DEX in the City: Why AI Agents Are Good for Crypto and Stablecoins

What It Covers

Edward Woodford, CEO of ZeroHash, discusses the crypto market structure bill Clarity, the White House meeting with banking and crypto executives, stablecoin growth post-GENIUS Act, and the convergence of AI agents with crypto rails. The conversation covers ZeroHash's decision to remain independent after acquisition discussions with Mastercard.

Key Questions Answered

• •Legislative Strategy: Focus on narrow, achievable reforms rather than comprehensive bills. Priority issues include updating security definitions to prevent regulation by enforcement, eliminating 1099 requirements for stablecoin transactions over $10,000, and addressing accounting treatment. Broad bills create vague language requiring years of rulemaking that agencies can interpret unfavorably, especially post-midterm elections when political dynamics shift.
• •Stablecoin Fragmentation: GENIUS Act creates massive fragmentation with multiple issuers and distribution channels. USDC now issues natively on 30 chains plus wrapped tokens and USDC-x constructs. This complexity increases value for infrastructure providers that abstract away multi-chain, multi-issuer complexity. Stablecoin-enabled accounts become the critical metric, measuring seamless access points rather than just transaction volume.
• •Banking Convergence: Traditional financial services companies totaling trillions in market cap will enter crypto within two to three years. Robinhood and Coinbase financials will look nearly identical as competitive threats force convergence. Morgan Stanley already offers Bitcoin, Ethereum, and Solana trading through ZeroHash infrastructure, demonstrating banks can operate with current clarity without comprehensive legislation.
• •Agent Commerce Infrastructure: AI agents require programmable money for billions of micro-transactions in agent-to-agent data transfers. Agents can pay for computational persistence across servers using crypto, creating operational freedom from centralized shutdown. Policy controls can be cryptographically embedded, requiring human or secondary agent approval beyond spending thresholds, enabling controlled autonomy rather than unrestricted operation.
• •AI Liability Framework: AI agents must roll up to natural or non-natural persons from KYC and monetary liability perspectives, similar to code liability where firms remain responsible for rogue software. This creates accountability chains but raises questions when agents self-improve, form autonomous communities, or engage in unintended commerce, requiring new frameworks beyond traditional employee scope-of-duty defenses.