Artificial intelligence

Citizens is six months into its Reimagine the Bank initiative, and the pace of change in the artificial intelligence realm is “overwhelming,” said Brendan Coughlin, the bank’s president.

“We're building tools for engineers to be more productive on code development, and every three weeks, it's like, well, Copilot’s in the lead. No, no, Claude’s in the lead … and your head wants to explode,” he said, referring to Microsoft and Anthropic AI products.

Banks must shake the old mindset of set it and revisit in five years, he said, as large language models and other firms “continue to leapfrog each other.”

“These things are becoming obsolete in matters of months, not years,” Coughlin said during a recent interview. “It's a really tricky dynamic that I think most organizations haven't had to deal with, ever.” 

Executives at the Providence, Rhode Island-based bank “intellectually knew that, but seeing that real-time, that you're revving the organization up, it’s pretty complicated, you’ve got everybody going in the right direction, and then you're like, ‘Whoa, wait a second – now this new thing just came out, and are we on the right path?’” he said. 

“The nimbleness around it has been pretty intense,” Coughlin said.

Citizens’ Reimagine plan involves taking “a giant step back,” Coughlin said. That means simplifying the $226 billion-asset lender’s business model through vendor consolidation and a smaller office footprint, and applying AI to reshape bank operations.

Through nearly 50 initiatives, the bank is targeting about $450 million in benefits by the end of 2028, with about two-thirds tied to expense efficiencies. Citizens has projected about $50 million of one-time costs to be offset by $45 million of benefits this year.

In January, CEO Bruce Van Saun said the bank has “a lot of wood to chop to actually execute this program.”

Achieving AI return on investment requires disciplined focus on areas that present tangible outcomes, and “intense, cross-functional enterprise energy,” Coughlin said.

Coughlin sees particular promise with changes to the bank’s call center and engineering operations, to bolster customer experience and help the bank compete with larger lenders spending far more on technology. 

As other banks tap AI for similar uses, “at the end of the day, this will be about execution,” Coughlin said. 

Call center changes are being tested, and Citizens aims to have 25% of calls answered by non-humans by the end of this year, in pursuit of a longer term goal of 50%. Coughlin said that’s a customer experience play as much as it is a cost-cutting effort, with the agent acting “almost like a personal concierge.” 

The bank is moving toward having engineers manage code-writing AI agents, versus writing code themselves. AI tools that have been deployed are making engineers 30% more productive, he said.

That boost “should get us down the field faster, competing with the mega-banks,” he said.

How it’s affecting the team

Those developments have already prompted changes for the bank’s roughly 17,400 employees, including upskilling and retraining through the Reimagine program. Coughlin said Citizens’ “willingness to ready our team … for the future of work” can be a differentiator for the bank.

Where AI agents can field requests for new debit cards, for example, a call center inquiry handled by a human may relate to retirement consultation or securing a mortgage, turning that role into one that’s “more strategic, important; it’s probably higher-pay,” Coughlin said. 

Currently, about 70% of the bank’s customer calls are answered overseas by a vendor. In the short term, “we’re going to shed our outsourced agreements before it touches folks that are directly employed by Citizens,” he said. Similarly, the bank plans to drop some third parties it works with on the engineering side. 

In the long term, “the more clerical, administrative roles that we have – some of them are outsourced, some of them are not – will likely not be needed at the same level of scale,” Coughlin said.

He likened it to retail banking changes over the past decade, with branch interactions becoming less transactional and more advice-driven.

Citizens doesn’t have a projection on how its overall headcount may be affected, but role transformation will happen broadly, he said.

Vendor and branch changes

As the bank streamlines its vendor roster, Coughlin said executives are mulling where they can consolidate business with one vendor rather than two or more, and which companies are the right ones to link with over the long term. 

When it comes to core providers, “the FISes and Fiservs don't have the here-and-now modern platform and they're telling you that they're going to build it for the future. And then you have these startups and challengers that are thinking about it in a more modern way, but they don't really have a platform that works today.” 

“Do we have confidence the incumbents will get modern at the pace we need?” Coughlin said. “Or do we want to wait it out a little bit and see if these challengers emerge with the right mindset, but a platform that actually works here and now?” 

Other simplification moves are easier, such as picking one customer chat solution to use across the bank instead of using different platforms in various divisions, he said.  

On the physical side, Citizens is exiting 19 small buildings so far and is weighing, over the next year or two, where to have its biggest office hubs, Coughlin said. Its largest office is in Johnston, Rhode Island. 

The regional bank is also plotting branch network changes and rolling out a new mobile app. About 15% of customers have access to the new app, and Coughlin expects that to hit 100% in the next 30 to 45 days. 

Changes include a more modern user interface; improvements to the login experience; faster functionality; enhanced customer data, including credit scores and logos of the companies they’re spending money with; better search and account opening capabilities; and more functionality for AI assistant CiZi. Coughlin said the bank plans to add more to the app over the next year or two. 

As for branches, Coughlin indicated the bank doesn’t share the aspirations of some peers who’ve de novoed in the Southeast, preferring to refine its existing network in the states it’s already in. Citizens counts about 1,000 branches across 14 states and Washington, D.C. 

The lender has a 2.1% branch share in New York, where 3% to 5% share is needed to achieve operating leverage, he said. “If we want to be a real player, we’re going to have to densify,” he said. 

And in New England, “we're going to do a lot of surgery around some of our legacy markets,” to position the bank for mass-affluent customers and above, he said. That means branches that are “more impressive, [on] the right corner, the right staffing.”

U.S. Bank recently rolled out a new artificial intelligence-powered tool that can review its designers’ work, flag any issues and suggest changes. 

The tool, Design Assistant, was developed in-house, and the bank’s hundreds of designers began using it in December, said Caleb Schmidt, head of experience design at the Minneapolis-based lender. His team designs all of the digital experiences the super-regional bank’s clients interact with.

Designer adoption of the tool is at about 90%, he said. 

As lenders across the industry have pursued AI-powered improvements, the idea for the tool spawned around mid-2025, during a mapping exercise that examined the existing design process and designer jobs within it, Schmidt said during a recent interview.

“There are things that we definitely want humans to only do,” from the perspectives of trust, regulatory and risk, he said, but there are also “manual, tedious” tasks that can be automated.

And in between those two, there are tasks that could be handled more quickly through the use of AI, while the human remains in the driver’s seat, Schmidt said.  

“We really spend a lot of time thinking about, how fast can we move from an idea to getting it in [clients’ or] customers’ hands, and … how can AI help us shorten that process?” he said.

U.S. Bank also considered where it could improve quality, reviewing data to see where issues pop up, whether that’s during the design process, when design and engineering begin working together, or once something is live, he said.

Designers at the $692 billion-asset bank work daily with platform Figma, so creating a tool integrated within that workflow was important, Schmidt said.

While other companies might buy a handful of different tools that the employee then has to understand when and where to use, “we knew that if we wanted our teams to use something that we built, it needed to be part of that broader ecosystem that they're working in every single day,” he said. 

U.S. Bank’s designers spent a couple of quarters developing the tool. “One of the things that we had to do to begin was to train the model on what good looked like,” he said. “That was probably the biggest load or lift.” 

Design Assistant considers the bank’s standards, including brand, content, tonality and accessibility, and connects those to the lender’s design system. As a designer is building a digital experience within Figma, the employee can invoke the tool, which considers what’s been designed and returns a list of things that need to be fixed, Schmidt said.  

That might be color contrast, for example, if a label on a button is too difficult for low-vision customers to read. The Design Assistant cites the specific issue and ties it to U.S. Bank’s standards, then gives the designer the option to make a suggested change, Schmidt said.

“The reason we don't do that just automatically, without that human-in-the-loop intervention, is that sometimes those were decisions that were made for a valid reason,” he said. “We want that human check to happen before it automatically fixes something.”

The tool is used for large tasks, such as scouring hundreds of pages built for a brand-new bank experience, and small ones, like optimizing existing pages by moving the location of a button, Schmidt said. 

The AI tool is saving designers time, Schmidt said – including hours of back-and-forth between designers and engineers – and ensuring what’s sent from design to engineering is defect-free. It also enables the bank to track where issues are cropping up. Such data could inform future employee training if patterns are observed, Schmidt said.

From an accessibility perspective, U.S. Bank is seeing 30% fewer defects because the tool catches quality issues before heading to the development side, Schmidt said. That’s saving the lender money, too, he said, because it’s more cost-effective to catch those defects earlier in the process.

Designers building the tool for their own use is emblematic of the culture of innovation the bank tries to drive, Schmidt said. “We want to empower our employees across the company to contribute these types of tools,” he said. 

U.S. Bank declined to say how much it invested in developing the tool. Return on investment will take some time to measure because it’s only been in heavy use for about one quarter, Schmidt said. But the bank is starting to make some correlations, such as increases in customer satisfaction as the tool has resulted in defect-free experiences.

As the bank refines Design Assistant, “we've got a lot more that we have planned and built into a roadmap,” including integrating more of the bank’s brand standards, he said. Incidentally, U.S. Bank is starting to roll out “a pretty substantial brand refresh” that balances expertise and trust with a premium or sophisticated feel, he said. 

Since the lender has done research and testing that found specific patterns or ways of problem-solving methods work better than others, the next layer of value the tool could offer might be proactively surfacing suggestions while a designer is building a flow, Schmidt said.

Sponsored content

By Glia

Bank mergers usually look like a win on a balance sheet, but the operational reality is often a "Day One siege". When two institutions combine, phone lines ring off the hook and digital queues become overloaded. The stakes are immediate: customers are three times more likely to switch banks following a merger. About 17% of customers move at least one account to a competitor shortly after an acquisition is announced.

While 33% of those who leave attribute it to poor service, another 36% cite the "emotional experience of change". This attrition starts when a new customer’s debit card fails to activate or they cannot find the login for a new mobile app. Their anxiety is real, and the traditional contact center infrastructure is rarely equipped to absorb it.

The Failure of Traditional Scale

Most banks try to manage this surge by ramping up third-party call centers or forcing existing agents into extreme overtime. These "band-aid" solutions often backfire. Outsourced service can feel impersonal, and overworked agents are prone to burnout. Relying on basic chatbots or static FAQ pages also fails because these tools cannot resolve the complex, high-anxiety questions that define a merger. When digital self-service tools are too basic, frustrated customers simply pick up the phone, further overwhelming the system.

During a transition, the contact center faces a chain reaction of pressure. A technical glitch in the mobile app leads to a call. Long wait times lead to customer frustration. Breaking this cycle requires a layer of intelligence that can act before the human agent is even reached.

Intelligent Automation as an Operational Shield

Intelligent automation changes this dynamic by acting as a shock absorber. Instead of forcing agents to handle thousands of routine inquiries—like "What is my new account number?" or "Where is my closest branch now?"—AI resolves these questions instantly. Modern conversational AI understands the specific context of a merger, providing accurate routing numbers and account update information across voice and digital channels.

By automating the routine, a bank preserves its human capacity for harder conversations. When an agent does need to step in, the AI provides interaction summaries and customer history so the agent can solve the problem immediately. This avoids the common frustration of asking a customer to repeat their issue three times. This efficiency reduces the systemic pressure that typically leads to service collapse during the first ninety days of an integration.

The 61% Benchmark: Redefining M&A Success

Recent market data provides a clear benchmark for what an AI-enabled integration looks like. In a $7 billion acquisition that grew a bank's customer base by 25%, the institution utilized intelligent automation to manage the transition. The results were significant: 61% of all calls were fully handled by AI.

The bank exceeded its performance targets and maintained low wait times while operating with fewer associates. By resolving more than half of the total interaction volume autonomously, the bank ensured that satisfaction scores remained high during a period of massive upheaval. This level of efficiency turns a chaotic growth event into a manageable, scalable process.

Strategic Growth Without Friction

Successful M&A scale requires institutions to protect their existing deposit base while simultaneously acquiring new assets. AI-powered automation ensures that while the organization grows, the customer experience remains stable. Deploying banking-specific intelligence converts the operational chaos of Day One into a highly controlled integration. Banks that use AI to handle the flood of routine data protect their agents, their reputation, and their new investments. Consistent, automated support builds the trust necessary to turn a newly acquired customer into a long-term advocate for the brand.

Despite the buzz around artificial intelligence’s potential to transform banking, it’s no “silver bullet,” according to a top bank analyst.

“It is a long, expensive and risk-constrained transformation,” Wells Fargo analyst Mike Mayo wrote in a note to investors in March. “Indeed, we can't find one extremely noteworthy new product or service from AI in these very early days (it's been a little boring from a product standpoint).”

Actual payoffs will take time and the human in the loop isn’t disappearing soon, Mayo wrote, following a bank and tech summit featuring input from JPMorgan Chase, Bank of America, Wells Fargo, Goldman Sachs, BNY, Amazon, Meta and a former Nvidia executive.

“The long-term upside,” Mayo wrote, “belongs to scale players that pair proprietary data with disciplined process redesign, with the added benefit that trust is a moat.”

Sean Viergutz, PwC’s banking and capital markets advisory leader, has seen a shift toward that process redesign mindset in the last six to 12 months, as AI understanding grows. 

“Those that have really grasped the power of it don't use it as a tool,” Viergutz said during a recent interview. “They use it as a co-worker. They use it as a teammate.”

There’s more consideration of “how are we fundamentally changing [organizational] structures, process structures, ways of working, operating procedures to really harness and involve the digital workers part of this?” he said.

Goldman Sachs CEO David Solomon alluded to that in a shareholder letter mentioning the bank’s new operating model “propelled by AI.”

“It has become increasingly clear that our operating processes need to reflect the gains that will come from these transformational technologies,” Solomon wrote. 

Still, Viergutz said, “there’s a perception gap of, ‘Oh, AI can do absolutely everything,’ and that’s certainly not the case.”

While it can be “a step-level shift into being more efficient, being more accurate,” he said, “it's not going to do the job of the expert worker that understands the systems, the technology and what has to happen in the business flow of things, or the product flow of things.”

Among other areas, banks are employing AI in engineering and code development, know-your-customer and anti-money laundering applications, and investment banking, where automation means credit analysts aren’t spending hours putting pitchbooks together, Viergutz said. 

Mayo cited an estimate that one-third of bank jobs or portions of jobs may eventually be better handled by AI.

“For money center banks, there is more scale and productivity across massive headcounts. For trust/asset servicing banks, there's a greater focus on agentic AI for document-heavy, data intensive workflows. For regional banks, efforts are more for targeted efficiency (call centers, fraud, coding),” Mayo wrote.

The bank and tech summit Mayo referenced “reinforced our view that AI can help banks trend toward record efficiency. Banks are transitioning from pilots to production to profits. Yet, the question remains when AI dreams turn to dollars.”

ROI questions

Jamie Dimon, the CEO of JPMorgan, the biggest U.S. bank, told Bloomberg last October the bank’s annual AI-generated savings of $2 billion is about equal to what it spends yearly on the technology.

Other than JPMorgan, “almost no banks disclose savings,” Mayo wrote, adding he wonders if banks know the return on investment they’re currently getting.

It’s a common question right now, as the industry experiments with agentic AI, said Nina Owens, managing director and co-lead of financial services strategy at Publicis Sapient.

“We haven't had 20 years of agents, or even five years of agents, to really capture what the typical returns are,” she said in a recent interview.

Keri Smith, Accenture’s global banking data and AI lead, expects to see firms redesigning job roles to support AI adoption over the next six to eight months, as companies realize what’s needed to see results. She called about 8% of financial services firms AI front-runners, meaning they’re reinventing workflows, have organizational alignment and are already seeing results. 

Owens said several of her bank clients aim to revamp marketing operations with agentic AI, keeping the marketer or content manager in the loop while agents handle data-pulling or other rote work.

That’s become a big focus as banks seek to grow their businesses, she said, because agentic AI can quickly connect the data and identity dots so a bank can provide hyperpersonalized offers to customers. To get there, though, a lot of foundational work has to occur, related to data and modernizing systems, she noted. 

Can’t be ‘AI-ed overnight’

At Bank of America, an AI catalyst group has Jeff Busconi, the bank’s head of strategy, and Hari Gopalkrishnan, BofA’s chief technology and information officer, working with 18 senior business leaders representing different areas of the bank, “to make sure that we're driving AI into every single area,” said Dean Athanasia, the lender’s co-president, during a March appearance at an investor conference. 

Given the bank’s size, “it's a way to get it done and to share and leverage and get every bang for every investment dollar,” he said. 

BofA has said AI-powered virtual assistant Erica handles the work of about 11,000 people, and all 18,000 of the bank’s software developers use coding agents to optimize the development process, boosting productivity by about 20%. The bank has also rolled out Salesforce’s Agentforce, which enables the creation of AI agents to handle tasks, to 1,000 financial advisers, Athanasia said.

Executives at fintech Chime envision AI bolstering efficiency with compliance reviews and ensuring compliance with unfair, deceptive, or abusive acts and practices standards, CEO Chris Britt said in March at an investor conference.

“You still need some element of human involvement, but there’s so many ways to make these processes more and more efficient,” he said. 

It takes time to get it right, though, Britt said, because the fintech operates in a highly regulated business requiring “deep interactions” with banks and regulators.

“It’s not something that can just be AI-ed overnight,” Britt said.

Indeed, cautious approaches to AI remain key for big banks, Mayo emphasized, and precision is more important than speed.

“AI failures at banks have greater downside than others since they erode a bank's most important asset: trust,” Mayo wrote.

Customers Bank is expanding its partnership with OpenAI – such that the bank aims to use artificial intelligence to “reimagine its lending, deposit and payment lifecycles” in commercial banking, the West Reading, Pennsylvania-based lender announced in April.

Customers began its collaboration with OpenAI in 2023, deploying ChatGPT Enterprise.

After its next round of enhancements, to be developed over the next six to 12 months, the $26 billion-asset bank wants to reduce the time it takes to close a commercial loan – including underwriting, document collection and legal negotiations – to roughly seven days, Customers Bancorp CEO Sam Sidhu told CNBC. That’s down from between 30 and 45 days. 

By the same token, it can take more than a day to open accounts for complex commercial clients. That could be cut to less than 20 minutes by using conversational AI and automated document gathering, Sidhu said. 

Sidhu teased Customers’ AI ramp-up on the bank’s first-quarter earnings call, when he noted: “The prepared remarks you heard on my behalf today were delivered by my AI clone, not read by me.”

Despite leaning into its OpenAI relationship, Customers emphasized Monday that it wants to “[keep] customer relationships fully human.”

By letting AI handle lending-related tasks such as document collection, credit memoranda preparation, legal documents, and portfolio and collateral monitoring, “bankers can focus on structuring the right deal and servicing their clients,” the bank said in a press release.

Still, Sidhu seemed keenly aware of the human trade-off.

“When you have an autonomous agent, you’re essentially creating a digital worker … and they can work around the clock,” Sidhu told CNBC.

AI has written roughly half of Customers’ software code, saving about 28,000 hours of work, Sidhu said. That’s the equivalent of not hiring about 15 full-time employees.

“This is an opportunity for us to potentially slow that hiring … and do more revenue per employee,” he said.

Sidhu sees the OpenAI partnership as an avenue to a “fundamental re-engineering of how Customers Bank operates,” according to the release.

Three-quarters of the bank’s team uses tools powered by OpenAI, Sidhu said.

“This strategic collaboration with OpenAI gives us the frontier models, engineering expertise and ability to co-create a roadmap toward becoming an AI-native bank,” he said.

But the partnership is more than deploying an off-the-shelf suite, Customers said.

OpenAI’s technical teams are engaging with the bank directly onsite to build custom AI capabilities by tapping Customers’ processes, data and institutional knowledge. That system could then be sold and customized to other banks.

OpenAI’s chief revenue officer, Denise Dresser, said the artificial intelligence company is pleased to help Customers “build a more intelligent operating model that empowers employees, strengthens client service, and sets a new standard for regional banking,” according to a statement.

“By the end of 2026, our bankers will spend more of their time on the work that creates value for clients and our shareholders,” Sidhu said.

Corporate and investment bank executives implementing technology innovation programs are experiencing lackluster results. More than 80% said new product implementations aren’t boosting revenue, according to a Capgemini Research Institute report released in March. Another 51% said new products didn’t deliver on expected cost savings, either. 

As consumer AI expectations rise, technology leaders are limited in how they can respond. Executives said 29% of annual IT budgets are set aside for transformative technologies, while a larger chunk – 43% – is devoted to maintaining legacy systems. Capgemini surveyed 150 senior leaders from corporate and investment banks and 600 senior executives from nonbank financial institutions and global corporations. 

“Client demands have shifted dramatically, and while CIBs have invested heavily in AI, many are struggling to move beyond the pilot stage,” Catherine Chedru-Refeuil, global head of corporate and investment banking at Capgemini, said in a press release.

CIOs and technology leaders will play a pivotal role in adapting to the structural headwinds corporate and investment banks are facing as customers push for new products and services. 

More than half of corporate and investment bank customers expect real-time responsiveness, 49% want personalized engagement, and 40% are looking for innovative products. However, fewer than 20% of customers find that banks are meeting those expectations, according to the Capgemini report. 

The customer demand for a modern experience comes at a time when revenue is expected to decelerate. Over the next five years, the annual growth rate is expected to slow from 6.5% between 2022 and 2024 to 5.4%, due to competitive pressure, technology challenges and weakened customer relationships, the report found. 

“Nonbanks and fintechs are scaling at speed, and expectations continue to rise,” Kartik Ramakrishnan, CEO of Capgemini’s financial services strategic business unit and a member of the group executive board, said in the report. “For incumbent banks, aging technology, tougher cyber-resilience requirements, and expanding data and compliance demands are squeezing margins and crowding out investment capacity.” 

Banks will need to assess how to modernize legacy infrastructure, redesign operating models and embed artificial intelligence across business operations to pull ahead, Ramakrishnan said. 

Getting there means technology leaders will need to introduce flexible operating models and enable accelerated change, as well as strengthen data and technology foundations with large language models that can support real-time interactions and elevate decision-making, the report said. 

Growing an AI workforce through reskilling, rethinking roles and reinforcing behaviors that support AI-enabled work methods will also play a significant factor in business transformation. Yet Capgemini found 40% prioritized hiring outside AI talent while only 23% invested in internal reskilling. 

Meanwhile, building governance into AI processes and platforms will enable banks to better scale the technology, particularly as almost two-thirds of executives cited high regulatory compliance costs as a business pain point, according to the report. 

“To succeed, CIBs must adopt a disciplined approach: creating enterprise-grade platforms and cultivating an ecosystem of trusted partners,” Chedru-Refeuil said.

Citi began piloting Spaces, a new capability enabling employee collaboration on artificial intelligence-assisted projects, in January, within its Stylus Workspaces AI platform, the bank told CIO Dive. 

Spaces lets teams of up to 15 people work together on documents, analyses and reports in a secure environment. The pilot group included 250 employees, and Citi plans to roll out the feature to the 182,000 employees with access to Stylus Workspaces in the coming weeks. 

“We know that most workflows and projects are a team effort and our goal with AI is to help alleviate extra touchpoints as much as possible,” Citi CTO David Griffiths told CIO Dive in an email. “By creating Spaces, we’re removing the extra step of sharing draft documents or reports outside the tool for feedback or approval.” 

Citi’s latest addition to its AI platform reflects the firm’s ongoing investments to bolster AI capabilities. The financial services industry has been working to deploy AI tools in search of productivity gains and greater efficiencies. 

Creating a shared workspace within the AI platform will help employees speed up reviews, approvals and decision-making, as well as encourage collaborative brainstorming, Griffiths said. 

“Prior to Spaces, colleagues were running individual conversations within Stylus Workspaces and then sharing out the results or findings to their team outside the tool,” Griffiths said. “We created the Spaces feature to allow colleagues to collaborate on Stylus Workspaces conversations in one place.” 

Employees use the Stylus Workspaces platform to summarize research, draft reports, manage projects and create emails, with use cases growing weekly, Griffiths said. Stylus Workspaces, which is underpinned by Google Gemini and Anthropic Claude models, uses Citi data to provide responses to employees, while also searching the web for information outside of Citi’s data ecosystem. 

“We’ve also begun integrating some of the most commonly used tools and platforms around the firm, such as Jira, allowing colleagues to complete work in one place,” Griffiths said. 

Spaces comes a few months after Citi launched agentic AI tools on Stylus Workspaces, allowing employees to extract insights from large datasets and streamline workflows. 

Citi CEO Jane Fraser said AI is driving new sources of efficiency as the company assesses where the technology fits into 50 of its most complex processes. 

“This bank is being truly transformed in terms of its operational capabilities, its controls and its tech infrastructure compared to five years ago,” Fraser said during a quarterly earnings call in January. 

Bank executives see ample opportunity in agentic AI, but getting the most out of the technology means rethinking how it’s incorporated into business processes. 

AI agents can “truly transform operations,” said Ted Paris, senior vice president and head of analytics, intelligence and AI at TD Bank U.S.

However, Paris said executives across the industry are paying attention to reports about AI project failure rates, which has pushed him to carefully consider how to incorporate AI so it becomes successfully implemented in the company's environment. 

“I reflect on that a lot,” Paris said. “It’s not an issue of getting the model wrong. It’s more about how well you’re integrating these things into your business processes so that you get adoption and usage.” 

AI represents a “massive opportunity” for TD, CEO Raymond Chun said on the bank’s fourth-quarter 2025 earnings call in December. TD implemented roughly 75 AI use cases generating $170 million in value in 2025, he said. The bank expects AI to drive $200 million in incremental value in 2026. 

“We are prioritizing our AI investments with use cases focused across categories such as customer acquisition, customer insights and risk management,” Chun said during the call.

As the bank scales AI, leaders have worked to involve the full ecosystem and understand how employees and customers will be interacting with the tools, Paris said. While the bank is implementing an AI strategy that includes traditional and generative AI, it’s also assessing, testing and building its agentic capabilities. 

TD added a generative AI virtual assistant in its U.S. contact center. The Knowledge Management System handles questions from agents and provides consistent responses to reduce client hold time on calls and resolve issues faster while lowering the frequency with which problems are escalated to supervisors, TD said in an email. 

When bringing in AI agents, Paris said it’s critical to involve partners, including risk, legal, IT and change management, to help ensure “our colleagues are ready for this.” 

“You have to start from the beginning in thinking through: How are we going to adopt this? What’s it going to look like in its full production environment, where people are going to have to engage with it?” he said. 

Crafting an AI agent strategy

Financial services firms can employ various AI agent strategies, Paris said. One involves finding tasks that can be easily automated and inserting agents into those processes. Banks “can get a lot of benefit out of that,” Paris said. 

But another way forward is to rethink business processes to take an “agent-centric approach,” Paris said. 

“That means: Let me now pretend like I had these agents from the beginning as opposed to simply plugging them into a process that was designed for a different type of work environment,” Paris said. “Some of these work processes were created years and years ago.” 

Tools for enterprise AI agent design and deployment have matured to a point where financial services firms will take greater advantage of the technology this year, according to an Accenture report released in January. 

Indeed, there are multiple types of AI agents banks can consider that serve different purposes, Paris said, including administrative agents to help automate routine tasks, task-oriented agents to execute actions within workflows or orchestrator agents that help dictate what’s happening with other agents. 

“You need them all, with anything you’re doing,” Paris said. 

Morgan Stanley’s wealth management business, for example, is building super agents to serve as an orchestration layer, Jed Finn, the bank’s head of wealth management, said at the UBS Financial Services Conference in February. 

Meanwhile, Citi updated its AI platform Citi Stylus Workspaces last year to include agentic AI capabilities, enabling employees to condense multi-stage workflows into a single, automated process and gather insights from large datasets. 

Regardless of a bank’s approach to incorporating AI agents, vendors are zeroing in the technology’s potential in financial services. 

Software provider Creatio recently released six AI agents for banking that focus on specific revenue generation and operational efficiency tasks. Salesforce unveiled Agentforce for Financial Services last year, while AWS launched mortgage approval agents powered by Amazon Bedrock, a generative AI platform. 

Looking ahead, Paris said he wants to lean in on creating end-to-end projects. 

“This is not like the traditional world, in which you build a model and you can just deploy it,” Paris said. “There’s far more involved in the ecosystem and, to get the most out of it, you really want to make sure you’re completely rewiring it, not just automating individual bits and pockets.”

Bank of America is advancing its artificial intelligence strategy as it finds ways to move beyond task-specific pilot projects and toward process transformation. 

The bank established an AI strategy centered around governance, innovation sessions and problem solving. Last spring, the bank reported that more than 90% of its roughly 213,000 employees are using its Erica for Employees virtual assistant and documented more than 3 billion client interactions with the bank’s customer-facing version of Erica.

But the bank isn’t sitting idle. During the Semafor World Economy 2026 event in April, Hari Gopalkrishnan, chief technology and information officer for Bank of America, outlined four primary areas, including end-to-end process transformation, scale and reuse, governance and return on investment, that are driving AI project success.

“The big pivot from last year to this year, I’d characterize in four dimensions. The first is shifting from proofs of concept that were focused on small tasks to actual end-to-end process transformation where you’re going after big opportunities that are going to transform either revenue, client experience or expenses,” Gopalkrishnan said.

The technology can save employees time and improve relationships with customers, Gopalkrishnan said. The bank’s wealth management firms recently rolled out AI-Powered Meeting Journey, leaning on its Salesforce CRM data, to provide AI-enabled assistance for its financial advisers before, during and after client meetings. 

“Today, we can understand prospects, identify them, create meeting summaries, help meetings and the whole process goes from days and weeks to hours,” he said. 

Governance and ROI

Scale and reuse is a major focus area for the bank, Gopalkrishnan said.

Bank of America, which allocates 30% of its $13.5 billion technology budget for new initiatives, including AI, is focused on shifting away from individual teams building applications to enterprisewide capabilities that can be reused repeatedly, he said. The bank supports roughly 3,000 processes and is looking at AI as a foundational element to be scaled and used across those operations, he added.  

Scaling AI across the enterprise leads to another primary focus area: governance. 

“This stuff is very hard to govern,” Gopalkrishnan said. “If you overdo it, you stall innovation. If you underdo it, you introduce a lot of risk.” 

Enterprises expect to increase generative AI spending by nearly 40% in 2026, which means AI will be more deeply embedded into business workflows, creating a need for stronger guardrails, according to Gartner. 

Lastly, ROI has become a critical component of the bank’s AI strategy, Gopalkrishnan said. AI adoption is expected to trim banking industry costs by up to 20%. However, legacy IT systems that make scaling AI difficult and lack of alignment on what translates to AI ROI are muddying the waters. 

“A year ago, we probably just said, ‘Let’s try a bunch of stuff out,’” Gopalkrishnan said. “We’re at a point where we say, ‘Before we try a bunch of stuff, let’s understand where this is going to take us from an ROI perspective.’” 

Data is the foundational component behind all of the bank’s work with AI, Gopalkrishnan said. Along with data, implementing a compute strategy and understanding the economics of running AI models is key, and where FinOps plays an important role, he added. 

“These models aren’t cheap, they take a lot of hardware to run,” he said. “It’s very easy to spend a lot of money and find out that you’ve got nothing in return.” 

Upskilling employees also plays a significant part in Bank of America’s AI strategy, Gopalkrishnan said. The bank established an academy focused on reskilling and upskilling employees on AI. The academy provides employees with different AI training programs, from basic prompt engineering to AI design and development. 

The bank filled 44% of jobs in the past few years through internal mobility, partly as a result of upskilling, he said.

Bank of Montreal opened an institute in April focused on responsible AI innovation, application and governance and the development of quantum computing capabilities.

Kristin Milchanowski, who’s spent the last year-and-a-half as BMO’s chief AI and data officer, will lead the BMO Institute for Applied Artificial Intelligence and Quantum in an expanded role.

“BMO has been applying AI across the bank for decades to personalize client experiences, augment our teams and automate our business,” Milchanowski said in a prepared statement. “These capabilities continue to evolve rapidly, alongside the emergence of Quantum technologies, creating both extraordinary opportunity and significant responsibility.”

The institute will help further embed an enterprise-wide AI agenda, advance its quantum strategy and “[keep] us accountable and agile in this era of exponential change,” she said.

BMO has been using AI for decades, the bank said.

Quantum computing, though, is in its experimental stages. Quantum computers could have the ability to solve problems too complex for today's most powerful computers.

BMO joined the IBM Quantum Network – a global web of academic institutions, startups and Fortune 500 companies collaborating with IBM to advance quantum computing – early last year, gaining access to IBM’s quantum infrastructure and thus enabling it to develop and deploy quantum solutions across operations.

Other banks that have joined the IBM Quantum Network include Barclays and HSBC. BMO is the first Canadian bank to join.

According to IBM, quantum computing will lead a significant transformation of the financial services landscape in the next five years, and there are competitive advantages for banks that adopt quantum early.

“BMO is committed to building AI and Quantum capabilities that are innovative, trusted and centered on our clients,” Steve Tennyson, BMO’s chief technology and operations officer, said in a prepared statement.

As artificial intelligence reshapes banking, small lenders know they can’t ignore the technology, said BNY’s head of community banking, Shofiur Razzaque. 

“There's skepticism around the controls: ‘How do I keep us safe?’ There’s no skepticism around, ‘We need to do this; otherwise, we’re in trouble,’” Razzaque said.

The New York City-based bank launched an initiative in September that offers free training, led by BNY senior leaders, for 1,000 community bank executives on topics such as AI, cybersecurity, data analytics, corporate finance and strategic planning.

Dozens of community bank leaders attended program events in December, Razzaque said. As the initiative is tailored to meet community bank needs, an upcoming session for human resources practitioners will be led by a BNY HR leader. 

BNY has $2.2 trillion in assets under management, and the custody bank’s clients include other financial institutions. Razzaque’s community banking and solutions group offers local lenders services such as payment solutions within BNY’s global payments and trade business. 

In engaging with community bankers, it became apparent there was a desire for more information on AI, cybersecurity and digital assets, Razzaque said in a recent interview.

As a large bank, BNY sees the move to assist banks with less than $10 billion in assets “as a duty,” he said.

“These players matter,” he said. “If we help them, the entire financial system gets stronger. Is there an aspect of it where there's commercial opportunity? That's secondary.”

Editor’s note: This interview has been edited for clarity and brevity. 

BANKING DIVE: Tell me about your role and this unit at the bank.

SHOFIUR RAZZAQUE: When we set this team up, we thought about two ways we could help community banks. One is through access to our products, services and capabilities. My job on that aspect is, how do we open the door for BNY? How do we bring them in? So I go out, I meet community banks, I'm at conferences, getting to know these organizations and explaining how we can help them on their journey. 

The second aspect: On some of these emerging capabilities, whether it's AI, digital assets, treasury service trends, we have deep experts who do this for a living. Smaller organizations face challenges such as running on a really lean staff, so they're always looking for some form of education, guidance, insight. So we introduced a free financial education program, allowing community bank leaders, middle managers to take part in sessions on topics such as AI.

Are bank executives turning to their peers more than consultants for this kind of information?

Consultants add a different level of value. But when you are a banker, when you're talking to your own colleagues or folks that have the same background, you're speaking the same language. And when there's no cost associated with that, and you could really ask about use cases and how to increase adoption, I think it's different. It goes from hypothetical, strategic to really practical discussion.

What’s your take on the AI opportunity within community banks?

These banks know every customer that walks through the door. The relationship banking is so pure. But when you go to the back office, a lot of them are doing things the way they've done for many years. That's where AI could really unlock potential, if you can free up employees to interact with customers while you apply AI to some of these mundane tasks. Adoption is scary, but they have to embrace AI; otherwise, if you don't start using and applying it in some shape or format, you fall behind.

Using AI, at first blush, would seem to inhibit the relationship banking community lenders are known for – not so, in your mind? 

The area where AI unlocks is growth. When you're in a smaller organization and what you're focused on is increasing your loan portfolio and increasing your client base, I don't imagine AI doing that. I imagine banker Mr. Smith, who knows the local clients, that's who's going to do that. If you give Mr. Smith more time to focus on that, because you remove some of the tasks from his day to day, that’s the growth. 

What are the hurdles to adoption for these banks?

When you run a bank, you're regulated, you're risk-oriented. You're always worried about controls. But also, not knowing where to start, and almost overthinking it and trying to use it in too many different places. And, not having the right partners who can enable it for you. You don't necessarily need to build your own AI to use it. A lot of these institutions could leverage it through their Microsoft engagement. They just need a blueprint and answers to some of these scary questions, such as, what are the guardrails? What are the controls? 

One in two banks is already investing in AI in some way. Adoption into some core operations still slightly lags, but I think the adoption rate will be much higher in the next few months.

How can community banks begin weaving AI into their operations? 

At BNY, we jumped onto this AI journey several years ago. Today, all employees have access to our AI tool, which is called Eliza, and 50% of our non-engineering employees are using it to build their own agents. We got there with simple use cases. Informing the team: AI is available. You can use it to help write emails. You can use it to review your notes. It’s these small adoptions in community banks. Encourage the team to use it, create an environment where it’s talked about often, and then build to some of the more complex operational uses, which come fast.

Are there areas of the bank in which community lenders should avoid using AI – perhaps customer service?

No, I don't think there's any spots they shouldn't, but I think there's some areas where it might be easier to integrate on Day One. I don’t think interrupting client engagement is the right place to start. But if you can improve some back office operations, create efficiency, it might be the ideal area to start.