Fetch.ai

Fetch.ai is an artificial intelligence and blockchain company that operates a decentralized network combining these technologies to enable autonomous economic agents to perform tasks and make decisions independently.^[21] The platform facilitates data sharing, machine learning, and automation to create scalable solutions across various industries. It was launched in March 2019 by Toby Simpson, Humayun Sheikh, and Thomas Hain.^[1]​^[2]​

Overview

Fetch.ai is a platform that enables a decentralized digital economy powered by autonomous AI agents. These agents are intelligent programs that act on behalf of users, devices, or organizations to perform tasks, make decisions, and exchange value. Fetch.ai provides the infrastructure and tools needed to build, deploy, and manage these agents, including the uAgents communication protocol, the Agentverse discovery platform, and a native blockchain that facilitates secure transactions using the FET token. The system allows agents to interact with one another, access shared resources, and collaborate to complete complex workflows in real time.

The platform also integrates an AI Engine that translates human input into agent-driven actions, enabling dynamic and adaptive task execution. Developers can use frameworks like LangGraph or AutoGen and decentralized computing resources like Fetch Compute to create scalable, purpose-driven agents. Once registered on the Agentverse and connected through the Almanac smart contract, these agents become discoverable and can transact securely with others. By combining AI, blockchain, and decentralized infrastructure, Fetch.ai creates a unified environment where autonomous agents operate efficiently across supply chains, data services, digital marketplaces, and more.^[8]​

Artificial Superintelligence Alliance Merger

On July 1, 2024, Fetch.ai, SingularityNET, and Ocean Protocol announced the formation of the Artificial Superintelligence Alliance through a multi-token merger, combining their respective tokens—FET, AGIX, and OCEAN—into a unified token under the FET ticker. This consolidation aims to streamline operations, unify tokenomics, and advance the development of decentralized artificial intelligence infrastructure. In October 2025, Ocean Protocol Foundation withdrew its directors and resigned as a member of the Alliance, leaving Fetch.ai and SingularityNET as its primary organizational participants alongside subsequent ecosystem partners.^[18]​^[19]​

The merger included the launch of migration tools via the SingularityDAO dApp, enabling holders of AGIX and OCEAN to convert their assets to FET. As part of the transition, AGIX and OCEAN began delisting from exchanges, while FET continued trading uninterrupted. The alliance is coordinating with exchanges and aggregators to support the token migration process.

Future phases of the merger include deploying the new ASI token across multiple blockchains, onboarding community members, and introducing new migration contracts for unconverted tokens. All FET tokens on Fetch.ai’s mainnet will automatically convert to ASI upon the mainnet upgrade. The migration contracts, including the OCEAN–FET bridge managed by Fetch.ai that converts OCEAN to FET at a rate of 0.433226 FET per OCEAN, are planned to remain open long-term to ensure accessibility.^[15]​^[18]​

Funding

In March 2021, Fetch.ai received funding from GDA Group, a Toronto-based digital asset firm that invested $5 million in the Fetch ecosystem over an undisclosed period. The funding was announced to be used for the research and development of Fetch's applications, which focus on autonomous artificial intelligence (AI) agents deployed on behalf of large firms like Bosch and others.^[6]​

In March 2023, Fetch.ai raised $40 million from market maker and investment firm DWF Labs. The investment was said to be used to deploy decentralized machine learning, autonomous agents, and network infrastructure on its platform.^[5]​

Fetch Name Service

On July 27, 2023, Fetch.ai announced an update to Fetch.ai Wallet version 0.14, which gives access to Fetch Name Service (FNS) directly within the wallet. The Fetch Name Service, created by AzoyaLabs, provides a Name System for the Fetch.ai blockchain. It allows users to map a human-readable domain to wallet addresses. With wallet upgrade 0.14, users can buy and register any unique .FET domain name directly from within their wallet.^[7]​

Technology

Since Fetch.ai is a multi-utility network, it combines a three-pillar framework consisting of autonomous economic agents, an open economic framework, and Smart Ledgers.

The Fetch.ai ecosystem uses a principle of AI known as Multi-Agent Systems (MAS), which allows agents to work in harmony despite being heterogeneous. An agent is a part of the software that represents an entity. Through its autonomy, it can operate independently (without its owner’s influence) and continuously make decisions on its own behalf. ^[4]​

Autonomous Economic Agents (AEAs)

AEAs are intelligent and independent programs that operate on the owner’s behalf. The goal of these agents is to generate economic value for their owners. The Fetch.ai platform provides an environment where agents seeking value are placed near those with that value. Therefore, each agent sees the world as being reorganized for its needs. These autonomous economic agents can be made to represent real-world objects, people, or assets to allow them to communicate and transact. ^[3]​

Open Economic Framework (OEF)

The Open Economic Framework (OEF) contains protocols, languages, and market mechanisms that facilitate communication and exchange of information between agents. The layer on top of the raw protocol and ledger provides the environment for all economic agents to operate. This enables a decentralized space where agents can find each other and exchange information.^[3]​^[4]​

Smart Ledger

The Fetch Smart Ledger, the Fetch.ai mainnet, is a smart contract-enabled ledger consisting of Distributed Ledger Technologies (DLTs) blockchain elements. It contains built-in AI based on notarized DAG technology that supports millions of economic agents and logs all transactions made on the blockchain. ^[4]​

Consensus, Staking, and Governance

The Fetch.ai/ASI network operates as a proof-of-stake blockchain in which a set of validators commit new blocks and participate in consensus by broadcasting cryptographically signed votes.^[22] Validators secure the network by staking FET (ASI) tokens, and the candidates with the most stake delegated to them become active validators. If validators double-sign blocks, remain frequently offline, or fail to participate in governance, a portion of their staked tokens, including delegated stake, can be slashed, with penalties varying according to the severity of the violation.^[22]​

Token holders who do not run validator infrastructure can delegate their FET (ASI) stake to validator candidates, contributing to network security and sharing in staking rewards distributed to bonded stake.^[22] On-chain governance is carried out through formal proposals that any FET holder can submit, including parameter-change and software-upgrade proposals.^[23] Proposals enter a 5-day voting period once a minimum deposit threshold (minDeposit) is met, during which bonded FET holders cast token-weighted votes choosing Yes, No, NoWithVeto, or Abstain.^[23]​

Within this system, the FET (ASI) token functions as the medium of exchange for transactions, smart contract execution, and agent interactions on the Fetch Ledger, while also serving as the staking and governance token that underpins validator security, delegation, and proposal voting across the ASI Network.^[24]​

Products

Agentverse

Agentverse is a cloud-based platform that enables users to build, deploy, and manage autonomous agents without dealing with backend infrastructure. It features an in-browser integrated development environment (IDE) that simplifies coding, editing, and running agents. With built-in tools for logging, file management, and code revisions, users can create functional agents using any framework and leverage predefined templates to streamline deployment. Once launched, agents remain continuously online, scalable with message volume, and can securely interact with blockchain contracts through built-in wallets. The platform supports various Python libraries, and agents can be listed in the Almanac for discovery by other agents or users.

Key features include blockchain-based identity for transparent registration, mailroom services for handling offline messages, and secure, isolated Python environments. Agentverse is framework-agnostic and supports various applications, from basic scripts to complex AI agents. Its API allows decentralized communication, real-time interaction, and programmatic integration. Agentverse is designed to reduce DevOps overhead by handling deployment and scaling of agents on behalf of users. As a directory for over 2.7 million agents, it enables scalable, trustless, and resilient agent operations within a decentralized ecosystem.^[9]​^[16]​

AI Engine

The AI Engine is designed to interpret human text input and convert it into actionable tasks by directing requests to suitable software agents. It functions as a network of interconnected large language models (LLMs) and agents, forming a scalable AI infrastructure. As more agents connect to the system, its capabilities expand, enhancing the quality and variety of services it can provide. The core purpose of the AI Engine is to understand user input in natural language and match it with the most appropriate agents registered in the Agentverse to complete the requested objectives.

The AI Engine uses contextual understanding to assess user goals and preferences. It analyzes ongoing trends and past user interactions—individual and collective—to transform seemingly random input into meaningful responses. During its training and current open beta phase, the system also solicits user feedback when uncertainty arises, helping to verify that its recommendations align with user intent. This feedback loop supports the AI Engine’s goal of anticipating and adapting to user needs over time.

A key feature of the AI Engine is its smart routing mechanism. Evaluating agents' performance history and capabilities registered in the Almanac ensures tasks are assigned to agents with the most relevant expertise. When a single agent cannot fulfil a request, the AI Engine constructs dynamic multi-agent workflows in real time. These workflows enable multiple agents to work collaboratively as independent components orchestrated by the AI Engine, allowing for the resolution of more complex user requests.^[8]​

World’s First AI-to-AI Payment for Real-World Transactions

On December 18, 2025, Fetch.ai announced what it described as the world's first AI-to-AI payment for a real-world transaction.

The milestone was achieved when a Fetch.ai agent, utilizing DeltaV, autonomously discovered, booked, and paid for a room at Hotel Satoshis. This event demonstrated the platform's ability to enable personal AI agents to coordinate and transact for services without direct user intervention at each step.

AI-to-AI payments allow personal AIs to move beyond simple assistance and into full execution, securing reservations, booking experiences, and completing purchases automatically, before opportunities are lost. This marks a shift in how AI interacts with the real world, giving users the ability to delegate time-sensitive tasks while maintaining control.

The capability was rolled out to users in early 2026.^[17]

ASI1

ASI1 (also referred to as ASI:One) is a Web3-native large language model created by Fetch.ai, built specifically for use in decentralized environments with autonomous AI agents. Unlike general-purpose LLMs, ASI1 is optimized for agentic reasoning, enabling it to independently plan, adapt, and execute complex multi-step tasks without ongoing user input. It supports context-aware responses and maintains memory across interactions, making it suitable for long-running or dynamic agent workflows.

The model integrates with the ASI wallet, which uses the FET token to enable secure on-chain transactions and interactions within the Fetch.ai network. ASI1 can be accessed via API and is available in three configurations—mini, fast, and extended—allowing for flexibility based on the required task complexity and performance. Its primary applications include decentralized finance, autonomous supply chain processes, personalized AI assistants, and smart contract automation, emphasizing functionality in agent-operated, decentralized systems.^[10]​^[11]​

Newer ASI:One features include a Card Playground that allows developers to construct and test interactive agent cards and user interfaces before connecting them to live agents, and Community Skills, which function as modular capabilities that users can activate to extend what their personal AI can do. The platform also offers Athena, a deep-work mode designed for multi-stage research and workflow projects that can chain together analysis, tool use, and deliverable creation, with each asi1.ai account receiving a limited allowance of free Athena projects to trial the feature.^[19]​^[20]​

ASI-1 Mini

ASI-1 Mini is a Web3-oriented large language model optimized for agentic workflows while running efficiently on just two GPUs. Built as part of the ASI: initiative, it introduces an architecture that combines a Mixture of Models (MoM) and Mixture of Agents (MoA), enabling dynamic selection of specialized models and coordination of autonomous agents. These agents manage real-world tasks like API integration, database operations, and decentralized workflows, allowing ASI-1 Mini to handle multi-step processes with adaptability, speed, and precision. Its reasoning system includes four distinct modes—Multi-Step, Complete, Optimized, and Short—tailoring decision-making to the task and supporting real-time corrections and improved transparency.

The model is deeply integrated into the ASI: ecosystem, powered by $FET and accessible through the ASI wallet. It forms the foundation of a community-driven AI framework that allows users to stake, train, and co-own foundational and specialist models within the Cortex collection. ASI-1 Mini also addresses the black-box problem by enabling continuous reasoning and more explainable outputs, which is critical for high-stakes applications. Upcoming expansions include agentic tool-calling, multimodal support, and vastly increased context windows—scaling to 10 million tokens—to support complex enterprise use cases like legal review and financial analysis. Through ASI:, the Web3 community gains a direct stake in the growth and value creation of next-generation AI models.^[11]​^[12]​

ASI:One Mobile

ASI:One Mobile is the first mobile interface for Fetch.ai’s ASI1 large language model. It enables users to interact with agent-based AI systems directly from their smartphones. The app connects to Agentverse, Fetch.ai’s decentralized agent network, allowing users to trigger task execution, access specialized agents, and receive structured responses powered by contextual data and knowledge graphs.

This initial release focuses on usability and exploration, offering functions like itinerary planning, technical support, and research assistance through intelligent agents. ASI:One Mobile is built on Fetch.ai’s agent infrastructure and is expected to expand over time with features such as session memory, deeper on-chain integrations, and additional tools for complex workflows.^[13]​

ASI Token

The ASI token is the product of a multi-phase merger between three decentralized AI projects: Fetch.ai (FET), SingularityNET (AGIX), and Ocean Protocol (OCEAN). The goal is to create a unified token and ecosystem under the Artificial Superintelligence Alliance (ASI) to lead development in decentralized AI.

The merger unfolded in two main phases starting on July 1, 2024. AGIX and OCEAN tokens were converted into FET on the Ethereum blockchain in the first phase. AGIX and OCEAN were delisted from exchanges while FET trading continued uninterrupted. Holders of AGIX and OCEAN tokens on centralized exchanges, like Binance, had their tokens automatically converted to FET at fixed rates. In contrast, self-custody holders manually migrated their tokens through the official SingularityDAO decentralized application. This phase also included rebranding the project across platforms under the Artificial Superintelligence Alliance banner.

Phase two focused on deploying the ASI token and transitioning from the FET ticker to ASI at a one-to-one ratio. Additional migration contracts facilitated the conversion of any remaining AGIX and OCEAN tokens. Concurrently, the Fetch.ai mainnet upgraded to the ASI network, improving the blockchain’s performance, security, and scalability. Staked FET tokens automatically converted to ASI, while liquidity providers adjusted their positions manually. All trading pairs were updated to reflect the ASI token after the upgrade.

To support the migration, approximately 1.48 billion new FET tokens were minted to cover conversions from AGIX and OCEAN holders. This brought the total FET supply to about 2.63 billion tokens, which eventually rebranded as ASI tokens. The migration and network upgrade process was carefully planned and tested to minimize disruptions and ensure a smooth user transition.^[14]​

In October 2025, Ocean Protocol Foundation withdrew from the Artificial Superintelligence Alliance, while the OCEAN→FET bridge operated by Fetch.ai remained open at a fixed rate of 0.433226 FET per OCEAN for holders who still choose to convert. Unconverted OCEAN tokens continue to trade on major exchanges, and although Ocean’s governance role in the Alliance ended, the tokenomics integration from the merger, including the prior OCEAN-to-FET conversions, remains largely intact.^[18]​

Partnerships

• Minima – Fetch.ai collaborates with Minima to integrate Minima’s Integritas on-chain verification system with ASI-1 autonomous agents, aiming to enhance data integrity and reliability for agent-driven applications.^[25]
• AITECH / Solidus AI Tech – Through a strategic partnership with Solidus AI Tech, Fetch.ai’s ASI:One large language model is listed on the AITECH AI Marketplace, providing developers and enterprises with access to tools for building and deploying intelligent agents via a dedicated AI marketplace.^[26]
• Secret Network / Secret VM – Fetch.ai and Secret Network work together on infrastructure for verifiable and privacy-preserving AI agents, combining Fetch.ai’s agent framework with Secret Network’s confidential virtual machines that can produce attestation proofs about the code an agent executes.^[27]
• Rivalz – Rivalz Network integrates Fetch.ai’s ASI-1 large language model into its Agentic Data Coordination System (ADCS), enabling decentralized applications to access AI inference directly on-chain for data coordination and analysis.^[28]
• Functionland – Fetch.ai partners with Functionland to combine the ASI-1 Mini LLM with Functionland’s NPU-powered edge hardware, creating an autonomous AI ecosystem that operates at the edge without reliance on centralized cloud infrastructure and is intended to improve privacy, efficiency, and user control.^[29]
• The Game Company – Fetch.ai collaborates with The Game Company to integrate ASI-1 Mini into TGC’s AI-powered cloud gaming platform, using Fetch.ai’s model to enhance in-game intelligence and adaptive gaming experiences.^[30]
• Zus Network – Fetch.ai forms a strategic partnership with Züs Network, leveraging Züs’s decentralized storage infrastructure to provide secure and scalable storage solutions intended to support Fetch.ai’s AI agent marketplace and enterprise-focused applications.^[31]
• Alibaba Cloud – Fetch.ai builds decentralized AI solutions on Alibaba Cloud, using Alibaba Cloud’s infrastructure to host and scale its blockchain-based AI agent network and to support deployment of agent-based services to enterprise users.^[32]
• Injective – Injective and Fetch.ai collaborate to bring autonomous AI agents to decentralized finance, focusing on having Fetch.ai agents interact with DeFi applications on the Injective blockchain to automate trading and other on-chain operations.^[33]