JPMorgan Chase ranked first overall in the AIQ 75, carrying the flag for U.S. commercial banks including BNY (15), Bank of America (19), Citigroup (51), Wells Fargo (53), and Capital One (62).

The financial industry outpaced 16 other sectors, with engineering and construction in a distant third. Globally, financial institutions were set to invest 2% of their revenue in AI initiatives this year, in line with the tech sector and outpacing other sectors, according to consulting giant BCG.

It’s not hard to see the appeal of AI to financial firms. Uses for the technology include voice bots that handle customer calls, more personalized pitches from advisors, speedier document preparation, and AI tools that bolster cyber defenses, accelerate software development cycles, and improve the productivity of underwriting staff.

At Bank of America, chief technology and information officer Hari Gopalkrishnan spearheads a roughly $14 billion annual technology budget, which includes more than $4 billion earmarked for AI and other new initiatives. The focus has been on Erica, an AI virtual financial assistant that’s already handled more than 3.5 billion client interactions.

An internal version of the tool, called Erica for Employees, is now used by 90% of Bank of America’s workers. This summer, Gopalkrishnan debuted the AI agent EricaAssist to more than 18,000 customer service representatives. The generative AI-powered agent provides contextual guidance for service reps, helping reduce average call times by nearly one minute per interaction.

Bank of America has also deepened its usage of AI to help software developers more efficiently write code, draft prep documents ahead of client meetings, and help investment bankers develop pitchbooks. “Pretty much everything we demoed a year ago is now live in a scaled fashion,” says Gopalkrishnan, who ascended to his current role in July 2025.

His vision is for AI applications within the bank to stretch from copilots that collaborate with workers to agentic systems that can take autonomous actions. Gopalkrishnan has been more conservative on the latter, and he feels that decision has been validated amid rising incidents of AI agents going rogue.

In March 2026, Brian Saluzzo joined Citigroup as chief information officer, after almost four years at Google, but also with almost three decades of expertise in the financial services sector.

Saluzzo says Citi’s AI investments focus on four areas: accelerating business growth, workforce enablement, productivity and reengineering complex workflows, and cybersecurity.

Some of the more expansive AI deployments have included the April 2026 debut of an AI wealth advisor called Citi Sky, which was built using technology from Google Cloud and Google DeepMind. There’s also Citi Stylus, which summarizes documents and performs translation. It can compare multiple documents and has been updated with new capabilities.

Saluzzo says 87% of Citi’s employees have adopted Stylus, resulting in 65 million enterprise-focused AI interactions.

“We encourage active experimentation, because we have guardrails to make sure that things don’t go awry,” says Saluzzo.

Koren Maranca, head of AI for wealth management at Morgan Stanley, says the firm struck a partnership with OpenAI in 2022 that predated the launch of ChatGPT in November of that year and collaborated with the AI hyperscaler when the bank created its first-ever generative AI tool in September 2023, which is now known as AI@MS.

“We’re always looking at the innovation that’s coming and really are committed—all the way from the top down—to invest in that innovation well before it becomes mainstream,” says Maranca. Morgan Stanley works closely with several frontier AI companies, including Anthropic and xAI.

One of Morgan Stanley’s main priorities with AI is to find ways to transform its entire workflow. The goal is to take the processing tasks that can be handled by large language models and combine them with the personal relationships that its financial advisors have. Efficiency gains from AI are also intended to create more capacity for advisors to expand their book of business.

“We’re evolving toward orchestrated systems where multiple agents work together across workflows and enterprise systems,” says Maranca.

Saul Van Beurden, head of AI and co-CEO of consumer banking and lending at Wells Fargo, says the San Francisco–based bank has focused its efforts on the highest impact use cases. A return on investment needs to be clear, either helping generate revenue, boost productivity, cut costs, or reduce risk.

Every business line and corporate function across Wells Fargo has a leader that’s accountable for the entire life cycle of any AI use case that’s explored, from ideation to pilot to full deployment. A centralized IT product team guides them, but they are empowered to experiment independently.

“We want to give you the keys of the use cases that you think you need to do, but you need to show to us that these are the highest-impact use cases in terms of ROI,” says Van Beurden.

Wells Fargo has also made Microsoft 365 Copilot available to 148,000 employees. AI is also being widely used by retail bankers and wealth advisors to help them better prepare for discussions with clients. For licensed bankers in the branches, Wells Fargo has seen a 31% increase in product sales, which include new deposit accounts or new credit cards opened, when aided by AI. Referrals to AI-enabled wealth advisors have doubled, according to Van Beurden.

Prem Natarajan, the chief enterprise AI officer and chief scientist at Capital One, says his company has recently put more emphasis on its agentic AI infrastructure, working to reduce latency and lower token costs. That has led Natarajan to embrace customized, open-weight models and build specialized models with less context that’s more targeted for the desired outputs.

This approach has led to more thorough and consistent case summarization for customer service agents, who now edit just one-fifth of the copy that’s drafted by AI, according to Natarajan. “You’re just massively reducing the stress in the system,” he adds. “It’s more satisfying for everyone.”

In 2026, Capital One launched an enterprise-wide AI learning hub to over 60,000 associates, but Natarajan says he also focuses on “actually training the AI itself in Capital One” to understand the company’s unique approach to financial services.

BNY debuted its proprietary AI platform, called Eliza, in 2024. The model-agnostic platform was designed to have unified governance and a control plane. 

The bank ended last year with 160 AI initiatives in production, which has grown to nearly 400 across the company today. Nastassja Hagan, head of applied AI at BNY, says 70% of the bank’s employees regularly use AI in their workflows, with more than 50% deemed “power users.”

Hagan says early upskilling efforts were focused on ethics and responsible AI usage but are now more expansive, with three tiers of curriculum training available to BNY’s 47,000 employees.

“We made AI widely available at the bank and we had a huge focus on upskilling, which has continued through this year,” says Hagan, adding, “People don’t view AI as optional, and we’re not really having to establish that it’s important. It’s more about execution of the hard problems where we’re really focused.”