Adapt or be arbitraged: Why fintechs and banks must reinvent themselves for the age of AI agents

Co-authored with Nick Gasbarro

The financial services industry may be on the verge of a seismic shift. AI agents acting on behalf of users are poised to fundamentally alter how financial products are chosen, optimised and consumed.

We have long been proponents of digitally adroit fintech businesses building financial products that solve customer pain points. But the next evolution, beyond better interfaces or smarter chatbots, is AI tools actively managing consumers’ or SMBs’ money, seeking yield, reducing fees and reallocating capital in real time. Traditional incumbent banking models reliant on idle deposits, switching friction or opaque pricing will come under direct threat.

As agents increasingly act on behalf of consumers and small businesses, banks and fintechs will compete on a new playing field where autonomously optimized capital allocation is king. The promise of “self-driving finance” could finally become a reality.

From idle deposits to intelligent movement

Traditionally, incumbents benefit from the inertia of customers. Deposits left sitting in checking accounts, earning minimal interest, provide a reliable source of margin for banks. As was discussed in our recent report, Fintech’s Next Chapter: Scaled Winners and Emerging Disruptors, the average deposit balance per user at traditional banks globally is around $15,000.

In contrast, leading challenger banks like Chime average just $970 per user. Despite the gap, challenger fintechs are gaining ground, with global deposit growth rates outpacing incumbents by 37 percent. This momentum is driven by customer-centric design, faster innovation cycles and higher yields. On the current playing field, as consumer trust deepens and scale accelerates, average deposit balances at these challengers are expected to grow meaningfully. This illustrates how easily incumbents can confuse inertia with loyalty.

This playing field will be redefined even further. Picture a scenario where an AI agent manages your deposits continuously. It shifts funds daily, or even hourly, between high-yield accounts, government bonds, money market funds or decentralised finance pools, all calibrated to your unique risk profile, spending habits, tax considerations and liquidity needs. The consumer no longer needs to lift a finger. Financial optimization becomes ambient, automated and tailored to individual risk preferences. It has decision authority within clearly identified boundaries.

This vision is not only rapidly becoming a reality but also could upend the economics of the system. Of course, this technology remains bleeding edge and granting real authority to agentic systems is far from straightforward. And we’ve yet to see a company truly breakthrough in this space. But given the marginal cost difference between a human and an AI agent, financial services players are incentivised to innovate quickly.

The compression of margins Is inevitable

Consider what happens when switching costs become near zero and intelligent agents can spot yield discrepancies and act in real time, arbitraging away the inefficiencies by which financial institutions have historically profited. Traditional banks rely heavily on deposit-funded lending, with net interest income (the difference between what banks earn from lending and what they pay out on deposits) making up over half of total net revenue, per the quarterly reports of big banks like Bank of America, Wells Fargo and JPMorgan Chase.

A disruption in this model and revenue stream would have meaningful impacts. Most banks are not built for the level of scrutiny associated with agentic AI evaluating hundreds of providers simultaneously and reallocating deposit funds at the first sign of inefficiency. They have relied on information asymmetry and consumer apathy, which now may be diminishing rapidly.

The new moats in a post-inertia world: Advantage disruptors

In this agentic future, traditional incumbent strengths such as brand recognition, branch presence or regulatory moats will need to be redefined as they approach defending this huge profit pool. Some of these advantages may hold water, but only if they are integrated into a model that is optimised for transparency, programmability and efficiency. The winners will likely share some of these four competitive strengths:

1. Ultra-low-cost organization

Operating with a low-cost base will be critical in a world of compressed spreads. Institutions with cloud-native technology stacks, modular core systems and real-time settlement capabilities will be positioned to compete sustainably. Large banks and community banks with branch-centric models stand at a relative disadvantage under this framework.

2. API native infrastructure

What was once tolerable friction in traditional processes may soon become costly inefficiencies. Unlike humans, AI agents operate without marginal cost pressure and expect seamless, instant execution. A prime example of this friction is analog customer onboarding, where processes like paper forms or manual approvals will make a bank completely invisible to the AI layer.

This digital disconnect will severely impede a bank's ability to offer competitive, real-time services and leverage AI for things like deposit arbitrage. Consequently, future financial platforms must be fully API-accessible, offering real-time data, open integrations and clear permissioning. These capabilities will no longer be a differentiator. Rather they will become the new baseline for market participation, a future which digitally native fintechs have inadvertently been preparing for and are now well positioned to exploit.

3. Brand built for an agentic future

We often say to not confuse customer loyalty with inertia. This takes on a whole new meaning in this agentic future. Customer acquisition and retention will center around trust and results. Trust is foundational in the relationship between financial services providers and customers. In some ways institutional AI agents will need to market to and build trust with their customer AI agent counterparts as much as humans.

That means your institution needs to be machine-trusted: transparent on fees, agile and effective in transfers, compliant with regulations and resilient under stress. While incumbent banks are known and trusted institutions, they’re often not well liked by their customers, with Net Promoter Scores (NPS) stuck in the teens or low twenties. Meanwhile, fintechs are not only gaining trust but also winning hearts, with customer NPS scores soaring into the 60s, 70s and beyond.

4. Mass customization at scale

As agents handle more of consumers’ financial lives, demand for hyper-personalized financial products will grow. Whether it’s tax-optimized savings accounts, dynamically priced insurance, or real-time credit lines tailored to daily cash flow, institutions that can construct bespoke offerings algorithmically will hold a competitive advantage. Incumbents have masses of data on their customers which, if oriented properly, could unlock hyper customization. At the margin, banks must self-cannibalize in order to defend ground. Capital One’s insurgency 30 years ago proved that a one-size-fits-all approach breaks down under marginal economics.

Strategic imperatives for banks and fintechs

1. Redefine distribution strategy: you will be competing as much for the end-user’s attention as you will for the trust of the AI that acts on their behalf. Your value proposition must be designed accordingly. Identify and eliminate every process, product or policy that introduces delay or opacity.

2. Make your brand machine compatible: assess your tech stack and capabilities to ensure your institution can be evaluated by agents through structured data, transparent pricing and reputational signals that are machine-readable.

3. Prepare for a new age of competition: if your business relies on passive consumer behavior, it will be systematically disintermediated. You must either move up the value chain, offering advisory, structuring or aggregation, or drive operating costs down dramatically.

Final reflection

Just as Credit Karma, a QED investment, reimagined how people shopped for financial products in a digital age, there is another opportunity to reshape how people shop for financial products in an agentic future. This time, though, the arbitrageurs are autonomous agents working for everyday consumers. The potential impacts of this future are significant, and its arrival may be closer than many anticipate. The underlying technology is already in place, with AI advancing at a rapid pace. Companies like Albert, Kudos and MaxMyInterest are trying to implement some version of this methodology.

Customers stand to benefit from this evolution with bespoke financial solutions that are constantly working on their behalf at their fingertips. Gone will be the days of below-market yield on cash savings accounts. As noted, the average global bank deposit balance stands at $15,000. With the FDIC reporting an average savings rate of just 0.38 percent APY, the typical consumer currently earns a meager $57 in annual interest.

However, a world where consumers collect 4.5 percent (let alone more) on their deposits would see that average consumer gain over $675 in annual interest, representing a substantial improvement in financial well-being.

Incumbents and fintechs alike must begin preparing their tech stacks, product offerings, cost base and brand positioning for this optimized, agentic future.