AI Agents Take Control: The Rise of Agentic Finance in the Financial World
Last week was a convergence of seemingly unrelated events in the financial world.
Google Cloud launched Gemini Enterprise for Financial Services, an AI-powered platform that streamlines KYC, credit analysis, investment portfolio monitoring, bond issuance, and financial research. Two days later, MoonPay announced its integration with Solana's Kamino, allowing users to initiate loans and earn yields through interactive interfaces like Claude or ChatGPT.
The Dallas Federal Reserve published a study on the impact of real-time account transfers and AI-powered comparison tools on bank deposits' stickiness. The next day, the Bank for International Settlements (BIS) highlighted Project Agorá, which completed real-world testing with 28 financial institutions and central banks.
Each event individually could be categorized under a familiar news label: Google as enterprise AI, MoonPay as crypto finance, Dallas Fed as bank research, and Project Agorá as central bank research on next-generation cross-border payment infrastructure. However, when combined, they reveal a more significant pattern.
The author argues that the real question isn't when AI will truly change finance but rather when it will be allowed to make financial decisions on behalf of humans and actually move funds. This distinction may be larger than the difference between search engines and ChatGPT.
In recent months, companies like Robinhood, Ramp, and Rocket Money have been developing products that allow AI to manage finances within predefined boundaries. These products don't replace human judgment but rather augment it by automating routine tasks and providing recommendations.
The author believes that the key concept in Agentic Finance is not automation but authorization. Just as a company's employees are authorized to use their credit cards, AI agents should be granted permission to manage funds within specified parameters.
MoonPay's recent developments can be seen as part of this trend, as it has been building a suite of financial capabilities for AI agents, including identity and permission management, transaction execution, financial operations, and asset management. These developments are aimed at creating a platform that allows AI to manage assets and liabilities within predefined rules.
The author also discusses the concept of Project Agorá, which is not an AI project but rather an experiment by the BIS and International Financial Corporation to improve wholesale cross-border payments using programmable currencies. The project has successfully completed real-world testing with 28 financial institutions and central banks.
The author concludes that Agentic Finance will change the way money flows through the system, making it more efficient and automated. However, this also raises questions about who will control these flows and how they will be governed.