long term, two goals are prioritized: AI-to-AI Contracts, where agents directly negotiate and execute contracts across chains and protocols; and Post-Quantum Cryptography Migration, upgrading signatures, key exchanges, and ZKP primitives to quantum-resistant algorithms to ensure network longevity and data integrity.

In sum, the component marketplace lays the foundation for model monetization, the PoUW mainnet fuels continuous model evolution, and the Agent DAO and quantum security upgrades point toward a self-governing, future-oriented decentralized intelligence network.

6. Conclusion: From Lightweight Models to Verifiable Decentralized Intelligence

The emergence of ultra-small models like Qwen3–0.6B shows that practicality is not solely a function of parameter size. More important is whether a model can deliver inference on edge devices with acceptable latency, benefiting a wider range of users and use cases. At the same time, Endless provides a verifiable on-chain runtime tailored to such widely accessible models: MCP unifies access to external tools and contracts, PoUW converts training and verification into economically accountable work, and the EDS token system automatically settles incentives based on result quality.

From a technical standpoint, local inference complements decentralized incentive-verification mechanisms. The former reduces adoption barriers; the latter ensures continuous improvement and data sovereignty via traceable incentives. Under this framework, users gain low-latency intelligent services and can also contribute compute, data, or fine-tuning efforts to the network and share in EDS-driven economic returns. This “lightweight inference + on-chain incentive” pairing offers a feasible path for deploying AI in decentralized settings, paving the way for multi-agent collaboration, component marketplaces, and future-proof security upgrades.