Pacific Meta, Inc. (Minato-ku, Tokyo; CEO: Shota Iwasaki) has developed “AutoFund,” an asset management product that uses AI agents to autonomously execute transactions on the blockchain, and has begun test operations using company funds.

As part of this initiative, the company is exploring potential collaborations with financial institutions and asset management firms to address regulatory, audit, and operational design considerations.

[Note] This announcement is not intended as an investment solicitation or investment advisory service. Regulatory and legal considerations related to this initiative are being reviewed and organized in consultation with external experts.

Overview of AutoFund

AutoFund is an asset management product in which AI references on-chain data and other sources, then makes rebalancing decisions and executes trades end-to-end based on market conditions. In the test operations, the company allocates assets for operations to AutoFund, where the AI agent references on-chain data and executes asset rebalancing in response to market conditions.

Transaction targets, conditions, and maximum amounts are pre-configured, and the AI’s execution is limited to within these parameters. This design allows the company to verify whether it is possible to automate decision-making and execution while maintaining controlled operations. Future service models will be considered based on the progress of regulatory and legal reviews.

Why “AI × Blockchain” Asset Management Matters

1. Why Blockchain Is Essential for AI-Driven Asset Management

On the blockchain, AI can directly move assets through programs. In traditional finance, humans needed to go through banks and securities firms, but blockchain removes this constraint, enabling automation from decision-making to execution. Furthermore, all transactions are recorded in a tamper-proof manner, and anyone can verify the AI’s operational behavior. AutoFund adopts blockchain to establish “autonomous AI execution,” “transparency,” and “permission controls” as a unified system.

2. The Growth of On-Chain Data Continuously Improves AI’s Operational Accuracy

All transaction data on the blockchain is public and structured data that AI can read directly. As more activity moves on-chain, the volume of data available for AI learning increases, improving the accuracy of operational decisions. In traditional financial markets, transaction data was siloed within each company’s internal systems, but on the blockchain, it is publicly available worldwide — a structural advantage for AI-driven asset management.

3. Humans Alone Cannot Monitor 24/7/365 Markets

On-chain financial markets operate around the clock globally. It is physically impossible for humans to simultaneously monitor hundreds of investment targets and continuously move assets at optimal timing. Autonomous monitoring, decision-making, and execution by AI agents has the potential to make operational capabilities — previously available only to specialist teams at major financial institutions — accessible to a broader audience.

AutoFund’s Technical Architecture

Safe (Smart Contract Wallet)

A wallet for storing and moving assets used in test operations.

Allowance Module (Safety Mechanism)

Sets on-chain operational limits per time period, ensuring transactions are executed only within the permitted range.

AI Agent (Operator)

The AI agent references real-time data, performs comparative analysis across multiple investment targets, and within constraint parameters, makes condition-based decisions and automatically executes rebalancing.

This AI agent is built on Coinbase Developer Platform (CDP) AgentKit.

By utilizing CDP Managed Wallet, the company has adopted a configuration where private keys are not held directly on its servers.

Operational Flow and Feature Overview

Note: All screen captures, videos, wallet addresses, amounts, and transaction data shown in this article are demo data and do not represent actual operational results.

Note: The following features are continuously being updated. Features may be added, changed, or consolidated.

1. Flow from Setup to Launch (Envisioned for Future Deployment)

1.1. Setting the Investment Policy — Instructing the AI on “How to Manage”

AutoFund’s greatest feature is the ability to configure investment policies in detail beforehand. Through a guided UI, users can intuitively customize the following items:

• Investment objective — Choose from 3 levels: liquidity-focused, stable operations, or growth-focused
• Currency policy — JPY only / USD allowed / currency-agnostic
• Diversification — Concentrated / moderately diversified / maximum diversification
• Protocol selection — Select or exclude protocols for operations, such as Aave V3, Compound V3, Morpho, etc. (Future expansion to other on-chain assets and market data is anticipated)
• Token selection — Specify tokens for operations such as USDC, WETH, DAI, etc.
• Chain selection — Choose chains for operations: Ethereum, Base, Arbitrum, Optimism, etc.
• Risk tolerance — Set on a 5-level scale from conservative to aggressive
• KYC pool access restriction — A feature to limit access to only KYC-verified pools is also planned, depending on compliance requirements
• Spending Limit (daily cap) — Set the maximum daily amount the AI can operate, enforced on-chain

The AI agent makes investment decisions based on the configured policy. Additionally, an “Let AI Handle Everything” button allows full delegation of policy setting, rebalancing, and risk management to the AI.

Note: Available protocols, tokens, chains, and other options may be re-evaluated based on financial regulations and legal considerations in each jurisdiction when expanding services to other companies.

1.2. Deposit and Start Operations — From Wallet Connection to Launch

Follow the on-screen instructions to easily start operations. Deposited assets are managed within the Safe, and no other assets are touched.

1.3. AI Rebalancing Proposals and Approval Flow — All Past Proposals Are Accumulated, Creating a Decision Audit Trail

The AI agent proposes rebalancing based on market analysis. A multi-AI debate system is implemented, where multiple AIs (analytical AI, optimistic AI, pessimistic AI, and decision AI) discuss before reaching a final decision.

All proposals are accumulated as history, recording the actual revenue impact of both “approved proposals” and “rejected proposals.” Furthermore, What-If simulations — “What would revenue have been if that proposal had been rejected?” or “What if it had been approved?” — are automatically calculated and visualized. Because operational decisions are preserved as data, they can be used to verify AI performance and improve investment strategies.

2. AI Chat — Recording and Querying Operational Discussions, and Assisting with Configuration

The AI chat feature allows real-time querying of operational logs, configuration details, and execution history in natural language during the company’s test operations. This enhances operational transparency (explainability) and is utilized for verification and improvement.

(Note: This feature is not intended as investment advice or solicitation.)

3. Multi-Sig Support — Supporting Multi-Person Operations

AutoFund also supports “multi-sig (multi-signature)” functionality requiring approvals from multiple parties. For example, operations such as “large transactions cannot be executed without approval from both the CEO and CFO” are possible. This is designed for corporate operations and joint management with CFOs and auditors.

4. Dashboard — Real-Time Visualization of Operational Status

Displays deposited asset balances, real-time yields for each investment target, and security status at a glance. By connecting a wallet, users can check actual operational balances and status in real time. Active positions, cumulative P&L, and AI execution history are all visible on the dashboard.

Product Positioning and Future Outlook

AutoFund is currently in the test operations phase using company funds. Technical verification on the Sepolia testnet has been completed, and the transition to mainnet will proceed in stages.

Pacific Meta aims to build an Operating System (OS) for corporate management in the AI × Blockchain era. AutoFund is part of this vision, demonstrating the practical potential of AI agents in on-chain asset management through in-house implementation.

Roadmap

Short-term (In-house Operations Phase): For the time being, the company will use internal operations to address technical considerations (smart contract security, AI decision accuracy) and legal considerations (compliance with financial regulations).

Mid-term (Service Expansion to Other Companies): Before expanding services to other companies, all legal considerations — including compliance with financial regulations, applicability to investment management and investment advisory businesses, and relationship with crypto-asset exchange regulations — will be fully resolved. Joint business development with asset management firms is also within scope. Additionally, the company anticipates the growth of on-chain financial services utilizing smart contracts in Japan and envisions enabling access to these new financial services through AutoFund.

Long-term Vision: Going forward, assets beyond crypto — including equities, real estate, and commodities — are expected to move on-chain. The ideal is to evolve AutoFund into an AI-powered asset management product that supports all asset classes and serves every client.

Toward Joint Exploration with Financial Institutions and Asset Management Firms

In the world of finance, trust is paramount. Pacific Meta possesses technical expertise in blockchain and AI, but to truly deliver this vision to society, the company believes that financial industry knowledge, credibility, and regulatory compliance know-how are essential. By combining our technology with the expertise of financial professionals, we believe there are outcomes that can only be achieved together.

Based on AutoFund’s technical foundation, we are seeking partners to jointly explore the practical implementation of AI agent-driven on-chain asset management through collaboration and technology provision.

We envision partnerships with the following types of organizations:

• Asset Management Companies — Joint exploration of governance design (permission controls, limit design, audit design) and operational process verification for AI agent-driven on-chain operations
• Securities Firms / Trust Banks — Organizing regulatory and compliance considerations, exploring monitoring/audit frameworks, and jointly examining future service models
• Fintech Companies — Implementation and operational design around wallets/custody/AI agents, and integration testing with existing business workflows

Inquiries from the financial industry are especially welcome.

Contact: https://pacific-meta.co.jp/en/contact

Development Background — Comment from CEO Shota Iwasaki

“As someone who regularly uses blockchain financial services, I was constantly researching lending protocol interest rates and DEX liquidity provision yields, manually rebalancing across multiple chains every day. Frankly, it was quite challenging. That’s when I started thinking, ‘This should be left to AI.’

Pacific Meta has been managing part of its corporate assets in crypto since 2025. From these real-world experiences emerged the concept of AutoFund — where an AI agent autonomously executes rebalancing while humans only set risk limits.

All financial data on the blockchain is public, structured data that AI can read directly. Because of this transparency, I believe AI-driven asset management can continuously improve in accuracy through verification. If we can extend these benefits not only to institutional investors but also to ordinary individuals, this could be a step toward the democratization of finance.”

About Pacific Meta, Inc.

Pacific Meta, Inc. provides comprehensive support for blockchain business launches and operations, from business strategy development and token design to fundraising and marketing. Within three years of founding, the group has supported over 260 projects cumulatively, working with projects across 41+ countries as a global team that supports business expansion from both Japan and overseas. To help blockchain and token-powered businesses reach social implementation quickly, the company provides end-to-end support across talent, know-how, networks, and investment.

Company Overview

Address: HAMAMATSUCHO PREX 8F, 2-2-12 Shiba, Minato-ku, Tokyo

Representative: Shota Iwasaki, CEO

Founded: August 10, 2022

Business: Blockchain Accelerator

Capital: 630 million yen (including capital reserves)

Website: https://pacific-meta.co.jp/en/

Contact

Pacific Meta, Inc. — PR Department

Contact form: https://pacific-meta.co.jp/en/contact/

This press release is for informational purposes only and is not intended as an investment solicitation or advisory service. It does not suggest or guarantee future price movements of crypto assets or related products.