Banks Are Building the Customer Before the Product Even Launches
There was a time when launching a new credit card or banking product meant months of painstaking preparation — regulatory reviews, customer recruitment, compliance sign-offs, and carefully managed pilot programs. Today, a growing number of the world's largest financial institutions have found a shortcut that bypasses much of that friction entirely: they simply build the customer themselves.
Artificial intelligence-generated synthetic profiles — digital stand-ins that statistically mirror real consumer behavior without representing any actual individual — are rapidly becoming a standard tool in the product development arsenal of major banks. These AI clones cost almost nothing to generate, carry none of the compliance exposure tied to real customer data, and can be produced at virtually unlimited scale. The result is a fundamental shift in how banks conceive, test, and bring financial products to market.
What Are Synthetic Customer Profiles and Why Do Banks Want Them?
Synthetic customer profiles are AI-generated datasets that replicate the statistical characteristics of real consumer segments without containing any personally identifiable information. Think of them as highly realistic digital dummies — modeled on real-world behaviors, spending habits, credit patterns, and demographic traits, but entirely fabricated from the ground up.
For banks, the appeal is straightforward. Working with real customer data in a testing environment creates immediate legal and regulatory exposure. Data privacy regulations such as GDPR in Europe and a patchwork of state-level laws in the United States impose strict limits on how personal financial data can be used, shared, and stored — even internally. Recruiting actual customers for product pilots is slow, expensive, and logistically complex. Synthetic profiles eliminate all of those obstacles at once.
Beyond cost and compliance, synthetic data offers something real customers cannot: infinite flexibility. Banks can generate thousands of synthetic consumer segments on demand, stress-test product designs against edge-case personas, and iterate rapidly without waiting for customer feedback cycles. The synthetic consumer doesn't just compress timelines. It fundamentally changes the pace and economics of financial product development.
Which Banks Are Already Using AI-Generated Synthetic Data?
Adoption of synthetic customer data is spreading quickly across major financial institutions on both sides of the Atlantic, and several household names are already well into deployment.
U.S. Bank is using synthetic audiences to model consumer segments — including high-net-worth households — for the purpose of testing messaging strategies and refining marketing campaigns before launch. Rather than exposing real client data to internal analytics teams, the bank generates representative digital populations that reflect the behaviors and preferences of its target customers.
JPMorgan Chase has taken the approach into risk management and product design, generating synthetic financial data to simulate market behaviors. By modeling how artificial consumers might respond to varying interest rate environments, credit conditions, or product structures, the bank can pressure-test new offerings in ways that would be far too costly and slow using traditional methods.
In the United Kingdom, NatWest, Monzo, and Santander have each built out synthetic data ecosystems primarily for the purpose of training AI models. As AI becomes more central to banking operations — from fraud detection to customer service automation — the need for large, clean, representative training datasets has grown dramatically. Synthetic data fills that gap without putting real customer information at risk.
The FCA Steps In: Bringing AI Testing Inside a Regulatory Framework
The rapid spread of AI-driven synthetic testing has not gone unnoticed by regulators. In the United Kingdom, the Financial Conduct Authority (FCA) has moved proactively to bring the practice inside a structured oversight framework through its AI Live Testing initiative — described as the first program of its kind in the financial sector.
The first cohort of the FCA's initiative launched in October, with NatWest, Monzo, and Santander among the participants. A second cohort began in April, expanding the program to include Barclays, Lloyds Banking Group, and UBS. The use cases being tested span a wide range of critical banking functions, including agentic payments, anti-money laundering detection, and know-your-customer (KYC) verification — areas where AI has enormous potential but also significant risk if deployed without proper governance.
Testing across both cohorts is scheduled to conclude by the end of 2026, with a comprehensive evaluation report due in the first quarter of 2027. The FCA's involvement signals that regulators are not simply observing this shift from the sidelines — they are actively shaping the conditions under which AI-powered synthetic testing can be conducted responsibly.
Governance Questions Still Loom Large
Despite the enthusiasm from both industry and regulators, important governance questions remain unresolved. Chief among them is the question of accuracy: how faithfully do synthetic profiles actually replicate the behavior of real customers, and what happens when they don't? A product tested exclusively against synthetic data could perform very differently once it encounters the full complexity of real human behavior, financial stress, or market disruption.
There are also deeper questions about model bias. If the AI systems used to generate synthetic profiles are trained on historical data that reflects existing inequalities in access to credit or financial services, those biases could be baked into the synthetic populations — and then reinforced through the products designed against them. Regulators and institutions alike will need robust frameworks for auditing not just the products that emerge from synthetic testing, but the synthetic data itself.
A New Era for Financial Product Development
The shift toward AI-generated synthetic customers represents one of the most significant changes in financial product development methodology in decades. By decoupling testing from the regulatory and logistical burden of working with real customer data, banks are gaining a speed and flexibility advantage that was simply not available to previous generations of product teams.
As the FCA's evaluation report approaches and more institutions build out their synthetic data capabilities, the industry is moving toward a new normal — one where the first customer to experience a new banking product may never have existed at all. Whether that is a feature or a risk, the answer will depend heavily on how well governance frameworks keep pace with the technology driving them forward.