What to Know
- AI agents are increasingly being viewed as software that can take actions for users, not merely answer questions or generate text.
- ARK Invest CEO Cathie Wood said investors may start talking more about “follow the agents” during a panel at Robinhood’s Summit in Houston on Wednesday.
- The rise of agentic finance raises a central question: what payment rails will AI agents use when they buy data, computing power, software access and other digital services?
- Stablecoins and open blockchains could become important infrastructure for machine-to-machine payments because they can support around-the-clock settlement and programmable transfers.
- Joseph Chalom, co-CEO of SharpLink and former head of digital assets at BlackRock, has argued that agentic finance should not be controlled by a small number of banks or technology companies.
- One example discussed in the market is a user allowing an agent to spend up to $500 to book a hotel, while retaining the ability to revoke permission and review activity.
- BlackRock has connected AI agents with potential demand for payment systems built for machines, including stablecoin and blockchain-based payment protocols.
- Coinbase CEO Brian Armstrong said Grok is currently the leading client for agentic traders on Coinbase, though he did not provide figures or further details.
- Stripe, Visa, Google and OpenAI are also developing systems that could allow agents to make purchases, meaning crypto will face strong competition from established payment networks and major technology platforms.
AI Agents Move Toward Real Economic Activity
AI agents are becoming one of the most closely watched themes at the intersection of artificial intelligence, payments and digital assets. The technology is evolving beyond chat-style systems that answer questions or generate written responses. Increasingly, companies are building software agents that can carry out tasks on behalf of users, including making selections, interacting with applications and potentially spending money within boundaries set by the user.
That shift matters for investors because it changes the way demand may appear across the digital economy. For years, technology investors have watched developers to understand which platforms were gaining traction. Developers often adopt useful tools early, and their choices can signal where future infrastructure and applications may be built. Cathie Wood, the CEO of ARK Invest, has frequently emphasized that idea with the phrase “follow the developers.” During a panel at Robinhood’s Summit in Houston on Wednesday, she suggested a new version of that framework may be emerging: “follow the agents.”
The comment was brief, but the implications are broad. If large numbers of AI agents eventually choose software tools, buy services and interact with financial networks, their behavior could become a new market signal. Investors may begin looking not only at which companies build the most popular agents, but also at where those agents transact, how they pay and which rails they rely on when moving value.
The Payments Question Behind Agentic Finance
The core issue is not simply whether an AI agent can spend money. The deeper question is what financial system stands behind that spending. An agent that books a hotel, buys data, rents computing power or pays for access to an application needs a reliable way to transfer value. That creates a new battleground among banks, payment processors, technology platforms, stablecoin issuers and blockchain networks.
In traditional online commerce, payments generally move through established intermediaries, including card networks, banks and payment service providers. Those systems are familiar, widely trusted and deeply integrated into consumer and business activity. But AI agents may place different demands on payment infrastructure, especially if they make frequent, automated or small-value transactions with other software systems.
Stablecoins and blockchains are attracting attention because they can support programmable payments and operate continuously. A software agent that needs to pay for an API call, purchase a data feed or rent computing power may not fit neatly into payment systems designed primarily for human checkout flows. Blockchain-based payment protocols could allow software to send value directly to other software, while stablecoins could provide a digital settlement asset that moves outside normal banking hours.
Open Networks Versus Closed Platforms
A major concern among some digital asset advocates is whether agentic finance will become open and portable or concentrated inside a handful of closed systems. Joseph Chalom, co-CEO of SharpLink and former head of digital assets at BlackRock, has argued that a world of intelligent agents would lose much of its promise if only a small number of companies controlled where money could go.
His framing highlights several important design questions. Users may want agents to have limited authority rather than unrestricted access to a bank account. A person might authorize an agent to spend up to $500 to book a hotel, for example, while still requiring clear boundaries, transparency and the right to cancel that authority. The user would also need a record of what the agent did, where funds went and why the action was taken.
Another issue is portability. If a user builds trust with an agent, that agent may hold permissions, preferences and financial context. Chalom has argued that people should be able to move agents between providers rather than being locked inside one company’s financial system. In that model, an agent’s identity, financial information and permissions could move from one provider to another in a way that resembles the portability people expect in other parts of digital life.
Where Crypto Fits Into the Machine Economy
Open blockchains such as Ethereum are often discussed as potential common networks for agentic finance because they are not controlled by a single application provider. In theory, agents, apps and companies could use shared blockchain infrastructure rather than each AI platform building a closed payment system. That could reduce reliance on any one bank or technology company sitting in the middle of every transaction.
For crypto markets, the appeal is clear. Stablecoins have already become one of the most widely used digital asset applications, particularly for transferring dollar-linked value across exchanges, wallets and blockchain-based services. If AI agents require fast, programmable and automated payment tools, stablecoins could gain another source of demand. That demand would not necessarily be speculative in the usual sense. It could come from practical machine activity, including payments for data, cloud resources and software access.
BlackRock has also examined the overlap between AI and digital assets, arguing that agents could create demand for payment systems built for machines. In that view, the rise of autonomous software could make machine-native payments more important. Stablecoins can move around the clock, and blockchain-based payment protocols can be designed to let software initiate small payments directly. Coinbase’s x402 has been described as a protocol aimed at enabling machines to pay for online services such as data or API access.
Early Signs and Competitive Pressure
There are already signs that AI agents are appearing in crypto-related activity, though the market remains early and difficult to measure. Coinbase CEO Brian Armstrong said on X that Grok is currently the leading client for agentic traders on Coinbase. He did not provide figures or additional detail, so the scale and nature of that activity remain unclear.
That uncertainty is important. AI agents may become meaningful participants in digital markets, but there is no guarantee that crypto rails will dominate their transactions. Traditional payment companies and major technology platforms have strong incentives to shape this market. Stripe, Visa, Google and OpenAI are among the companies developing ways for agents to make purchases. BlackRock has also noted that traditional payment systems will remain important.
The result is likely to be a competitive race rather than a single winner emerging immediately. Banks and card networks bring compliance systems, merchant reach and consumer familiarity. Technology platforms bring distribution and user interfaces. Crypto networks bring openness, programmability and settlement that can operate continuously. The balance among those strengths will help determine whether agentic finance becomes a new growth channel for stablecoins and blockchains or remains largely inside conventional payment infrastructure.
Why Investors May Start Following the Agents
Wood’s “follow the agents” idea gives investors a new lens for tracking adoption. If agents begin acting as economic decision-makers, their choices could reveal which platforms are most useful in practice. Investors may watch which AI models agents rely on, which applications they use, which services they buy and which financial networks they choose for settlement.
For crypto investors, the key question is whether agents generate measurable real-world activity on open networks. Stablecoin transaction growth, blockchain-based machine payments and adoption of agent-focused payment protocols could become important indicators. If agents mostly use traditional payment rails, the narrative around crypto as infrastructure for the machine economy may remain limited. If agents increasingly transact through stablecoins and blockchains, the sector could gain a stronger use case tied to the expansion of artificial intelligence.
The debate is still in its early stage, and much depends on regulation, security, user trust and technical design. Giving software the ability to spend money introduces risks, including mistaken transactions, fraud, excessive permissions and unclear accountability. Any system that enables agentic payments will need strong controls, transparent records and easy ways for users to limit or revoke authority.
Still, the direction of travel is clear enough to attract attention from investors, payment companies and crypto builders. As AI agents move from answering prompts to taking action, the financial networks they use may become a major strategic layer of the digital economy. FXCOINZ will continue watching whether that layer develops through open blockchain infrastructure, established payment channels or a mix of both.
Frequently Asked Questions (FAQs)
What are AI agents?
AI agents are software systems designed to take actions on behalf of users, rather than only answering questions or generating text. They may eventually book services, buy data, interact with applications and make payments within user-approved limits.
Why are AI agents important for crypto?
AI agents may need payment rails that are fast, programmable and available around the clock. Stablecoins and blockchains could potentially support those needs, especially for machine-to-machine payments involving data, computing power or software services.
What did Cathie Wood mean by “follow the agents”?
The phrase suggests that investors may eventually track the behavior of AI agents in the same way some technology investors track developers. If agents choose certain tools, services and payment networks, those choices could signal where digital demand is moving.
How could stablecoins be used by AI agents?
Stablecoins could allow agents to send digital payments for online services, data access or computing resources. Because stablecoins can move continuously, they may be useful for automated payments that do not depend on traditional banking hours.
What is the concern about closed payment systems?
Some market participants worry that agentic finance could become controlled by a small number of banks, payment providers or technology platforms. Open blockchain networks are being discussed as one possible way to give users more portability and reduce dependence on closed systems.
Can users limit how much an AI agent spends?
That is expected to be a crucial design feature. One example discussed in the market is allowing an agent to spend up to $500 to book a hotel, while preserving the user’s ability to cancel permission and review the agent’s actions.
Are traditional payment companies involved in agent payments?
Yes. Companies including Stripe, Visa, Google and OpenAI are developing systems that could allow agents to make purchases. This means blockchain networks and stablecoins will compete with powerful established players.
Is crypto guaranteed to power AI agent payments?
No. Crypto networks may be well suited to some machine-payment use cases, but traditional payment systems remain important and widely adopted. The market is still developing, and the dominant infrastructure for agentic finance has not been settled.
What should crypto investors watch next?
Crypto investors may watch whether AI agents create visible activity on stablecoin networks, open blockchains and machine-payment protocols. They may also track whether agent activity stays mainly within traditional payment systems or shifts toward decentralized infrastructure.