Exploring CARV's AI Chain Ecosystem

Key Insights

• CARV is developing an AI Chain ecosystem that aims to enhance the intelligence of crypto agents by enabling access to data from over 9 million CARV IDs, while strengthening privacy protections and value capture for users.
• In January 2025, CARV introduced the CARV SVM Chain and D.A.T.A. Framework, foundational components of its AI Chain ecosystem providing infrastructure and enhanced decision making capabilities for agents.
• CARV SVM Chain is a hybrid rollup on Ethereum, powered by a decoupled SVM. It aims to provide a high-performance environment for agents to operate by combining the scalability of Solana with Ethereum’s security.
• CARV ID powers the D.A.T.A. Framework by unifying user identity data from on-chain and off-chain sources. Data is enriched with tags and accessed by agents to facilitate deeper insights and more personalized interactions within the ecosystem.
• The D.A.T.A. Framework enables agents to AI agents to monitor user activity, trigger alerts, and execute transactions. Future upgrades will introduce additional features like memory sharing and risk monitoring that enhance their capabilities.

Primer

CARV was founded in 2022 by Chloe Gu and Victor Yu to give users greater control over their Web2 and Web3 identities and data, primarily in the gaming sector. That same year, CARV launched CARV Play, a gaming application that offers a unified interface to access multiple games and reward-earning activities. In 2023, CARV introduced CARV ID as an unique identifier that merges a user’s Web2 and Web3 identities into a single NFT-based profile based on the custom ERC-7231 standard. This identity solution powers CARV Protocol, which serves as a cross-chain decentralized identity and data layer enabling users to securely and privately manage, share, and monetize their digital presence.

In Q1 2025, CARV announced its transformation into an AI Chain ecosystem with the introduction of CARV SVM Chain and D.A.T.A. Framework. By integrating these advancements with its existing infrastructure and identity system, CARV aims to enable large-scale data sovereignty and accelerate progress toward artificial general intelligence.

Website / X (Twitter) / Discord / Telegram

Overview of the AI Chain Ecosystem

AI agents will drive the ecosystem by leveraging enriched user data from CARV ID to enhance decision-making capabilities and enable more personalized interactions. CARV’s purpose-built AI chain aims to ensure user data is handled privately and securely, while providing a dedicated environment for agents to operate.

Three core components power the ecosystem:

• CARV SVM Chain
• D.A.T.A. Framework
• Application Layer

CARV Labs will serve as the ecosystem’s accelerator and development arm. With $50 million in capital, it plans to catalyze growth through funding, incubation programs, and foundational infrastructure.

CARV SVM Chain

The CARV SVM Chain testnet launched on January 7, 2025. CARV SVM Chain is a hybrid rollup on Ethereum that combines Solana Virtual Machine (SVM) execution with zero-knowledge proofs for validity and faster withdrawals. It is built using SOON Stack, which provides a “decoupled” SVM capable of settling transactions on Ethereum.

Core features of CARV SVM Chain enabled by SOON Stack include:

• A Decoupled SVM offering 30,000 transactions per second, native fraud-proof mechanisms, and reduced data availability waste.
• Merkle trees in the SVM to enable state root verification, stateless execution, and inclusion proofs to enhance cross-chain interoperability and bridge security.
• Horizontal scaling through Transaction Processing Unit (TPU) distribution, Consumer-Producer architecture, and SIMD83 integration, allowing for theoretically unlimited growth.

CARV SVM Chain will integrate with CARV Protocol to manage the end-to-end data flow within the ecosystem. The protocol supports permissionless computation, verification, and data storage on the CARV SVM Chain. It consists of four modular layers, with the following two playing key roles within the AI Chain ecosystem:

• Computation and Training Layer: handles the processing and analysis of aggregated data used by agents to unlock insights and train models. It employs Trusted Execution Environments (TEEs) like Intel SGX and Zero-Knowledge proofs (zk-SNARK) for verification. Agents operating on CARV SVM Chain will execute tasks within TEEs, ensuring data is processed privately and securely.
• Verification Layer: comprises community-operated verifier nodes that validate Zero-Knowledge proofs and TEE attestations. Verifier nodes provide additional security for CARV SVM Chain by cross-checking these results to confirm the integrity of AI agent operations and transactions. This ensures that computations remain verifiable and tamper-resistant.

D.A.T.A. Framework

The D.A.T.A. Framework (Data Authentication, Trust, and Attestation) is a plugin for web3 agent frameworks designed to enhance AI agents’ capabilities through access to on-chain and off-chain data from CARV ID. It integrates with multiple chains—including CARV SVM Chain, Ethereum, and Solana—and supports both on-chain and off-chain storage solutions. The framework currently supports Eliza and Sentio, with plans to support additional web3 agent frameworks like Rig, G.A.M.E., ZerePy in the future.

Additionally, DeepSeek R1 is integrated into the framework to optimize agents’ cognitive architecture. DeepSeek can enhance agent decision-making through chain-of-thought processing for complex reasoning, multi-stage self-verification to improve decision accuracy, and reinforcement learning to enable agents to adapt strategies based on both historical and real-time blockchain interactions.

Integration with CARV ID

CARV ID is a unique identifier based on the custom ERC-7231 standard that merges a user's identities across multiple Web2 and Web3 platforms into a single NFT-based profile. With over nine million CARV IDs minted, its integration with the D.A.T.A. Framework can provide agents with a comprehensive understanding of various user behaviors and preferences. This aims to allow users who choose to share and monetize their data to receive more meaningful, personalized interactions while ensuring their data remains private and secure.

CARV ID data can also support the development of sophisticated, data-driven applications within the ecosystem. For example, gaming dApps could incorporate AI-powered non-player characters (NPCs) that evolve in real-time based on user data from CARV ID, fueling unique interactions and dialogue to create more immersive experiences for players. Additionally, reward distribution could be tailored to users based on historical engagement with the dApp and other platforms.

Contextual Tags and Metrics

CARV ID data is enriched with quantifiable metrics and descriptive tags that contextualize blockchain addresses and user activities. As a result, agents can recognize patterns and categorize users (i.e., distinguishing developers, whales, memecoin traders) more effectively, enabling them to generate deeper insights and make more informed decisions.

D.A.T.A. Query Generation Methods

The D.A.T.A. Framework provides a backend service that supports SQL, key-value, or vector-based queries to retrieve on-chain and off-chain data based on user input. The framework supports two methods for query generation:

• Backend-Generated Query: An AI agent receives a user’s natural language query and sends it to the backend, which analyzes the user query, creates the corresponding queries, executes them, and returns the results. This allows agents to fetch on-chain and off-chain data based on user input to trigger alerts and actions.
• AI Model-Generated Query: Instead of relying on the backend to parse user input, developers can use AI models to generate custom queries. The AI model receives user prompts, formulates the queries, and then sends them to the D.A.T.A. Framework backend for execution. This allows agents to customize or adapt their data retrieval strategies to meet demands for more advanced analytics and decision-making requirements.

In both methods, the framework executes the generated queries and delivers structured response data. AI agents can then use this data to extract deeper insights, trigger actions, and conduct blockchain analyses to inform decision-making.

Ecosystem DevelopmentApplications of CARV Token

CARV is an ERC-20 token that was initially launched on Base in October 2024. It has a maximum total supply of 1.00 billion tokens and will be deployed on CARV SVM Chain network to function as the governance token and native gas token. CARV will support the AI Chain ecosystem in several ways, including but not limited to (i) pairing with AI agent tokens in liquidity pools, (ii) being staked by agents for veCARV(s) to participate in governance, (iii) serving as payment for agents and projects to access gated CARV ID data.

AI Agent Capabilities

The D.A.T.A. Framework currently supports on-chain activity alerts, cross-chain insights, and actions like token transfers and airdrops. This enables D.A.T.A. Framework agents to detect cross-chain market opportunities in real time, trigger user alerts, and execute agent-to-human and agent-to-agent transactions based on predefined strategies.

The following additional features will be released soon to unlock the full range of capabilities for D.A.T.A. Framework agents:

February 2025

• Social Media Integration for Enhanced User Profiling
• Swarm of Agents for Modular Data Access

March 2025

• Memory Sharing Between AI Agents
• Centralized Data Repository for Knowledge Sharing
• Enhanced Querying with AI Model Integration

April 2025

• Real-Time Analytics for Blockchain Activities
• Customizable Data Fetching and Query Generation
• Risk and Compliance Monitoring

Closing Summary

CARV is establishing an AI Chain ecosystem that aims to advance crypto agent capabilities and expand opportunities for user-led data ownership and monetization. Its CARV SVM Chain features a Decoupled SVM on Ethereum, combining the scalability of Solana with Ethereum’s security. Additionally, the chain aims to ensure end-to-end data privacy and security for users through TEEs, zero-knowledge proofs, and verifier nodes. This infrastructure will support the decentralized ecosystem where AI agents operate, collaborate with other agents, and engage with dApps and end-users.
Furthermore, the D.A.T.A. Framework can provide agents with real-time access to enriched, structured user data from on-chain and off-chain sources to enhance their decision-making capabilities and enable more context-aware actions. As the D.A.T.A. Framework and CARV SVM Chain evolve, CARV is positioned at the intersection of AI innovation and the growing demand for data sovereignty.