Deploying AI in banking is no longer optional. The hard part is showing the roadmap and how it will deliver results.
Alexandra Mousavizadeh, co-founder and co-CEO of Evident, which tracks the use of AI in the financial industry, said some AI capabilities are “high stakes” for banks at the moment. Think back-office functions like legal document review and routine onboarding operations. But beyond that, Mousavizadeh Bank needs to double its “competitiveness.”
“If a particular area accounts for a large portion of their business, banks will double down on AI usage there,” she says. For companies with large wealth management businesses, that might mean helping advisors better analyze customer data. For banks with a retail focus, that may mean prioritizing chatbots and customer engagement.
Banks are pouring billions of dollars into AI, and the technology is expected to redefine 44% of the work done in banks by 2030, according to consulting firm ThoughtLinks. JPMorgan has grabbed the top spot in Evident’s AI maturity rankings, investing at least $2 billion in the technology and deploying the tools to more than 300,000 employees.
But some investors and analysts are starting to question returns. Executives at several bulge bracket banks answered questions on recent earnings calls about when the productivity and revenue gains from AI will start showing up on their balance sheets. Banks are under pressure to demonstrate how AI is helping them gain an advantage.
Mousavizadeh said banks that take a “centralized” approach to technology decision-making are often able to move faster and incorporate AI more seamlessly.
Her comments about focusing on embedding AI into core areas of business echo those of Dan Priest, chief AI officer at consultancy PwC. He previously told Business Insider that the return on investment for companies that took a “crowdsourcing” approach to AI deployments was “pretty disappointing.”
Priest said the move to a “top-down” approach has become more effective, allowing clients to focus on fewer tools and master a smaller set of tasks more deeply.
While AI agents are a top concern as banks race to prove that AI investments are worthwhile, Mousavizadeh said their adoption is still in its infancy, especially in external roles such as bankers and traders. Agents and humans will likely be used together for the foreseeable future as banks, some of the most highly regulated companies, consider what guardrails to put in place in such cases.
For the past six months, Goldman Sachs has been working with Anthropic to develop a joint autonomous AI agent that can automate internal tasks such as trading, transaction accounting, and customer onboarding. The bank’s technology chief said he expects the agency to launch “soon,” and it’s premature to think it will lead to job losses, CNBC reported.
How banks are measuring AI success
Over the past year, banks have changed the way they measure AI success, moving from tracking specific use cases to expanding capabilities, Mousabizadeh said. They are thinking about how to apply capabilities from one line of business to other lines of business and create internal “architectures” that allow AI to reconfigure workflows across the company.
Achieving this scale requires a combination of top-down and bottom-up approaches. Mousavizadeh said banks need to instill technology among all employees, and mandatory training often yields better results.
But duty alone is probably not enough.
“AI is interesting. It requires a culture of creativity,” Mousavizadeh added.
Mousavizadeh said banks have a new “North Star” as they look to an AI-integrated future. What will a fully AI-integrated bank look like in a few years?
“You have to work backwards from there,” she says.
This means not just applying AI to existing products, but creating new systems inspired by the banks of the future and running them faster than competitors.
