Powering Agentic AI in financial services: from Open Banking to Smart Data

Agentic AI could transform finance, but it needs Smart Data to thrive. Peter Cornforth explores what’s next, from Open Banking to autonomy.

Agentic AI is generating a major buzz in financial services, and for good reason. It promises to transform how we interact with our financial providers, making tedious, time-consuming tasks automatic and intelligent.

Let’s face it: we have a bit of a love-hate relationship with financial services. We value the services but not necessarily the service. We know we should shop around for better deals, switch providers, or optimise our policies, but the reality is we usually don’t. The effort of comparing offers, understanding the fine print, and navigating complex sign-up processes often ends with the familiar: ‘I’ll get around to it later.’

Enter agentic AI, autonomous (or semi-autonomous) systems designed to act on your behalf in financial decision-making. These agents can analyse options, negotiate, and even execute decisions based on your preferences, potentially transforming the consumer experience.

The rapid advances in AI over the past few years have made this vision increasingly feasible. But while the algorithms are advancing, the supporting infrastructure, particularly data access, is playing catch-up.

Smart Data, smarter finance

As with all AI systems, agentic AI needs data to function. And not just any data – it needs secure, granular, real-time access to personal financial data, information about available products and services, and broader contextual inputs to make smart, tailored decisions on your behalf.

This raises a critical challenge.

Unlike generative AI models like ChatGPT or Gemini, which are trained mostly on public data, agentic AI must tap into highly sensitive personal data that’s protected behind multiple layers of security and regulatory compliance. Add to that the complexity of modern financial lives, with accounts spread across banks, insurers, investment platforms, pension providers, and more, each using different data formats and authentication standards, and the difficulty becomes clear.

The future of agentic AI in finance depends not just on better AI models, but on building the rails that allow secure, proportionate, and seamless data access across the financial ecosystem.

Open Banking +

Open Banking, introduced in 2018, has been a game-changer. It gave consumers the power to share their payment account data securely and in machine-readable format with service providers of their choice. Today, nearly 7 million UK account holders use Open Banking-enabled data services every month – a number that continues to grow, despite the fact that the available data is limited to payment accounts.

The next step forward is Smart Data, an evolution that builds on Open Banking’s success by extending data-sharing capabilities to other areas of financial services (and beyond). The goal: to give consumers and businesses greater control over a wider range of their data, unlocking innovation and competition across multiple sectors.

One area that could benefit enormously is insurance, a personal pet hate of mine.

Imagine having a personal AI agent that could find the best insurance deals tailored to your specific circumstances, and switch policies based on your preferences. Sounds great, right? But to do this effectively, that agent would need access to a wide array of data:

• Your insurance policy history, from both current and past providers;
  
• Your transaction data, to spot major purchases that might affect coverage;
• Your property information, pulled from address data, land registry records, and legal documents;
• Signals from your spending to identify upgrades like an alarm installation;
  
• And finally, comparison data from across providers to check for better deals and missing data that could improve your insurability.

That’s just for insurance. Multiply that across pensions, savings, energy, telecoms, and more, and you start to see the scale of opportunity Agentic AI presents when paired with truly interoperable Smart Data.

What needs to change

To unlock this potential, Smart Data infrastructure must evolve, building on lessons learned from Open Banking. Two key challenges stand out:

1. Regulatory fragmentation

Open Banking works smoothly because the data sits within a single regulatory perimeter. But the real power of agentic AI lies in its ability to combine data from multiple sectors – banking, insurance, utilities, housing to create insights that no single data holder can provide.

If each sector maintains its own regulatory scheme, we risk creating a system where the AI agent must register and comply with multiple regulators. While that might be manageable at onboarding, the ongoing complexity of adhering to different standards and rulebooks could degrade the user experience, pushing everything toward a lowest common denominator approach.

2. Authentication friction

As in the insurance example, an AI agent may need to access dozens of different data holders, all of whom currently expect the customer to authenticate separately. If that model persists, the benefit of having an autonomous agent diminishes fast.

What’s needed is a delegated trust model: one that allows a consumer to grant standing consent to their agent, authenticates once, and enables secure, revocable access across services. This should be underpinned by a robust Smart Data governance framework, with safety, auditability, and consumer control at its core.

Conclusion

Agentic AI and Smart Data could go together like strawberries and cream – a powerful combination that could finally bring about truly proactive, personalised financial services.

But this won’t happen on its own. It will require coordinated effort from regulators, industry, and infrastructure providers to solve the challenges ahead of us, some of which will be costly to handle. That said, given the scale of the opportunity, in terms of both economic impact and consumer benefit, it’s no surprise this is shaping up to be one of the hottest topics in financial services in 2025.

About the author

Peter Cornforth is a Technical Specialist at the FCA. He has been building seamless, secure financial solutions in Open Banking for 8 of his 17 years in financial services and is passionate about consumer-led innovation. Views expressed here are his own and not necessarily the FCA’s.