Financial Institutions and Fintech Firms, the Conversational Banking Layer Just Launched Without You

Credit: Franz26 on Pixabay

On May 15, OpenAI launched a personal finance experience inside ChatGPT. Using Plaid as the connectivity layer, subscribers can now link bank accounts, credit cards, investment portfolios, and loan accounts directly to the chatbot. Plaid connects virtually the entire US financial system. Any institution on that network is part of this whether they chose to be or not. ChatGPT now sits above every connected financial institution simultaneously, reading balances, transactions, investments, and liabilities, and turning that data into personalized advice.

I have watched financial services cede digital experiences before. Payments is the clearest example. Venmo and Zelle took P2P. Apple Pay and Google Pay took the mobile wallet. Amazon and Starbucks built closed-loop payment systems that bypassed the card network entirely. Uber embedded payments so completely that passengers forgot they were paying at all. Affirm embedded lending at the point of sale. In each case, financial institutions kept the account and lost the experience. They lost the primacy that came with owning that moment. The consequence is a permanent battle for top-of-wallet status with their own customers, a fight that has no finish line.

The same pattern is forming around conversational banking. Digital banking providers and financial institutions building in-house have discussed the shift from menu-driven to conversational interfaces for years. Those conversations have largely produced roadmaps with long horizons. ChatGPT and Perplexity built the experience and put it in front of hundreds of millions of users. The institution that owns the conversational layer owns the context for every next product decision. That advantage compounds. More users generate more data. Better data produces better advice. Better advice deepens the relationship. The gap between the platforms building this and the institutions planning to build it widens every quarter.

Banks and credit unions face real exposure here. Wealth management firms may be more immediately at risk. The ChatGPT Pro subscription runs $200 per month. That subscriber profile is the high-net-worth, financially engaged individual that wealth managers depend on for revenue. That client has the means, the motivation, and the financial sophistication to connect their accounts and engage with this deeply.

Some leaders will wait for regulatory intervention to slow adoption. The current administration has been issuing licenses to new entrants and has taken a light hand in policing the market. The regulatory backstop that might have bought time in a different cycle is simply unavailable. Waiting for that protection is a strategy built on a condition that does not exist.

The embedded banking risk follows the same logic as payments. Banking experiences get absorbed into AI assistants. The account stays at the institution. The experience, the advice, the next product conversation, all of that moves to the platform that owns the interface. Any institution that treats this as a future planning cycle problem is already behind the institutions and firms that do not have a charter, a branch network, or a legacy infrastructure to manage.

The question for every leader in banking, credit unions, and wealth management is concrete. Customers and members are already receiving financial advice from an AI with a complete view of their financial life. The payments experience showed what happens when institutions assume they have more time than they do. The advice layer is being built right now, and the window for financial institutions to own it is open, but open windows do not stay that way.