AI solutions are top of mind in financial services, with the majority of institutions already working toward streamlined integrations of existing systems for improved intelligence, security and efficiency. Gartner reports that 57% of financial institutions are already implementing or planning to implement agentic AI in the near future, for example.
However, the road to implementation hasn’t been without trials. Now, most institutions integrating agentic AI are grappling with a formidable challenge: ensuring the sensitive financial data of customers and businesses isn’t compromised.
“Almost every conversation I have with a CIO or chief risk officer starts with AI agents, but it quickly turns into a conversation about data foundations, governance and trust,” says Rinesh Patel, global head of industry for financial services at Snowflake. “Firms have moved past experimentation and are asking how they put agents into production safely, against data they can audit, explain and stand behind.”
Here’s how AI agents are working in financial services, and how leaders are thinking through security.
Click the banner below to learn how organizations are unlocking artificial intelligence’s potential.
How Agentic AI Helps with Financial Services
As financial institution tech teams consider just how agentic AI can be helpful to them, the Gartner report breaks down potential applications into four workflows:
- High-volume interactions, transactions and context
- Mature data, metadata, APIs, workflows and process orchestration
- Established authorizations, privacy, trust and security controls
- Experienced users of large language models and generative AI tools
One example, according to Patel, is in anti-money laundering investigations, where traditionally analysts must spend large amounts of time manually pulling extensive data from financial systems, including transactions, customer relationship management systems, sanctions lists, and exposed persons databases. Agentic AI does that now in seconds, generating reports for analysts to approve. “That shifts banks from looking backward to actively defending against financial crime.”
Patel adds, “AI agents are already helping financial institutions make faster decisions, manage risk in real time and deliver better customer experiences, whether that’s automating regulatory reporting at global banks or speeding up underwriting and claims at insurers.”
Overall, the rise in AI allows institutions to move from fragmented, siloed systems and processes to more unified data governance frameworks, which should ultimately be more secure, not less, than previous data environments.
READ MORE: What is artificial intelligence’s role in financial compliance?
Exploring Emerging Trends in Finance
Banks are adopting solutions such as Snowflake’s AI Data Cloud, which gives financial institutions one governed platform to unify their structured and unstructured data, run AI workloads, and securely collaborate with partners and third-party data providers, all without moving data, Patel says. He notes that these integrations break down into three converging trends.
“First, security is becoming data-centric. The perimeter is no longer the network, it’s the data itself,” he says. “Governance, access controls and policies now travel with the data, regardless of where it moves or which agent accesses it, so you can deploy AI across your enterprise without compromising the integrity of your most sensitive information.”
Next, he shares that zero trust is evolving beyond users to include AI agents, including every interaction, such as queries or tool calls, and every decision. Each one is verified, scoped and auditable, which means agents can be “traced, challenged and explained to a regulator” — a must-have for an AI-forward institution.
Finally, Patel says, unified governance frameworks are increasing in importance. Fragmentation is becoming “untenable,” a traditionally common issue as teams, tools and regions are siloed. “The trend is toward a single framework spanning structured and unstructured data, first- and third-party sources, across every cloud and jurisdiction with consistent policies, lineage and auditability,” he says.
A typical mortgage journey is a classic example: “AI agents can analyze applications, financial documents, property data and risk policies in real time, reducing underwriting from days to minutes. But that only works if governance, access controls, lineage and auditability travel with the data itself and enable banks to scale AI safely across the enterprise,” Patel says.
What Tech Leaders at Financial Institutions Should Do Right Now
To take this opportunity to become more streamlined and secure, Patel recommends three steps to banking leaders:
- Active unstructured data: Necessary context might be “locked” in formats AI can’t access, from contracts to communications. Make that accessible, and you’ve overcome the “biggest lock” most firms have.
- Create a semantic strategy: Shared definitions and meanings help AI interpret data consistently.
- Build a connected foundation of first and third-party data: this includes across clouds, jurisdictions and lines of business.
He warns that governance should be built into each step as you go, rather than “bolted on” at the end.
And he says that AI isn’t an all-purpose solution; instead, data strategy, including the data and context around it, matters more than ever.
“A governed AI agent can unify and synthesize this information in real time, surfacing risks, hedging opportunities, liquidity insights and geopolitical exposure before they become operational issues,” Patel says. “But only if the data foundation beneath it is unified, governed and connected. Without that foundation, agents operate in a vacuum. With it, they become the most powerful decision-making infrastructure financial services have ever had.”