In today’s newsletter, First Digital’s Vincent Chok unpacks the rise of “agentic finance,” where AI agents move beyond advice to execute financial transactions, making crypto the essential financial backend for this machine-driven economy.
Then, in “Ask an Expert,” we asked three leading AI systems—Grok, Gemini, and Claude—two questions about AI payment usage and the steps needed for scalability.
Note: Answers were generated by AI assistants and reflect each model’s perspective. They should not be construed as financial or legal advice.
-Sarah Morton
AI agents in crypto: what advisors need to know
The explosive growth of AI agents
AI agents have become one of the hottest topics over the past year. A recent PwC survey of over 300 companies found that 79% already use AI agents in some form. This explosive growth reflects a broader shift: AI agents are evolving from advisory roles to execution roles.
Originally implemented to assist with chatbot services and copiloting roles, AI systems now actively plan, decide and act on predefined parameters set by humans, including financial transactions. The result is the early formation of “agent finance.” This is a new primitive where AI agents essentially perform economic actions within pre-defined rules such as limits, permissions and targets.
Breakdown of agent financing
Agent financing can be understood in three layers. The agentic trading layer focuses on discovery and decision making. For example, an AI agent can search for the best hotel deal for an upcoming trip. The agent payment layer handles execution, where the agent completes a transaction once it is approved.
Finally, the asset management layer represents the entire stack, where the agent can manage portfolios, handle payments and dynamically optimize financial strategies based on real-time market trends. While this may seem like we are giving AI agents full autonomy, this is not the case. It is conditional delegation where users retain control through constraints while offloading execution.
Theoretically, AI agents have a use case in the financial field; however, they do not fit neatly into existing traditional financial infrastructure. Structurally, AI agents lack direct access to global banking rails and are designed to operate 24/7. This structural mismatch is where crypto comes into play.
Stablecoins offer AI agents access to programmable, always-on money, blockchains enable instant and global settlement, and crypto-wallets provide permissionless access to funds. Essentially, these components form an economic layer better suited to machine-driven activity. Crypto is thus increasingly becoming the infrastructure for autonomous systems, rather than just being an asset class.
Use cases of AI agents
Early implementations are already visible. Machine-to-machine payments powered by API access and data providers have made intermediary rails stronger and faster. In the consumer context, autonomous commerce has allowed users to optimize retail research by using agents to get the best deals on travel, subscriptions and shopping.
Meanwhile, in crypto-native environments, trading agents are widely used for portfolio management, yield optimization and trading strategies. On the enterprise side, supply chain management and supplier payments have been easily automated via AI agents, reducing errors and resource consumption. At this stage, most activities remain business-to-business and infrastructure-driven, rather than consumer-facing.
Beyond use cases, AI agents also play an integral role in driving new investable categories as well as demand for crypto itself. Since AI agents cannot operate on existing infrastructure rails, the demand for agent-native wallets, stablecoin payment rails, and data or computer marketplaces is growing.
Coinbase, for example, has launched x402, an open payment protocol designed for agent-native transactions. This shift is particularly relevant for micropayments, where high transaction volumes and low value make traditional rails ineffective. For the first time, non-human users are participating in the financial system and driving activity. AI agents have become a new class of ‘user’ for crypto networks.
Risks and future prospects
Despite the momentum, we are still in the early stages and there are risks and limitations. Security is the primary concern, especially around rogue or exploited agents performing unintended transactions. Issues around authorization, liability and regulatory treatment are still under review and are being actively defined. For widespread adoption, we need to build trust with users. This comes through regulatory clarity from all stakeholders involved, so that projects can be built with clarity and confidence, while protecting users’ funds and interests.
Over the next twelve months, this technology will continue to grow and mature. Signals that matter include growth in agent-driven transaction volume, the rise of agent-native wallets and payment protocols, and deeper integration between stablecoins and AI-powered systems. Finally, regulatory clarity will greatly influence the pace and extent of adoption across different industries and areas.
In conclusion
AI agents are not a theoretical concept; they already perform transactions in restricted environments. As the trend develops, crypto is increasingly emerging as the financial backend for machine-driven economies. For now, this is an infrastructure and long-term thematic game; however, this is changing with increasing adoption rates. Advisors should track it as a next-wave driver of crypto tools.
– Vincent Chok, CEO and co-founder, First Digital
Ask an expert
This week we’re doing something a little different. Instead of one expert, we have a panel of experts – AI experts. Below, we posed two questions to three leading AI models, asking about the present and future of AI payments. While there were common themes – particularly about what is needed for future growth – there are also some clear differences. We hope you find this experiment as fun and thought-provoking as we have.
Q1: What AI payment use cases do you see today?
Grok xAI |
Fraud detection is the most mature use case – AI detects anomalies, rules-based systems miss. Intelligent payment routing dynamically selects the best railway according to price and speed. New agent payments let autonomous AI handle B2B treasury tasks and machine-to-machine micropayments using stablecoins, with Visa, PayPal and Mastercard starting to verify these agents. |
|---|---|
| Gemini Google | Three “Agentic Commerce” breakthroughs stand out: autonomous financial management and DeFi returns management, where AI performs cross-chain swaps in real time; “Inference Economy”, where agents settle machine-to-machine micropayments instantly via stablecoins; and hyper-personalized wealth implementation – AI-powered direct indexing that handles fractional trading and daily tax loss harvesting on a scale that no human advisor could handle manually. |
| Claude Antrop | AI payments are moving from concept to infrastructure faster than most advisors realize. Clearest use cases: programmatic billing for AI services and agents that autonomously spend funds on computing or subtasks without human approval per transaction. On the wealth side, AI enhances rebalancing and tax loss harvesting, where humans are still in the loop on most material decisions today. |
Q2: What will it take for AI payments to scale?
Grok xAI |
Three gaps must be closed: standardized agent identity with cryptographic authorization proof and clear rules of responsibility; governance and explanation options for regulators, including AML compliance in automated asset flows; and high-quality, real-time data infrastructure that bridges the gap between fiat and crypto. Advisors must demonstrate AI augments rather than replace fiduciary duties. |
|---|---|
| Gemini Google | Three things must converge: Verifiable agent identity using DIDs and cryptographically signed credentials so institutions can distinguish authorized AI from rogue bots; standardized management protocols with universal consumption limits and kill switches enforced at the protocol level; and legal liability frameworks that clarify who is liable when an AI executes a flawed trade or falls victim to a rapid injection attack. |
| Claude Antrop | Three things are really missing: identity, accountability and spending control. AI agents lack a standardized identity layer – that’s the fiduciary risk that hides inside all the excitement. Programmable barriers built into the payments layer (like a corporate card with hard limits and audit trails) are needed, along with regulatory clarity on stablecoins. The technology is largely ready. What’s lagging is governance – and that’s an opportunity for advisers to get ahead of it. |
