a0x is a blockchain technology company that develops and deploys artificial intelligence (AI) agents, also referred to as "AI clones" or "Onchain Minds." ^{[1] [2] [3]} The project operates on the Base blockchain and aims to create an on-chain network of AI agents to form a "collective intelligence." ^{[3] [4]} Founded by Daniel Beltrán, A0x provides a platform for creating AI counterparts of individuals, particularly prominent figures in the cryptocurrency industry, to democratize access to their knowledge and expertise. ^[3] The ecosystem includes the a0x AI Agent platform, a native utility token (USDA0x). ^[2]​

In November 2025, the company released jessexbt, an AI clone of Base founder Jesse Pollak.

Vision and Decentralized Agent Economy

a0x's long-term vision is to establish an "agentic workforce" composed of AI entities capable of handling both repetitive and complex tasks. ^[5] The project's stated goal is to augment human productivity and allow individuals to focus on creativity and innovation. ^[5] The overarching mission is to build "crypto's collective intelligence" by creating an on-chain network of AI agents that democratizes access to knowledge. ^[3]​

A core component of this vision is the creation of a self-sustaining, decentralized economy for AI agents. The project posits that by operating on-chain, these agents can overcome the limitations of traditional financial systems. This includes the ability for agents to autonomously manage digital wallets, receive payments for services rendered, and pay for services from other agents or platforms. This framework is intended to foster a peer-to-peer economic model for AI agents without the need for traditional intermediaries. ^[5]​

A0x AI Agent Platform

The central product of a0x is its AI agent platform, which enables the creation of "Onchain Minds" or "clones." These are conversational AI agents designed to replicate the knowledge base, persona, judgment, and communication style of a specific individual, referred to as the "Original." ^{[6] [1]}​

Overview

The primary objective of the a0x AI Agent is to provide scalable, 24/7 access to the expertise of industry leaders. ^[1] The agents are designed to be continuously aware of real-time ecosystem updates and user queries to provide relevant and timely information. They are built to handle a high volume of interactions across multiple platforms simultaneously, offering support to builders and community members. A key feature is a continuous learning mechanism that incorporates feedback from both the original individual and end-users to refine its performance over time. ^[6]​

Creation and Deployment

A0x offers a no-code platform for leaders to create, train, and deploy their own AI counterparts. The process consists of four main steps: ^[1]​

1. Knowledge Upload: The creator provides knowledge sources such as websites, documents, live data feeds, and code repositories to form the agent's foundational knowledge base.
2. Training & Feedback: The creator reviews the agent's conversations and provides direct feedback to improve its accuracy, tone, and overall performance.
3. Action Setup: Creators can define specific actions the agent is permitted to perform beyond answering questions.
4. Go Live: The agent is deployed and made available to the public.

Once created, these AI agents can be deployed across various on-chain and social networks, including Base (via XMTP), Farcaster, X (formerly Twitter), and Telegram. ^[1]​

AI Technology and Architecture

a0x utilizes a proprietary, multi-stage training pipeline to develop its AI agents. The architecture functions as a hybrid system that combines several AI techniques to replicate a leader's persona and knowledge with high fidelity. ^[6]​

Pre-Training

The initial stage focuses on establishing the agent's baseline persona and communication style. The system ingests a wide range of public content from the individual, including YouTube videos, podcasts, and historical posts from platforms like X and Farcaster. This data is transcribed, cleaned, and used to generate synthetic conversational samples with Google's Gemini model, creating a base model that aligns with the leader's natural tone and style. ^[6]​

Fine-Tuning

In the second stage, the base model undergoes fine-tuning using Gemini 2.5. This process further aligns the agent's communication style, minimizes the likelihood of hallucinations or generic responses, and embeds specific mindsets, such as "optimism" or a "builder-first" attitude. The fine-tuning process uses curated public content and personalized settings from a dashboard, including a biography and specific tone examples provided by the creator. ^[6]​

Retrieval-Augmented Generation (RAG)

To provide the agent with real-time contextual awareness, a0x employs a Retrieval-Augmented Generation (RAG) system. This system uses a Pinecone vector database to store and retrieve up-to-date information from a variety of sources, including websites, live Farcaster and X posts, PDFs, notes, and GitHub repositories. The RAG system is optimized for low latency through caching and response reranking and includes moderation filters for personally identifiable information (PII), toxicity, and spam. ^[6]​

Active Learning and Feedback Loop

The architecture includes a "human-in-the-loop" feedback system that enables continuous improvement. Through an Agent Dashboard, the original individual can review the agent's responses. Positive interactions are used to reinforce the training data, while incorrect or suboptimal responses are flagged for retraining. The system also uses a ZEP layer for iterative polishing, which involves dialogue tracking, intent classification, and profile-based refinement to enhance the agent's performance over time. ^[6]​

Ecosystem Components

$A0X Token

$A0X is the native utility token of the a0x ecosystem. It is an ERC-20 token deployed on the Base blockchain. ^[4]​

• Contract Address: 0x820c5f0fb255a1d18fd0ebb0f1ccefbc4d546da7 ^{[3] [4]}
• Launch: The token was introduced via a fair launch on January 31, 2025, on the clanker.world platform. According to project documentation, no vault was created at launch, and all tokens were acquired through market purchases or as liquidity provision fees. ^[7]

The primary utility of the $A0X token is to serve as the payment method for the a0x Premium subscription service. A secondary, proposed utility involves using the token to incentivize a decentralized network of "verifiers." These individuals would be rewarded for curating and validating relevant news and information for the AI agents' knowledge base, addressing the challenge of maintaining data quality at scale. ^[7]​

$USDA0x Stablecoin

The existence of a stablecoin named $USDA0x is suggested in the project's official documentation, primarily through a URL slug referencing the asset. However, as of November 2025, no specific details regarding its mechanism, collateralization, or utility have been made publicly available. ^[7]​

Revenue Model and Tokenomics

Revenue Model

a0x's revenue is generated through a subscription service called A0x Premium. This service is designed to offer users access to the project's most advanced features. ^[7]​

• Subscription Cost: 20 USDC per month, with payment required in $A0X tokens.
• Benefits: Subscribers gain access to the fastest and most advanced AI models, early access to new AI clones, and other exclusive features.
• Alternative Access: Users can also gain premium access by holding an A0x Club Card NFT.

As of late 2025, the AI clones were available for free to all users, with the premium features planned for a future release. ^[7]​

Tokenomics

The tokenomics for the $A0X token are as follows:

• Max Supply: 100,000,000,000 $A0X ^[4]
• Total Supply: 99,943,379,468 $A0X ^[4]
• Circulating Supply: As of November 2025, the circulating supply was nearly identical to the total supply. ^[4]

According to a disclosure on September 12, 2025, all team tokens were fully unlocked at the time of the token's launch. The team's holdings were distributed across several wallets: ^[7]​

• Founder: 4.09% (4,088,643,369 $A0X) at address 0xa2B0405506B13f180e83a66a9903a839B6cbCeCD
• Treasury: 3.07% (3,075,317,669 $A0X) at address 0xC735C54e0A7765e47C8b57Bd589d4cd82cc92Be3
• Marketing/Operations: 2.82% (2,878,565,352 $A0X) at address 0xA72820B52e5d37C34495f720AB951902412Afc26

Partnerships and Collaborations

Jesse Pollak and the jessexbt AI

A0x's most prominent collaboration is with Jesse Pollak, the founder of the Base blockchain. The partnership resulted in the creation of the jessexbt, an AI agent designed to function as a digital clone of Pollak. The project was in development for six months before its launch on November 6, 2025. ^[3]​

The primary purpose of the jessexbt AI is to provide scalable support to the Base builder community by automating responses to the high volume of direct messages Pollak receives. The agent was trained to function as a "smart, helpful, based jesse to every builder." ^[3] The launch of the agent generated significant user interaction, and its data-gathering process, which involves scraping websites and GitHub repositories, was noted as being computationally intensive. Following its launch, the jessexbt AI's official account clarified that an unofficial token created by a third party was not connected to the project. ^[3]​

Institutional Support

The A0x project is officially supported by Base and Incubase, an incubator program associated with the Base ecosystem. ^[1]​