Beyond chatbots: How agentic AI will transform financial services

Revolutionizing Finance: The Critical Role of AI Governance in Agentic AI Adoption

Ritesh Singhania, Co-founder & CEO of Zango, addressed the audience at Web Summit Lisbon 2025, highlighting the unique challenges of adopting AI agents in financial services. His company builds AI agents for risk and compliance teams within this highly regulated sector, acknowledging the widespread excitement for AI’s potential while pointing out distinct hurdles faced by banks and insurers.

Mr. Singhania revealed a significant “confidence gap” in the financial industry regarding AI agent deployment. While many AI agents exist in sandbox environments, very few actually reach production. He cited a statistic from the Bank of England, indicating that less than four percent of AI agents developed in sandboxes successfully transition into live operational use.

Illustrating this challenge, Mr. Singhania shared Zango’s experience: it took 11 months to move an AI agent from discussion to production with a major bank. A staggering 80% of this time, eight months, was dedicated to convincing the bank to open its APIs. This underscores that building trust within financial institutions is the most formidable obstacle.

The vision for agentic AI in back-office operations, such as streamlining bank account openings, involves a sophisticated network of AI agents working collaboratively. While chief AI officers and heads of transformation possess detailed architectural plans for such systems, these blueprints often fail to materialize into production environments.

The primary concern preventing widespread adoption is the fear of an AI agent “going rogue.” Mr. Singhania provided a hypothetical scenario where a customer support agent might inadvertently increase a customer’s credit limit, leading to heightened risk exposure for the bank and potentially sending incorrect messages across communication channels.

To mitigate these risks and enable responsible AI deployment, Mr. Singhania stressed the urgent need for a robust governance layer. This oversight mechanism would consist of a network of AI agents specifically designed to monitor, cross-check, and validate the actions and outputs of every other agent within the system, ensuring adherence to protocols.

This crucial governance layer empowers bank CEOs to confidently assure regulators that their institution is utilizing AI responsibly and with full accountability. Without such an overarching system of checks and balances, it becomes nearly impossible for financial leaders to provide the necessary assurances regarding AI’s ethical and compliant operation.

Mr. Singhania concluded by expressing his hope that the financial sector can avoid a prolonged transition period for AI agent adoption, contrasting it with the more than a decade it took for the industry to fully embrace cloud computing. A robust governance framework is key to accelerating this vital transformation.