what advisors need to know

In today’s newsletter, Vincent Chok from First Digital unpacks the rise of “agentic finance,” where AI agents are moving beyond advice to execute financial transactions, making crypto the essential financial backend for this machine-driven economy.

Then, in “Ask an Expert,” we posed two questions to three leading AI systems — Grok, Gemini, and Claude — about AI payment use cases and the necessary steps for scalability.

Note: Responses 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 most trending topics over the last year. A recent PwC survey of over 300 companies found that 79% are already adopting AI agents in some form. This explosive growth reflects a broader shift: AI agents are evolving from advisory roles to execution roles.

Initially deployed to help with chatbot services and copiloting roles, AI systems are now actively planning, deciding and acting on predefined parameters set by humans, including financial transactions. The result is the early formation of “agentic finance.” This is a new primitive wherein AI agents essentially execute financial actions within predefined rules such as limits, permissions and goals.

Breaking down agentic finance

Agentic finance can be understood in three layers. The agentic commerce layer focuses on discovery and decision-making. For example, an AI agent can search for the best hotel deal for an upcoming trip. The agentic payments layer handles execution, where the agent completes a transaction once approved.

Finally, the asset management layer represents the full stack, where the agent can manage portfolios, handle payments and dynamically optimize financial strategies based on real-time market trends. While this may seem as if we are giving AI agents full autonomy, that is not the case. It’s conditional delegation, wherein users retain control through constraints while offloading execution.

Theoretically, AI agents do have a use case in the financial space; however, they don’t neatly fit in with 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 a financial layer that is better suited to machine-driven activity. Crypto is thus increasingly becoming the infrastructure for autonomous systems, rather than only 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 the inter-merchant rails stronger and faster. In the consumer context, autonomous commerce has allowed users to optimize retail research, using agents to get the best deals for travel, subscriptions and shopping.

Meanwhile, in crypto-native environments, trading agents are widely deployed for portfolio management, yield optimization and trading strategies. On the enterprise side, supply chain management and vendor payments have been easily automated via AI agents, cutting down on errors and resource expenditure. At this stage, most activity remains business-to-business and infrastructure-driven, rather than consumer-facing.

Beyond use cases, AI agents also play an integral part in driving new investable categories as well as demand for crypto itself. As AI agents can’t operate on existing infrastructure rails, demand is growing for agent-native wallets, stablecoin payment rails and data or compute marketplaces.

Coinbase, for example, has launched x402, an open payments protocol designed for agent-native transactions. This shift is particularly relevant for micropayments, where high transaction volumes and low value make traditional rails inefficient. 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 outlook

Despite the momentum, we are still in the early stages, and there are risks and limitations. Security is the primary concern, particularly around rogue or exploited agents executing unintended transactions. Questions around authorisation, liability and regulatory treatment are still under scrutiny and are being actively defined. For widespread adoption, we must build trust for users. This comes through regulatory clarity from all involved stakeholders, so projects can build with clarity and confidence while safeguarding user 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, emergence of agent-native wallets and payments protocols, and deeper integration between stablecoins and AI-driven systems. Finally, regulatory clarity will heavily shape the pace and scope of adoption across different industries and fields.

In conclusion

AI agents are not a theoretical concept; they are already executing transactions in limited 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 play; however, that is changing with rising adoption rates. Advisors should track it as a next-wave driver of crypto utility.

– Vincent Chok, CEO and co-founder, First Digital

Ask an Expert

This week we’re doing something a bit 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 — especially about what’s 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 are you seeing today?

Q2: What’s needed for AI payments to scale?

Keep Reading