The Agentic Shift: Why Crypto Infrastructure is Rebuilding for Autonomous Commerce

The Rise of the Machine Counterparty

For most of the last decade, crypto trading has been a human activity augmented by static code. While algorithmic trading dominates institutional flows, the retail experience has remained manual, tethered to buttons and screens. That paradigm is now shifting toward “agentic commerce”—a model where autonomous AI agents do not just assist users but act on their behalf within decentralized networks [20].

This week, the industry witnessed a synchronized pivot among retail giants. Robinhood announced that AI agents will “soon” be assisting its crypto traders, following a beta phase that saw 70,000 agentic accounts created for equities [15]. Simultaneously, Kraken and Revolut unveiled overhauls to their flagship applications, placing autonomous agents at the center of the user experience [34][59]. These are not mere chatbots; they are designed to monitor markets, backtest strategies, and execute trades without immediate human intervention.

Solving the Intent-to-Execution Gap

The technological hurdle in agentic commerce is not just intelligence, but trust. When a human authorizes a transaction, the intent is clear. When an AI agent triggers a trade based on a prompt or a perceived market signal, the line between an authorized action and a computational error—or a “hallucination”—becomes blurred. Researchers have already warned that these agents could be turned into botnets via hallucinations [82].

To mitigate these risks, a new layer of infrastructure is emerging to provide a “safety net” for machine-driven finance. The Genlayer Foundation recently formed a 27-firm consortium, backed by OKX and MetaMask, to develop a decentralized dispute resolution court specifically for AI agents [49]. This infrastructure aims to make AI-based payments, escrow, and dispute resolution interoperable across different networks. It represents an admission that existing legal and technical rails are insufficient for the speed and autonomy of AI-to-AI transactions.

Where Regulation Meets Autonomy

How do decentralized networks interact with governments when the primary actor is an agent? Regulators are currently ill-equipped to handle the “attribution problem”—the difficulty of assigning legal liability to an autonomous software agent. This is where the state still matters in digital finance: the legal definitions of a “broker” or an “exchange” are being tested by platforms like Hyperliquid and Phantom, which are urging the CFTC to modernize rules for on-chain protocols that act as impartial venues for these agents [96][107].

If the SEC or CFTC continues to apply legacy frameworks to autonomous agents, the result could be a fractured market where agentic commerce is siloed within offshore or highly restricted environments. However, the move by established players like Kraken to integrate these tools suggests a bet that US and European regulators will eventually provide a pathway for regulated AI-driven finance [21][34].

Synthesis: Infrastructure Beneath the Surface

The most critical infrastructure changes are happening beneath the UI. It is no longer enough to have a fast blockchain; the network must now support “delegated signing” and “intent-based” architectures that allow agents to operate securely without exposing private keys. MoonPay’s introduction of “MoonAgents” on Telegram illustrates this trend, allowing users to prepare transactions via AI while keeping keys on their own devices [123].

This trend represents a fundamental change in the movement of value. We are moving from a world of passive assets to a world of active, autonomous capital. The success of this transition depends not on the cleverness of the LLMs used by Revolut or Robinhood, but on the robustness of the secondary layers—the dispute courts, the intent-signers, and the regulatory frameworks—that allow humans to safely hand over the keys to the machines.

Sources

• [15] https://cointelegraph.com/news/robinhood-says-ai-agents-will-be-trading-for-crypto-users-soon
• [20] https://www.coindesk.com/coindesk-news/2026/07/10/meta-s-chief-data-officer-says-agentic-commerce-is-the-next-tier-of-business
• [21] https://cointelegraph.com/news/kraken-to-rebuild-app-around-ai-powered-trading-agents-report
• [34] https://www.theblock.co/post/407899/kraken-to-relaunch-its-mobile-app-with-agentic-trading-and-advise-at-the-center
• [49] https://www.coindesk.com/business/2026/07/10/okx-metamask-matter-labs-back-dispute-resolution-court-for-ai-agents
• [59] https://www.theblock.co/post/407865/revolut-integrates-its-crypto-exchange-with-ai-assistants-as-agentic-trading-spreads
• [82] https://decrypt.co/373196/ai-agents-botnets-hallucinations-researchers-warn
• [96] https://cointelegraph.com/news/phantom-hyperliquid-ask-cftc-to-modernize-rules-for-onchain-derivatives
• [107] https://www.theblock.co/post/407780/hyperliquid-policy-center-phantom-urge-cftc-to-stop-treating-onchain-protocols-like-traditional-brokers-and-exchanges
• [123] https://decrypt.co/373059/moonpay-ai-crypto-agents-telegram