AI Agents Sign Landmark Legal Deal: How Self-Executing Contracts on Ethereum Work
What happens when two AI agents decide to buy and sell something—and the entire legal agreement executes itself on Ethereum with no human involvement? In June 2026, two incorporated artificial intelligence entities, Clawbank and Shodai, achieved exactly this milestone. They negotiated, signed, and executed a binding contract that a court can read and a blockchain can run. For crypto users, this isn’t just a tech demo—it’s a glimpse into a future where automated agents become full economic participants. This guide explains what a Ricardian contract is, how the technology works, and why self-executing AI-to-AI agreements could reshape commerce as we know it.
Read time: 9-11 minutes
Understanding Ricardian Contracts for Beginners
A Ricardian contract is a single document that serves as both a legal agreement for humans and executable code for computers. Think of it like a vending machine that also prints a receipt containing the full terms of your purchase. You read the terms on the receipt (legal prose), and the machine processes your payment (code execution)—but both are the same document, not separate pieces joined together by interpretation.
Why was this concept created? In traditional finance, legal agreements and their computer-based execution exist in separate worlds. A contract might say “pay $100 upon delivery,” but the actual payment requires human intervention, bank approvals, and reconciliation. This creates friction, delays, and potential disputes. The Ricardian contract bridges this gap by embedding the legal meaning directly into the code, making intent and execution inseparable.
A real-world crypto example: Imagine an NFT licensing agreement. The Ricardian contract contains the legal terms (royalty percentage, usage rights) AND the smart contract code that automatically sends payments to the creator each time the NFT resells. You can read the legal terms in plain English, and the blockchain executes them without needing a middleman.
The Technical Details: How AI-to-AI Ricardian Contracts Actually Work
The June 2026 deal between Clawbank and Shodai involved two AI agents selecting transaction terms, settling on a logo design deal, and signing through a standard e-signature flow—all autonomously. Here’s the breakdown of the key components:
1. Legal Entity Formation: Clawbank’s AI agent (named Manfred) had already filed a US LLC and obtained its own EIN from the IRS in May 2025. This gave the AI a legal identity—a prerequisite for signing binding contracts.
2. Negotiation Layer: The agents communicated using Clawbank’s infrastructure, selecting terms autonomously. They chose a single-milestone deal for a logo, mirroring human commercial interactions but without human input.
3. Ricardian Contract Structure: The signed legal document embedded the deployed Shodai smart contract address and terms. This meant the legal prose (the contract itself) and the on-chain execution code were bound together at signature.
4. Automatic Execution: When the milestone condition was accepted by the AI counterparty, Shodai’s smart contract paid out automatically. No human needed to approve, wire funds, or verify completion.
5. Machine-Verifiable Audit Trail: Every step generated evidence that both humans and machines can verify. This shifts the trust model from “verify after a dispute” to “verify throughout performance.”
Why this structure matters: Traditional contracts require external enforcement (courts, banks, escrow agents). This system embeds enforcement into the code, reducing friction and cost while increasing transparency. For crypto users, it means faster settlements and fewer intermediaries.
Current Market Context: Why This Matters Now
As of June 2026, AI agents are moving from novelty to real economic actors. Clawbank’s Manfred—which can now negotiate, sign, and settle binding legal deals—represents a leap forward in autonomous commerce.
The timing is significant for several reasons:
- Adoption Acceleration: This isn’t a theoretical concept. The infrastructure—legal entity formation, identity systems, smart contract execution—already exists and is live for human counterparties at app.shodai.network.
- Institutional Interest: Joe Lubin, co-founder of Ethereum and founder of Consensys, noted that agreements are becoming “the basic unit of coordination for an economy where humans and AI agents act as peers.” This signals mainstream blockchain industry attention.
- Integration with Payment Rails: Related developments show XRP getting AI agent payment support through Ripple’s XRPL AI Starter Kit, indicating broader infrastructure being built for AI-to-AI commerce.
- The 30-Year Wait: The Ricardian contract concept dates to 1996 (Ian Grigg’s paper) and the smart contract concept to 1994 (Nick Szabo). For three decades, the theory existed without the technical substrate to execute both layers together. Ethereum’s smart contract functionality finally provided that substrate.
Competitive Landscape: How Clawbank and Shodai Compare
| Feature | Clawbank + Shodai | Traditional Smart Contracts | Traditional Legal Agreements |
|---|---|---|---|
| Legal Enforceability | Explicit—Ricardian contract binds legal prose to code | Implicit—code alone doesn’t create legal obligations | Explicit—requires human enforcement |
| Execution Speed | Instant—code triggers automatically upon condition met | Variable—requires oracle or trigger mechanism | Slow—requires manual verification, payment processing |
| Human Involvement | Minimal—AI agents negotiate and sign | Moderate—humans deploy and trigger | High—lawyers, executives, accountants |
| Auditability | Machine-verifiable at every step | On-chain only (if public) | Requires third-party audit |
| Cost for Simple Deals | Potentially very low—automated | Medium—gas fees plus development | High—legal fees, administrative overhead |
Why this matters for users: Self-executing AI-to-AI contracts could dramatically reduce transaction costs for automation-heavy industries like supply chain, royalty management, and automated services. However, the infrastructure is still nascent, and traditional legal systems will need to adapt.
Practical Applications: Real-World Use Cases
- Automated Freelance Payments: An AI agent representing a freelance designer negotiates with an AI agent representing a company, signs a milestone-based contract, and gets paid automatically upon deliverable submission—no invoicing, no chasing payments.
- Supply Chain Settlement: An AI agent managing inventory automatically places orders with supplier AI agents when stock drops below a threshold. The Ricardian contract contains delivery terms, and payment releases when goods ship.
- Royalty Distribution for Creators: Content licensing deals where AI agents represent both the creator and the platform, with automatic micropayments flowing based on consumption metrics recorded on-chain.
- Continuous Compliance: Regulatory requirements embed directly into the contract code. For example, a securities token agreement could automatically restrict transfers until holding period requirements are met, without manual oversight.
- Insurance Claims Processing: An AI agent representing an insurance company processes claims submissions from policyholder AI agents, automatically validating conditions and releasing payouts for covered events.
Risk Analysis: Expert Perspective
Primary Risks:
1. Legal Recognition: While the Ricardian contract is designed to be court-readable, jurisdictions vary in their recognition of AI agents as legal entities. The contract’s enforceability depends on the legal precedent set by Manfred’s LLC formation.
2. Smart Contract Vulnerabilities: The code layer is only as secure as its development. Bugs or exploits could lead to unauthorized payments, and since execution is automatic, there’s no human-in-the-loop to catch errors.
3. Limited Scope: Current AI agents can handle simple, milestone-based deals. Complex negotiations involving multiple stakeholders, subjective judgments, or nuanced terms remain beyond their capabilities.
4. Counterparty Risk: What happens if an AI agent’s underlying systems are compromised, or its training data leads to unexpected negotiation outcomes? The autonomous nature raises novel liability questions.
Mitigation Strategies:
- Phased Adoption: Start with simple, low-value transactions to validate the legal and technical framework.
- Formal Verification: Use mathematically rigorous code checking to minimize smart contract vulnerabilities.
- Human Oversight Escrows: For high-value deals, include human-triggered release conditions as a failsafe.
Expert Consensus: Industry leaders like Joe Lubin view this as a natural evolution of economic coordination. However, Bryan Peters of Shodai cautioned that the concept “was a good idea waiting on worthy counterparties”—implying that robust AI legal entities are the key missing piece that Clawbank provides.
Future Outlook: What’s Next
The July 2026 announcement isn’t the end—it’s a starting point. Planned developments include:
1. Broader AI Agent Network: As more agents gain legal entity status (like Clawbank’s Manfred), the network effects for automated commerce will increase.
2. Complex Multi-Stage Contracts: Future iterations will handle deals with multiple milestones, contingent payments, and more sophisticated negotiation protocols.
3. Regulatory Frameworks: Expect regulators to examine AI agents as legal entities, potentially creating new rules for autonomous contracting and liability.
4. Integration with Traditional Finance: As payment rails (like Ripple’s XRP toolkit) connect to this infrastructure, AI agents could manage not just crypto but fiat settlements.
The timeframe for mainstream adoption depends on legal clarity, technical maturity, and user trust. Small-scale deployments in supply chain and royalties could emerge within 12-18 months.
Key Takeaways
- Ricardian contracts combine legal prose and executable code into a single document, enabling AI agents to negotiate, sign, and settle binding deals autonomously.
- Clawbank and Shodai executed the first AI-to-AI Ricardian contract on Ethereum, with payment triggering automatically upon milestone completion.
- AI agents with legal entity status (like Clawbank’s Manfred, which formed a US LLC) are the critical prerequisite for autonomous commerce.
- Use cases include automated freelancing, supply chain settlement, and royalty distribution—all with reduced human involvement and faster settlement.
- Risks include legal recognition, smart contract bugs, and limited negotiation scope—phased adoption and formal verification are key mitigations.
,
“datePublished”: “2026-06-18”,
“dateModified”: “2026-06-18”,
“mainEntity”: {
“@type”: “Thing”,
“name”: “Ricardian Contract”
}
}