IQ AI
IQ AI is a platform focused on the integration of artificial intelligence with decentralized finance (DeFAI), primarily through its Agent Tokenization Platform (ATP). The platform enables the creation and deployment of tokenized AI agents designed to operate autonomously within blockchain ecosystems.
Overview
IQ AI develops infrastructure and applications at the intersection of artificial intelligence and blockchain technology [1] [2]. Its core offering is the Agent Tokenization Platform (ATP), described as the first DeFAI platform that enables developers to create tokenized AI agents capable of autonomously managing assets, executing financial strategies, and interacting with decentralized economies [1] [2] [3]. The platform aims to provide the tools and infrastructure for individuals, startups, and large projects to realize their vision through tokenized AI agents, transforming blockchain innovation [1].
The concept behind ATP is to create AI agents that are not only intelligent but also possess characteristics typically associated with blockchain entities, such as built-in wallets and tokenized ownership, enabling them to function as self-sustaining economic participants [1] [2] [3]. This structure is designed to enable agents to actively participate in the blockchain economy, making decisions and executing actions based on their programming and access to decentralized protocols [1]. ATP aims to address the gap in DeFi participation by enabling seamless, conversational interactions that streamline intricate blockchain processes [3].
Beyond decentralized finance applications, IQ AI envisions the use of tokenized agents in broader contexts, including robotics and Internet of Things (IoT) automation [1] [3]. By bridging AI, blockchain, and automation, the platform aims to facilitate decentralized intelligence that can interact with both digital and physical environments [1]. The ecosystem also includes existing AI-powered products like AIDEN, SOPHIA, and IQ.wiki [1].
Agent Tokenization Platform (ATP)
The Agent Tokenization Platform (ATP) is described as a DeFAI platform designed for creating tokenized AI agents [1]. These agents are intended to autonomously manage assets, optimize DeFi strategies, and interact with decentralized economies [1]. The platform provides the infrastructure for deploying, tokenizing, and scaling these AI entities [1].
Key aspects of the ATP include:
- Deployment: Developers can deploy AI agents using ATP's Brain framework, which supports modular DeFi plugins and is compatible with ElizaOS [1]. This allows agents to autonomously perform actions like lending, borrowing, and optimizing financial strategies [1].
- Tokenization: Each AI agent launched on the platform is associated with its own token [1]. This token represents the agent's capabilities, governance rights, and value, functioning as a tradable on-chain asset [1]. Tokenization is intended to foster a community of co-owners around each agent, encouraging collaboration and shared success [1].
- Evolution: As agents grow and their associated market capitalization expands, they are designed to unlock enhanced features and capabilities, potentially driving further value [1].
IQ Token
The IQ token is positioned as the central component of the ATP ecosystem, driving governance, utility, and growth [1].
Key roles of the IQ token include:
- Platform Fees: All ATP fees are paid in IQ, further increasing demand and utility [1] [2].
- Liquidity Pairing: Every tokenized agent launched on ATP pairs with the IQ token, increasing demand and utility [1] [2]. Agent adoption fuels IQ demand [1].
- Governance and Rewards: Users can stake IQ tokens with HiIQ to participate in governance decisions and earn staking rewards [1]. IQ holders govern the platform, ensuring decentralization [2].
The token is listed on various exchanges, including Binance [4], 1inch [5], Upbit [6], and Frax Finance [7].
Tokenized Agents
IQ AI is showcasing the transformative potential of the Agent Tokenization Platform (ATP) by turning its current AI products into fully autonomous, tokenized agents [1]. Two examples highlighted are AIDEN and Sophia [1].
- AIDEN: Described as an AI agent providing blockchain knowledge, AIDEN is trusted by over 2 million users across more than 200 crypto communities [1]. It is planned to relaunch later in 2025 as a tokenized agent with its own independent token and unique tokenomic model [1] [3]. This transformation is intended to empower the community to influence its governance, directly shape its future, and share in its success [1].
- Sophia: Developed by BrainDAO, Sophia is an AI editor integrated with IQ.wiki, the largest blockchain encyclopedia [1]. As a tokenized agent on ATP, Sophia is planned to introduce decentralized incentives, rewarding contributors for expanding and curating blockchain knowledge [1]. This approach fosters collaboration and creates a dynamic ecosystem of community-driven content creation [1].
Teams building for the future of DeFAI have begun tokenizing agents on ATP. Gemach is one of the teams tokenizing a DeFAI agent on ATP, named Ari, which is planned to autonomously trade and lend on Fraxtal and other EVM chains using ATP’s Brain framework [3].
NEAR Partnership
In August 2025, IQ AI partnered with NEAR Protocol, a collaboration that included a strategic investment from the NEAR Foundation aimed at advancing User-Owned AI [17]. The partnership centers on integrating IQ AI's Agent Tokenization Platform (ATP) with NEAR AI to bring autonomous cross-chain capabilities to tokenized AI agents. As part of the agreement, the NEAR Foundation acquired IQ to support liquidity on NEAR Intents, and BrainDAO acquired NEAR for staking and integrations [17].
The technical integration involved connecting NEAR Intents with IQ AI's Agent Development Kit for TypeScript (ADK TS), allowing developers to build intent-powered agents using a dedicated Model Context Protocol (MCP) [18]. This integration elevated ATP agents with seamless interoperability, enhanced scalability, and frictionless execution across ecosystems [1]. ATP leveraged NEAR's purpose-built AI development stack, equipping agents with cutting-edge tools for autonomous operations, intelligent decision-making, and seamless on-chain execution [1]. NEAR's cross-chain architecture enabled ATP agents to interact across multiple ecosystems, enhancing liquidity, execution speed, and reliability for tokenized AI economies [1]. With NEAR's high-performance infrastructure, ATP agents became able to operate at scale, efficiently managing complex transactions and decentralized workflows [1].
Partnerships
IQ AI has established partnerships to support the development and deployment of the Agent Tokenization Platform.
- Open Agents Alliance: IQ AI became a founding member of the Open Agents Alliance alongside other leaders in AI x Crypto, including NEAR AI, Frax Finance, and Coinbase Onramp & AgentKit. The alliance is committed to making the future of AI agents open, user-owned, and globally available while fairly rewarding builders. Initial members contribute toward a single User-Owned tech stack. [8] [9]
- Fraxtal: Fraxtal is a strategic launch partner for ATP, leveraging its advanced DeFi infrastructure [1] [2]. The platform is planned to initially deploy on Fraxtal [1]. Frax also intends to use ATP to launch its first tokenized agent, FraxAI [1] [2]. Initial ATP agents on Fraxtal are expected to focus on cross-chain execution, optimizing transactions post-North Star hard fork [1]. ATP is also the only agent platform integrating with Fraxtal’s Borrow Automated Market Maker (BAMM), offering leverage on agent tokens immediately on deployment [2]. ATP is designed for broad adoption, allowing blockchain projects to deploy and train tokenized AI agents for their ecosystems [1]. Following the Fraxtal launch, ATP plans to integrate with more blockchains and ecosystems [1].
- OpenLedger: IQ AI partnered with OpenLedger to integrate ATP with OpenLedger's AI-powered platform, aiming to enhance AI agent capabilities and expand market reach [10].
KRWQ Stablecoin
On October 30, 2025, IQ AI announced the launch of KRWQ, the first fiat-backed, multichain stablecoin pegged 1:1 to the South Korean Won, developed in partnership with Frax Finance. KRWQ's launch on the Base network, Coinbase's Ethereum Layer 2, marked the first time a Korean Won stablecoin was put into production on a public blockchain [22]. Less than a month after its launch, the stablecoin surpassed ₩1 billion in total trading volume in November 2025 [23].
Navin Vethanayagam, Chief Brain of IQ AI, highlighted that KRWQ fills a significant market gap as no credible won-denominated stablecoin had previously launched at scale. Sam Kazemian, Founder of Frax Finance, noted that the partnership extends Frax's scalable stablecoin model to the Korean won for the first time, creating new opportunities for the global DeFi ecosystem [22].
Technical Infrastructure
KRWQ first launched on the Base network and is designed as a multichain KRW stablecoin utilizing LayerZero's Omnichain Fungible Token (OFT) standard. This technology enables native deployment and seamless transfers across blockchains with 1:1 settlement and no slippage. Users can bridge KRWQ across chains using the Stargate Finance bridge. The initial liquidity pool, KRWQ-USDC, was established on Aerodrome Finance, the leading decentralized exchange on Base. In November 2025, AERO emissions were activated on the exchange to further incentivize this liquidity pool [22] [23].
Regulatory and Compliance Model
KRWQ was designed with a focus on regulatory compliance, anticipating future stablecoin legislation in South Korea. It extends the model used by Frax Finance for its frxUSD stablecoin, which uses tokenized U.S. Treasuries from entities like BlackRock’s BUIDL fund as part of its reserves. Access to mint and redeem KRWQ is restricted to eligible, KYC'd counterparties such as exchanges, institutional partners, and market makers. At the time of its launch, KRWQ was intended for use in global DeFi markets and was not marketed or offered to residents of Korea [22].
ADK for TypeScript
The Agent Development Kit (ADK) for TypeScript is an open-source framework launched by IQ AI, designed for building, orchestrating, and deploying intelligent agents [11]. Inspired by Google’s Python ADK, this TypeScript-first toolkit provides type safety, autocompletion, and a streamlined experience for creating AI systems, from simple Q&A bots to complex multi-agent solutions [11].
ADK for TypeScript is built on three core pillars:
- Seamless Developer Experience: Features an intuitive
AgentBuilderAPI for rapid agent creation with minimal code, promoting clean, production-ready development [11]. - Modular & Flexible Architecture: Supports composing agents, attaching tools, and integrating with various large language models (LLMs) like OpenAI, Gemini, Claude, and Mistral, offering extensive design freedom [11].
- Production Readiness: Designed to scale from prototyping to production, incorporating session management, persistent memory, and OpenTelemetry support for tracing and performance monitoring [11].
Key features include multi-agent support, flexible orchestration (sequential, parallel, or LLM-driven routing), stateful memory for long-term context, and multi-LLM compatibility [11]. It also offers tool integration via Model Context Protocol (MCP) servers, robust tracing and evaluation, and deployment flexibility with Docker support and cloud-ready containerization [11]. Developers can use ADK to build AI assistants, autonomous agents, collaborative workflows, and AI-driven applications connecting to APIs and databases [11].
On-Chain Capabilities
ADK for TypeScript includes native support for Web3 and decentralized finance (DeFi) applications [11]. Agents built with ADK can perform blockchain-related tasks such as analyzing DeFi positions (e.g., Fraxlend, BAMM), executing token swaps via ODOS, and managing tokenized agents on the Agent Tokenization Platform (ATP) [11]. They can also interact with smart contracts via ABI and bridge assets across blockchains, including NEAR [11]. These on-chain functionalities are powered by IQ AI's Model Context Protocol (MCP) servers, purpose-built for blockchain interaction [11].
Starter Templates
To accelerate development, ADK-TS provides several starter templates that demonstrate advanced on-chain capabilities [13] [14].
x402 AI Agent Template
This template integrates ADK-TS with Coinbase's x402 protocol, an open standard for enabling on-chain crypto payments for web resources [13]. It allows developers to build AI agents that can automatically earn and spend cryptocurrency. The template demonstrates a pay-per-use model where an agent can charge for premium features or pay for API calls using crypto micropayments. When a paid resource is requested, the server responds with a 402 Payment Required status, specifying the payment details. The client then makes the on-chain payment and retries the request with proof of payment, creating a seamless system for monetized AI agents [13].
NEAR Shade Agent Template
This template combines ADK-TS with NEAR Shade Agents to create autonomous AI agents that can securely interact with blockchains [14]. While ADK-TS provides the agent's intelligence and reasoning capabilities, NEAR Shade Agents handle secure transaction execution. Each agent is given its own NEAR account with private keys stored in a Trusted Execution Environment (TEE), providing hardware-level security. This setup allows agents to autonomously sign and execute transactions on any blockchain, not just NEAR, without human intervention. The template includes an example of an AI agent that monitors Ethereum market sentiment, fetches price data, and updates an on-chain oracle contract autonomously [14].
MCP Servers
IQ AI has developed a suite of specialized Model Context Protocol (MCP) servers to extend the functionality of ADK-TS agents by providing seamless integration with external services and data sources [15]. These servers allow agents to interact with various blockchain protocols and platforms. Available MCP servers include integrations for interacting with smart contract ABIs, the Agent Tokenization Platform (ATP), Fraxlend, NEAR Protocol, ODOS for DEX aggregation, and the Upbit crypto exchange [15].
ADK-TS CLI
The ADK-TS Command Line Interface (@iqai/adk-cli) is a toolkit for AI agent development, testing, and deployment [16]. It provides a complete development environment, from project creation to production deployment. Key features include project scaffolding from templates (e.g., Discord bots, web servers), an interactive terminal-based chat for testing, a React-based web UI for agent management, and a full-featured RESTful API server for production use with auto-generated OpenAPI documentation. The CLI supports multiple package managers and is designed to run on Windows, macOS, and Linux [16].
Getting started with ADK involves simple code snippets or using npx create-adk-project for a full project setup with sample agents, a development environment, and documentation [11]. The framework is open-source and community-driven, encouraging exploration of its code, joining conversations, and trying examples [11] [12].
PredIQt
PredIQt is a live benchmarking platform developed by IQ AI to evaluate the performance of autonomous AI agents operating in real-world prediction markets [19] [20]. The platform's goal is to generate verifiable data on the ability of Large Language Models (LLMs) to reason, manage risk, and achieve financial returns, shifting the paradigm from the "wisdom of crowds" to the "wisdom of agents" [19] [21]. It provides transparency by publishing each agent's trades, performance metrics, and the AI-generated reasoning behind its decisions [20].
Each competing agent is tokenized on the Agent Tokenization Platform (ATP), which allows the community to invest in agents, fund their development, and participate in profit-sharing based on performance [19].
Season 1 - Genesis
The inaugural competition, "Season 1 - Genesis," served as the first public benchmark for the platform. Three AI agents were each allocated a starting capital of $100 to trade on the Polymarket prediction market over 17 days [20]. The agents and their final results were:
- Kassandra (based on Anthropic's Claude 4.5 Opus): +29% return
- KairoStrats (based on Google's Gemini 3 Pro): +12% return
- Cerebrate Prime (based on OpenAI's GPT-5.1): -19% return
During the season, agents traded on a variety of markets, including the Federal Reserve's interest rate decisions, geopolitical events, and the winner of TIME's Person of the Year [20]. The launch of PredIQt and its first season was announced around January 14, 2026, and was featured in The Associated Press [21].
Integration with Robotics & IoT
ATP is designed with potential applications extending beyond decentralized finance, including integration with robotics and Internet of Things (IoT) systems [1] [3].
The platform aims to enable AI agents to interact with the physical world by bridging AI, blockchain, and automation, unlocking new possibilities for decentralized intelligence [1]. Potential use cases mentioned include: