From the Hong Kong Consensus Conference to the Current Potential Projects and Directions of AI x Web3
TechFlow Selected TechFlow Selected
From the Hong Kong Consensus Conference to the Current Potential Projects and Directions of AI x Web3
Whether in the main venue or satellite sessions, discussions about AI x Web3 were ubiquitous.
Navigating Web3 tides with focused insightsAuthor: Anci, Core Contributor at Biteye
AI and web3 are widely recognized as the two driving forces propelling humanity into the next phase of technological growth. Following ChatGPT's disruptive AI experience, on-chain AI has quietly evolved from meme to infrastructure, emerging as the most promising and sustainable breakout sector in the eyes of the web3 community.
At Consensus Hong Kong 2025, the convergence of AI and web3 was a hot topic—discussions about AI x web3 echoed throughout both main and side events. The Biteye team dove deep into the action, uncovering numerous high-potential AI projects, and brings you these frontline insights.
1. AI Infrastructure
1. AI Agent Launch Platforms and Frameworks
Following the early-year surge of $AI16Z and $VIRTUAL, platforms and frameworks for launching AI Agents have seen explosive growth over the past six months. These projects provide low-barrier platforms enabling developers and everyday users alike to own and use AI Agents, marking one of the key focal points of this current wave of AI initiatives.
0G Labs is the first decentralized artificial intelligence operating system (deAIOS), building an AI-dedicated Layer 1 to connect computing resources, data, and models, forming a distributed AI development ecosystem. Developers can leverage 0G’s modular tools to build more efficient and transparent AI Agents while ensuring fair reward distribution. Users can run nodes to contribute computing power to the decentralized network and earn rewards.
DeAgentAI is an innovative platform focused on decentralized AI Agents, committed to advancing multi-agent systems (MAS). Through DeAgentAI, users can create, manage, and coordinate networks of AI Agents applicable across various scenarios such as business automation, data analysis, and personalized recommendations.
Autonomys Network is a decentralized infrastructure stack designed to enable secure, autonomous human-machine collaboration. Anyone with just a solid-state drive (SSD) can join the network as a node. Users can create dedicated AI agents capable of autonomous actions such as booking services, managing funds, scheduling meetings, or even participating in on-chain governance.
Gaia Network is a decentralized AI infrastructure platform supporting the distributed development and operation of AI Agents and applications. By integrating distributed storage, computing, and data verification via blockchain technology, Gaia Network addresses AI challenges around privacy, scalability, and accessibility. Instead of centralized servers, GaiaNet builds a distributed network of edge computing nodes controlled by individuals and enterprises, hosting fine-tuned AI models based on domain-specific knowledge from node operators.
Questflow is a decentralized network composed of multiple AI Agents. Users simply describe their needs, and the AI Agent network autonomously completes tasks. Multiple agents seamlessly collaborate, leveraging collective intelligence and capabilities to achieve efficiency and speed far surpassing that of any single AI model or human.
2. Decentralized AI
Decentralized AI represents the ultimate goal of on-chain AI. Currently, many projects are actively working toward this vision by decentralizing compute, data, and models, aiming to break Big Tech’s monopoly on large language models (LLMs) and empower individuals with ownership over their data and models.
Vana is building a decentralized user data sovereignty platform, transforming personal data into financial assets. As an EVM-compatible Layer 1 blockchain, Vana introduces Data Liquidity Pools (DLP), allowing users to contribute and verify data on-chain to generate tokenized assets for AI model training. Using Proof of Contribution and privacy-preserving technologies, Vana ensures data security and quality. It recently secured investment from YZi Lab, with CZ serving as advisor.
Hyperbolic is an open-access AI cloud platform that aggregates global computing resources to deliver affordable, scalable GPU resources and AI services. The platform supports AI inference, on-demand GPU access, and monetization of idle hardware, helping enterprises, researchers, and individuals reduce costs and efficiently utilize compute. Hyperbolic currently powers well-known on-chain AI projects like Virtuals and AI16z, and announced at this conference its upcoming launch of an AI Agent framework to further advance innovation in AI computing and applications.
OpenLedger is a next-generation network focused on AI and blockchain, providing decentralized economic infrastructure that enables developers to access high-quality data, fine-tune specialized language models (SLMs), and deploy them as paid services. SLMs are its core advantage—custom-built for blockchain, DeFi, and smart contracts. Leveraging the Open LoRA framework, OpenLedger efficiently runs thousands of fine-tuned models on a single GPU, reducing costs while maintaining high throughput. Combined with Datanets and Model Factory, it offers a complete solution for data aggregation, model optimization, and secure deployment.
IO.NET is a decentralized compute platform offering on-demand access to GPU and CPU clusters, eliminating the need for users to invest in expensive hardware or infrastructure. Through the IOG Network—a decentralized physical infrastructure network composed of independently operated hardware nodes—users gain instant access via IO Workers to distributed GPU clusters suitable for high-performance computing and AI model training, all without purchasing costly equipment.
Aethir is an innovative platform specializing in distributed cloud computing infrastructure. Aethir Earth is a bare-metal GPU cloud purpose-built for AI computing, delivering high-performance capabilities; Aethir Atmosphere is a cloud GPU network optimized for gaming, enabling low-latency, high-quality gameplay experiences. Users can also contribute their own GPUs to the Aethir platform as providers and earn rewards.
MinionLab is a decentralized autonomous AI agent network where agents, known as "Minions," operate on users’ devices to mine real-time data from the internet. Device owners earn MINION tokens by supporting the network and running these autonomous data-mining agents. By leveraging idle computing resources, MinionLab offers a more elastic, cost-effective, and scalable approach to data collection, advancing the AI industry toward decentralized, user-driven development.
GAIB is an economic-layer solution targeting AI and high-performance computing, aiming to create a new asset class and economic system by financializing and tokenizing GPU resources. Its core concept treats GPUs as assets and computational power as currency. GAIB leverages a decentralized compute network, fractional ownership, and blockchain technology (L1/L2) to solve existing challenges in resource allocation and financing, meeting the massive demand for high-performance computing in the AI era.
Kite AI is a decentralized Layer 1 blockchain platform designed specifically for the AI economy. Using its innovative Proof of AI (PoAI) consensus mechanism, Kite AI unlocks fair access to and rewards for AI assets—including data, models, and agents. Kite AI aims to build an open, collaborative ecosystem where data contributors, AI developers, and end users all benefit from the decentralized AI economy.
Automata provides middleware-level privacy protection and traceless computation for decentralized applications (DApps). It offers developers and users an efficient, trustworthy, and decentralized environment that maximizes data privacy and user autonomy.
Public AI envisions an open, transparent AI data platform supporting multimodal data collection and annotation, including text, audio, video, and map data. It also delivers efficient, low-cost AI-assisted data labeling. AI assistants perform initial labeling, followed by comprehensive review from AI validators to ensure data quality. Using a fair and transparent Proof of AI consensus mechanism, Public AI enables everyone to contribute data and participate in AI training.
3. Verifiable AI
One major challenge in AI development is the opacity of training processes and inability to guarantee output accuracy. Many projects are now exploring technologies like zero-knowledge proofs (ZKP) and trusted execution environments (TEE) to enable verifiable AI training and ensure reliable outputs.
Phala Network is a decentralized cloud computing platform providing trusted privacy computing and AI inference services for on-chain applications. Its offerings include a confidential computing network based on Trusted Execution Environments (TEE), supporting smart contracts, AI model training and inference, and data privacy protection. Phala Network holds natural advantages in high-privacy use cases such as DeFi, DeSci, NFT marketplaces, and on-chain AI Agent development, using TEE to ensure data security and verifiable computation.
Brevis is a decentralized computing engine focused on verifiable off-chain AI and blockchain computation. By integrating zero-knowledge proofs (ZKP), Brevis enhances privacy and efficiency, particularly in financial data protection. It supports the expansion of smart contracts and Web3 applications, providing a secure computing environment that ensures privacy for on-chain AI inference and transactions while maintaining verifiability—ideal for DeFi, prediction markets, and privacy-centric AI inference.
Verisense Network is an innovative platform dedicated to decentralized data validation and trusted AI. It helps developers verify data sources, ensuring authenticity and integrity of training data. It also enables auditing and verification of AI decision-making processes, offering transparent training data and reasoning workflows to eliminate the “black box” problem and increase trust in AI decisions. Ideal for AI models used in DeFi, prediction markets, and compliance monitoring, Verisense ensures reliability of on-chain data.
2. AI Use Cases: Potential and Expectations
Compared to the rich landscape of AI infrastructure, standout real-world AI use case projects remain relatively scarce. Beyond the familiar Twitter bot AIXBT, speakers at this event highlighted other promising examples such as @BukProtocol in the Virtuals ecosystem offering travel booking services, NBA commentator @HeyTracyAI, and DeFAI Agent @askjimmy_ai built on the ARC framework—each expanding the possibilities for AI Agent applications.
Narra is a GameFi AI Agent platform on Berachain that leverages its AI engine to generate dynamic, real-time narrative content, interact with players, advance storylines, and deliver personalized experiences within gamified environments. Narra also supports the creation and interaction of AI-NFTs—digital assets endowed with intelligent behaviors capable of dynamic engagement with users.
AI Travel is an AI-powered travel assistant that customizes travel plans through conversational interfaces. It can call hotels on behalf of users to book rooms at optimal prices and offers browser plugins for booking and price-comparison of hotels, flights, and activities.
HeyTracyAI is an AI Agent for sports commentary in basketball, backed by NBA champion Tristan Thompson, delivering real-time game analysis and predictive insights.
AskJimmy is an AI Agent platform focused on finance and trading, aiming to build a decentralized, multi-strategy hedge fund fully operated by AI Agents.
3. Traditional Projects Transitioning to AI
Following the trend, many established web3 projects announced their AI transformation strategies during the event.
Legacy chains such as Sui, Near, Flow, and Aptos actively participated in AI-focused sessions, emphasizing that AI Agents significantly simplify complex blockchain interactions, making web3 more accessible and addressing long-standing usability challenges. The tokenization of AI Agents and their ability to collaborate opens up new possibilities for on-chain AI. These legacy chains expressed clear goals to lead in AI adoption, pledging full support through architectural upgrades, account innovations, and hackathons to encourage developer innovation in on-chain AI applications.
Eigenlayer, previously focused on restaking services, stated at its co-hosted AI satellite event that it is building a Decentralized Trust layer, offering Verifiable Cloud services. This will provide on-chain proofs for off-chain operations such as AI training, inference, and prediction, accelerating the development of verifiable AI Agents.
4. Challenges and the Future
As a central theme of Consensus Hong Kong 2025, discussions around AI and Web3 were intense and inspiring. While painting optimistic visions, several speakers acknowledged ongoing challenges in on-chain AI development—such as unreliable models, ambiguous prompt intentions, storage and hardware limitations, and privacy concerns. These hurdles present not only technical obstacles but also fertile ground for innovation. In the long term, there is strong optimism about the future of on-chain AI, with hopes that continued improvements in infrastructure, novel use cases, and community collaboration will jointly drive the integration and prosperity of AI and Web3.
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
@BiteyeCN
