By Futurist Teresa Grobecker and Futurist Thomas Frey
There is $23 trillion sitting in checking accounts around the world right now, earning almost nothing.
Not because the people who own it can’t do better. Not because better options don’t exist. The top-of-market savings rate in the United States is currently more than ten times the national average rate paid on savings accounts. That gap is worth thousands of dollars a year to an ordinary household — money that is simply not being collected because moving it requires effort, attention, and the willingness to interrupt the daily inertia of life.
Banks have known this for decades. They have quietly built their profit models around it. The financial term for this phenomenon is “customer inertia,” and it is one of the most valuable and least examined assets on any bank’s balance sheet. Trillions of dollars sit in low-yield accounts not because customers are satisfied, but because friction is a more powerful force than rationality when it comes to financial behavior.
That friction is about to be engineered away. And when it is, nothing about the economics of retail banking will be the same.
AI agents are about to eliminate customer inertia—automatically moving money toward better returns and potentially reshaping the economics of global banking overnight.
The $23 Trillion Problem — And Opportunity
McKinsey’s Global Banking Annual Review 2025 made a finding that every banking executive should be reading slowly and carefully: if just 5 to 10 percent of the world’s low-yield checking balances migrate to top-of-market rates — prompted by AI agents acting on consumers’ behalf — total deposit profits across the global banking industry could decline by 20 percent or more.
Let that settle. A single-digit percentage shift in one category of deposit behavior could remove one-fifth of the banking industry’s deposit profitability. The mechanism driving this shift is not a new regulation, a market crash, or a competitor stealing customers. It is simply an AI agent doing what a good financial advisor would do: noticing an obvious inefficiency and fixing it.
“Imagine you have an AI agent that says: ‘Hey, you could save $2,000 a year by moving your money,'” McKinsey senior partner Pradip Patiath said in October 2025. Then the agent doesn’t just say it. It offers to move the money. Right now. In the background. While you continue with your day.
This is not a hypothetical. The infrastructure for exactly this kind of autonomous financial action is being built and deployed today by the same companies that sit at the heart of global commerce. Visa has launched its Trusted Agent Protocol — a framework enabling AI agents to search, compare, and complete payments on behalf of consumers. Mastercard has introduced Agent Pay, its own infrastructure for secure agentic transactions. Microsoft and PayPal have launched Copilot Checkout. BBVA has embedded its banking app directly into OpenAI’s ChatGPT in Germany and Italy, letting customers engage with their financial products through a conversational AI interface. In 2025, fifty of the world’s largest banks announced more than 160 individual AI use cases — more than in any previous year on record.
The infrastructure moment has arrived. What comes next is adoption — and adoption, historically, arrives faster than any incumbent expects.
The Difference Between a Chatbot and a Co-Pilot
Most of what banks have deployed under the label of “AI” to date has been generative AI in its most passive form: a chatbot that answers questions, a tool that summarizes documents, a system that drafts credit memos faster than a junior analyst. These are genuinely useful. They are not transformative.
Agentic AI is a different animal entirely. The distinction is not technical sophistication — it is the nature of the action. Generative AI produces content. Agentic AI executes decisions. One tells you what your options are. The other acts on them.
Think about autopilot on a commercial aircraft. The pilots are present. They set the destination. They monitor the system and take control when judgment is required. But the plane moves through the air, adjusts for turbulence, and maintains altitude and heading without a human hand on the stick for the vast majority of the flight. This is the model of the AI financial co-pilot — what we call the MoneyMind.
You define your financial goals: retire at 62, maintain six months of emergency reserve, keep credit card balances at zero, maximize match in your 401k, take one international trip per year. From that point forward, the MoneyMind monitors every variable that affects those goals — your income, your spending, your rates, your portfolio, the interest environment — and takes action within the boundaries you have set. It sweeps idle cash into higher-yield accounts. It flags when a better mortgage refinance rate becomes available and models the break-even. It rebalances your investment portfolio when it drifts from your target allocation. It notices that you have a credit card carrying a balance at 22 percent interest and suggests a personal loan at 9 percent, then executes the transfer on approval.
None of this requires human intervention at each step. All of it requires human judgment at the goal-setting stage and human oversight as a continuous backstop.
Chatbots answer questions. Agentic AI acts on them—becoming a financial co-pilot that continuously manages money toward goals you define.
The Trust Equation Banks Cannot Afford to Ignore
Here is where the story takes a turn that is genuinely good news for banks — if they move quickly enough to claim the advantage.
McKinsey’s 2025 global consumer survey asked people which institution they most trusted to offer AI-powered financial services. Sixty-two percent said their primary bank. Only 19 percent said a major technology company.
This is a profound finding, and banks should treat it with the urgency it deserves. Consumers do not want Google managing their money. They do not want Meta knowing their account balances. They are deeply, instinctively uncomfortable with the idea of a technology company — even one they use every day for everything else — sitting at the center of their financial life. The trust that banks have spent generations building is not an anachronism. In the age of AI, it is a competitive moat.
But a moat only protects you if you build on it. The same survey found that 55 percent of Americans asked a large language model for financial advice in 2025 — up from just 10 percent in 2024. That is a fivefold increase in a single year. Consumers are already turning to AI for financial guidance. They are already trusting algorithms with their financial thinking. The only question is whether their bank is the entity delivering that AI experience — or whether someone else is filling the gap their bank left open.
Wells Fargo is already moving. The bank deployed an internal AI agent to 35,000 bankers across 4,000 branches to help employees find and surface information for customers. Within a short period, 75 percent of internal information searches were flowing through the agent — a staggering adoption rate for any new technology inside a large organization. If that kind of adoption is happening internally, consumer-facing deployment is not far behind.
Future banking loyalty won’t depend on where money sits—but on which AI system manages your entire financial life intelligently and continuously.
The New Geometry of Financial Loyalty
The implications of AI co-pilots extend well beyond deposit optimization. Consider credit cards, where 75 percent of revolving balances in the United States belong to prime and super-prime consumers — people with the credit scores and income to qualify for lower-cost debt alternatives. They carry these balances not because they can’t do better but because switching requires research, applications, and decisions they keep deferring. An AI agent eliminates the deferral. It identifies the gap, presents the option, and executes on approval.
For banks, this dynamic cuts both ways. An AI agent that is loyal to the consumer — rather than to the institution holding the deposit — will ruthlessly optimize away every inefficiency the institution depends on for margin. An AI agent that is offered by the bank itself, however, becomes the deepest possible hook for customer retention. If your bank’s MoneyMind is managing your savings optimization, your debt consolidation, your investment rebalancing, and your tax strategy, switching banks doesn’t just mean changing a routing number. It means starting over with a system that knows you.
The financial relationship of the future is not about who holds your deposit. It is about who manages your financial intelligence. That distinction will determine which institutions still exist in twenty years — and which ones become cautionary case studies taught in business schools.
This is the race that is already underway. The banks building their MoneyMind infrastructure today are not doing R&D. They are doing survival. And the consumers who adopt these tools earliest are about to discover something that should have been available to them for decades: a financial system that actually works for them, not around them.
In Part 4 of this series, we explore the force that will take the MoneyMind out of the bank app entirely — and weave it invisibly into every transaction, platform, and experience in daily life.
Related Articles
McKinsey & Company — How Gen AI Agents Threaten Retail Banks’ Customer Relationships https://www.mckinsey.com/industries/financial-services/our-insights/how-gen-ai-agents-threaten-retail-banks-customer-relationships
McKinsey & Company — The End of Inertia: Agentic AI’s Disruption of Retail and SME Banking https://www.mckinsey.com/industries/financial-services/our-insights/the-end-of-inertia-agentic-ais-disruption-of-retail-and-sme-banking
American Banker — Customer Use of AI Agents Will Depress Profits, McKinsey Warns https://www.americanbanker.com/news/customer-use-of-ai-agents-will-depress-profits-mckinsey