From messari by Nicholas Garcia
Key Insights
- FLock.io's system design integrates decentralized AI model creation, training, and deployment into a streamlined ecosystem. The system comprises three major components: AI Arena, FL Alliance, and AI Marketplace.
- FLock.io was founded by Jiahao Sun, an experienced innovator in artificial intelligence and decentralized systems. Jiahao has an extensive entrepreneurial background, including founding UiiTech and serving in leadership roles at Boolv Tech, where he specialized in AI innovations for computer vision and decentralized technologies.
- FLock.io has raised capital across two major funding rounds: a $6 million Seed Round in March 2024 and a $3 million Strategic Round in December 2024. Investors include Digital Currency Group, Lightspeed Faction, and Animoca Brands.
- The FLOCK token is the native utility token of the FLock.io network, enabling decentralized AI model training, fine-tuning, and validation. It incentivizes participants, ensures security, and supports governance within the platform's three core components.
- FLock has partnered with prominent protocols such as Animoca Brands, io.net, Chainbase, and Aptos, highlighting its strong presence in the crypto ecosystem.
Introduction
FLock.io is at the forefront of decentralizing artificial intelligence (AI) development, addressing the critical issues posed by centralized AI governance and accessibility. Traditional AI development remains dominated by a handful of large corporations that dictate the trajectory of technological advancements based on their objectives. This centralization amplifies biases, limits public participation, and restricts access to computing power, creating a system misaligned with the broader public interest.
To tackle these issues, FLock.io leverages blockchain technology to pioneer a decentralized federated learning platform that democratizes AI development. By integrating Federated Learning with blockchain, FLock.io creates a collaborative environment where data contributors, developers, and validators work together to build AI models. The platform empowers communities to propose, train, and refine models transparently while retaining full control over their data. This approach eliminates centralized bottlenecks, ensures fair distribution of rewards, and fosters the development of ethical, high-quality AI models.
At its core, FLock.io combines the strengths of decentralized governance and federated learning to create a next-generation AI platform. Its mission is to align AI development with societal goals, ensuring that communities—not corporations—shape the future of technology.
Background
Team
FLock.io was founded by Jiahao Sun, an experienced innovator in artificial intelligence and decentralized systems.
Jiahao’s academic background includes earning a Master of Science degree in Computer Science from the University of Oxford. He also served as an Honorary Research Fellow at Imperial College London, where his research focused on knowledge graph generation, natural language processing, and machine learning for applications in financial prediction and data annotation.
Professionally, he has held key roles at RBC Wealth Management, advancing from Lead AI Engineer to Director of Artificial Intelligence. During his tenure, he spearheaded the development of innovative systems like “NORA,” an AI-driven platform for sales and investments.
In addition to these accomplishments, Jiahao has an extensive entrepreneurial background, including founding UiiTech and serving in leadership roles at Boolv Tech, where he specialized in AI innovations for computer vision and decentralized technologies.
Investors
FLock.io raised $6 million in a Seed Round announced on March 27, 2024. This milestone was co-led by Lightspeed Faction and Tagus Capital, with additional participation from Digital Currency Group, OKX Ventures, Volt Capital, and several prominent individual investors, including Stefan George, Martin Köppelmann, and Jasmine Zhang.
In December 2024, FLock.io announced a $3 million Strategic Round led by Digital Currency Group, with participation from Lightspeed Faction, Animoca Brands, Fenbushi Capital, Gnosis VC, GSR, OKCoin Japan, Basis Ventures, and All For One Ventures.
Technology
System Design Overview
FLock.io's system design integrates decentralized AI model creation, training, and deployment into a streamlined ecosystem. The system comprises three major components: AI Arena, FL Alliance, and AI Marketplace.
Models in FLock’s system are initially trained and validated in AI Arena, a blockchain-based decentralized platform where training nodes submit models for evaluation by validators. The top-performing models are fine-tuned in FL Alliance using participants’ local data without compromising privacy. Once optimized, models are deployed in the AI Marketplace, where they are accessible for real-world applications. Feedback from the marketplace further improves these models, creating a continuous cycle of refinement and performance enhancement.
FLock.io Blockchain
The FLock.io blockchain is integral to enhancing system security, incentivization, and trust among participants. Through smart contracts and staking mechanisms, it ensures that rewards are distributed transparently while maintaining network integrity.
Incentivization
The blockchain layer automates reward distribution using smart contracts. Participants stake FLOCK tokens to engage in tasks, fostering a competitive and trustworthy environment. Rewards are dynamically allocated based on performance metrics, incentivizing high-quality contributions from training nodes, validators, and delegators.
Security
FLock employs a proof-of-stake (PoS) mechanism to enhance security and resilience against malicious behavior. Participants who act dishonestly, such as submitting poor-quality models or manipulating validation, risk having their staked tokens slashed. This economic disincentive promotes honest participation.
To mitigate specific attack vectors, FLock employs the following measures:
- Sybil Attacks: Minimum staking requirements prevent attackers from creating multiple fake identities.
- DoS Attacks: Rate limiting ensures no single participant can overwhelm system resources.
- Free-rider Attacks: Rewards are only allocated to top-performing nodes and validators, ensuring contributions are genuine and effective.
- FL Model Poisoning Attacks: Majority voting and slashing mechanisms identify and penalize malicious data submissions.
- Lookup Attacks: Dual validation datasets (A and B) prevent overfitting to a single dataset, ensuring true model quality.
FLock.io AI
AI Arena
AI Arena is FLock.io’s decentralized AI training platform that facilitates the creation of high-quality models. It supports a conventional machine learning paradigm, enabling training nodes to optimize models directly on their devices using public or local datasets.
Training nodes stake FLOCK to participate. Validators evaluate these models, and rewards are allocated based on task performance and ranking.
FL Alliance
FL Alliance enables collaborative model fine-tuning using a Federated Learning (FL) approach. Thousands of participants contribute to training a global model while preserving data privacy—no local data is shared during the training process.
Participants in FL Alliance are randomly assigned roles as Proposers and Voters:
The result is an optimized global model that combines diverse data sources while maintaining data sovereignty. The FL Alliance’s slashing and reward mechanisms further incentivize honest participation.
AI Marketplace
Participants and workflow in AI Marketplace
The AI Marketplace is the final stage of FLock.io’s system, where trained and fine-tuned models are deployed for real-world applications. This platform serves as a decentralized marketplace, allowing developers, businesses, and end-users to:
- Access and host models for specific use cases.
- Contribute data to improve existing models.
- Provide user feedback to drive continuous model refinement.
Models hosted on the marketplace can be integrated into applications through API endpoints. Examples include intelligent agents, trading bots, and domain-specific tools like BTC-GPT, which has already processed over 10,000 model calls.
Summary of Benefits
FLock.io’s technology stack offers a unique combination of decentralized training, secure collaboration, and real-world deployment. Key benefits include:
- Incentivized Collaboration: Transparent reward mechanisms encourage high-quality contributions from all participants.
- Data Sovereignty: Federated Learning ensures that participants retain full control over their local data.
- Scalability: Blockchain integration and decentralized task execution enable the system to scale seamlessly.
- Edge Computing for Decentralized AI Training: By unlocking the potential of private data through decentralized, privacy-preserving AI training on edge devices, FLock addresses niche use cases with strategic advantages:Utilization of domain-specific, private data to enhance model performance.Reduced latency through real-time training on edge devices.Lower infrastructure costs via idle compute resource utilization.
- Utilization of domain-specific, private data to enhance model performance.
- Reduced latency through real-time training on edge devices.
- Lower infrastructure costs via idle compute resource utilization.
- Security: PoS mechanisms, slashing, and validation safeguards protect against malicious activities.
- Real-World Utility: Models deployed through the AI Marketplace provide practical, scalable solutions across industries.
- Revenue Sharing in Decentralized AI: FLock redefines model ownership by enabling continuous revenue sharing for contributors:Contributors earn ongoing rewards as their models are adopted in real-world applications.Integrated monetization through platforms and application-specific use cases generates recurring revenues.A self-sustaining ecosystem incentivizes developers, data providers, and validators.
- Contributors earn ongoing rewards as their models are adopted in real-world applications.
- Integrated monetization through platforms and application-specific use cases generates recurring revenues.
- A self-sustaining ecosystem incentivizes developers, data providers, and validators.
Tokenomics
FLOCK Overview
The FLOCK token is the native utility token of the FLock.io network, enabling decentralized AI model training, fine-tuning, and validation. It incentivizes participants, ensures security, and supports governance within the platform's three core components: AI Arena, FL Alliance, and AI Marketplace. The FLOCK token utilities include:
- Staking: Participants stake FLOCK to join training tasks, validate contributions, or delegate tokens, incentivizing honest participation.
- Payment: Tokens are used to access models on the AI Marketplace and pay fees for participating in training and validation processes.
- Governance: Token holders participate in decision-making through a DAO, voting on allocation of resources for AI Arena and FL Alliance, adjustments to staking parameters and reward pools, and approval of new features and updates to platform rules.
The FLOCK token launched on December 31, 2024, and is live on Base. It is supported by several CEXs, including Bybit, MEXC, and Gate.io.
FLOCK Supply and Vesting
The total supply of FLOCK tokens is capped at 1 billion. Initial minting includes only the day-one unlock tokens, with additional tokens minted daily through a management smart contract. Key allocations include:
- Community Airdrops: Reserved for incentivized testnet participants based on activities like staking, training, and validation. Future incentives will be minted over a 60-month period with a 1% monthly decay.
- Emissions: Allocated to support R&D, task creation, protocol development, and ecosystem initiatives.
- Investors: Early supporters follow a 3-year vesting schedule (1-year cliff and 2 years of linear vesting). Locked tokens cannot be staked until vested.
- Team and Advisors: Contributors to FLock technology follow the same vesting schedule as investors (1-year cliff and 2 years linear). Locked tokens cannot be staked until vested.
Airdrop Details
The FLock airdrop allocates 50,000,000 FLOCK (5% of the total supply) to trainers, validators, delegators, model users, and community partners. Eligibility is based on contributions during the testnet phase, such as earning FLOCK tokens or achieving specific scores on the Model Store. To ensure fairness, a verifiable lottery system determines individual airdrop amounts, with Sybil resistance measures in place.
Use Cases and Partnerships Overview
AI Assistants
FLock has established itself as a leader in decentralized AI training, enabling the development of advanced AI assistants like Farcaster GPT and Scroll GPT.
- Farcaster GPT: Created during the AI Arena’s inaugural hackathon at the Base Onchain Summer Buildathon. Participants collaboratively trained a chatbot using Farcaster documentation, resulting in the winning model, Mistralai-Mistral-7B-Instruct-v0.3, hosted on Hugging Face.
- Scroll GPT: Leveraging Retrieval-Augmented Generation (RAG), Scroll GPT offers a live chat interface. Hosted on FLock, it fosters continuous improvement through user contributions and incentivizes engagement by awarding points.
AI Companions: Professor Grump with Akash
FLock’s partnership with the Akash Network introduces capabilities for running AI nodes on decentralized compute resources. By creating one-click templates, FLock simplifies the deployment of validators on the Akash Supercloud, enabling efficient AI model training.
Web3 Agents: Text2SQL Model
The Text2SQL model, developed in partnership with Chainbase, simplifies database queries through natural language commands. Developed on FLock’s AI Arena, it democratizes data access for users across various domains. By allowing users to query databases in plain language, Text2SQL bridges the gap between technical complexity and accessibility.
Privacy-Preserving Healthcare
FLock’s federated learning approach addresses one of healthcare’s most pressing challenges: data sharing. Collaborating with a leading British hospital, FLock demonstrated its ability to train AI models on sensitive healthcare data without compromising privacy. Initial tests in blood glucose prediction validated the effectiveness of this approach, showcasing its potential for broader applications in chronic disease management and global healthcare innovation.
Proof of AI
In partnership with io.net, FLock has launched the world’s first Proof of AI consensus mechanism. This innovation is designed to ensure that DePINs genuinely contribute compute resources rather than simulate participation to collect rewards. By deterring dishonest behavior and aligning incentives, PoAI enhances the profitability of genuine contributors and ensures the long-term sustainability of the network.
Aptos Move Language Coding Copilot
FLock has partnered with the Aptos Foundation to develop an advanced AI-assisted Move language coding copilot, tailored specifically for Aptos developers. By leveraging decentralized training capabilities, FLock’s Move-tuned LLM outperforms mainstream models like ChatGPT-4o in generating Move-specific code.
Partnership with Animoca Brands
FLock.io has signed a Memorandum of Understanding with Animoca Brands to collaborate on developing domain-specific AI models tailored for blockchain applications. The partnership focuses on creating four specialized AI models to support due diligence, investment analysis, market-making, and operational tasks for Animoca Brands’ ecosystem.
Roadmap
FLock.io’s 2025 roadmap outlines key milestones to drive the platform’s growth and innovation. In Q1 2025, the FL Alliance testnet will launch alongside partnerships with leading Web2 and Web3 firms, and plans to decentralize train.flock.io will be published. By Q2 2025, the mainnet for FL Alliance will go live, accompanied by the launch of custom model and application development services. In H2 2025, FLock aims to launch the AI Marketplace, introduce its Web3 Model benchmarks, and finalize its foundational model for crypto.
Closing Summary
FLock.io integrates blockchain technology and decentralized systems to enable collaborative development and deployment of AI. By combining Federated Learning with decentralized governance, FLock.io allows communities to create and refine AI models while maintaining transparency and data sovereignty.
The FLOCK token facilitates participation in the network by supporting incentivized collaboration, secure interactions, and decentralized decision-making. Backed by investors such as Digital Currency Group, Lightspeed Faction, and Animoca Brands, as well as partnerships with organizations including Chainbase, io.net, and Aptos Foundation, FLock.io aims to advance the development of decentralized AI systems.
FLock’s roadmap includes milestones such as the launch of the FL Alliance mainnet, the AI Marketplace, and the creation of Web3-focused foundational models. These initiatives address challenges in AI governance, security, and accessibility.
Let us know what you loved about the report, what may be missing, or share any other feedback by filling out this short form. All responses are subject to our Privacy Policy and Terms of Service.
This report was commissioned by FLock.io All content was produced independently by the author(s) and does not necessarily reflect the opinions of Messari, Inc. or the organization that requested the report. The commissioning organization does not influence editorial decision or content. Author(s) may hold cryptocurrencies named in this report. This report is meant for informational purposes only. It is not meant to serve as investment advice. You should conduct your own research, and consult an independent financial, tax, or legal advisor before making any investment decisions. Past performance of any asset is not indicative of future results. Please see our Terms of Service for more information.
No part of this report may be (a) copied, photocopied, duplicated in any form by any means or (b) redistributed without the prior written consent of Messari®.
All Comments