The US banking sector is experiencing the greatest technological transformation in its history, according to reports The Wall Street Journal. Bank of New York Mellon It already has “dozen” of digital employees with their own corporate credentials that work autonomously, while JPMorgan Chase deployed tools from ia to more than 200,000 employees and Bank of America Invest 4,000 million of dollars per year in artificial intelligence. This is not a futuristic promise: it is the operational reality of 2025 on Wall Street.
The Wall Street Journal It emphasizes that the concept of “Digital workers“Vary between institutions; some banks are called”AI agents”Although there is no consensus on the nomenclature, technology is consolidated in the financial sector. While Bny Mellon gives its digital agents full access to internal systems, other entities such as JPMorgan Chase explore different levels of integration and control, according to the WSJ.
Derek WaldronJPMorgan analysis director, considers that the idea of ”digital employees” helps conceptualize the role of AI, although it emphasizes that these systems differ from both human employees and traditional software, according to WSJ. Waldron pointed out that the question of how much access grant these agents must be resolved case by case.
JPMorgan Chase It has been consolidated as the undisputed leader in implementation of bank digital workers. Your platform Llm Suite It already works on the teams of more than 200,000 employees globally – more than half of its workforce of 313,000 people – with a value projection that Daniel Pinto, president and CO of the bank, raised from 1.5 billion to almost 2,000 million dollars For 2024, as reported Cio dive.
The numbers are overwhelming: the 60% of employees With access use the tools several times a day, according to data presented by Teresa HeitsenretherChief Data and Analytics Officer, during a conference with CNBC. Heitsenrether explains that development goes through three stages: Availability of models (completed), incorporation of proprietary data (in progress) and the definitive phase of autonomous agents that will perform complex multipaso tasks.
In practice, JPMorgan uses generative to create marketing content in social networksdesign itineraries of Customer trip from your agency acquired in 2022, summarize meetings For financial advisors and generate code, as detailed CNBC. Lori BeerGlobal Cio, confirmed in Aws Re: Invent that bankers and advisors already receive “ideas generated by AI to get better involved with customers.”
The financial impact is measurable: The AI tools helped To increase gross sales by 20% between 2023-2024, while during market volatility in April 2025, the systems allowed multiple applications for worried customers and increase the portfolio projected by 50% for the next three to five years, as confirmed Mary Callahan ErdoesCEO of JPMORGAN ASSET & WEALTH MANAGEMENTin statements to Reuters.
While other banks experience, Bank of America Harvest the fruits of starting early. Ericalaunched in 2018, currently handled two million daily interactions and exceeded the 2.5 billion interactions accumulated, according to Pymnts reports. But the real advance is in Erica for Employees (For employees): More than 90% of the 213,000 employees of the bank use it, which reduced calls to technical support in more than 50%, as confirmed by the institution itself in a statement in April 2025.
The internal model has evolved significantly. Hari GoparkrishnanHead of Consumer, Business, and Wealth Management Technology, explains in an interview with CIO that 60% of current interactions are proactive messages from Erica towards users, not reactive consultations. This paradigm investment – answering questions to anticipate needs – marks the Difference between basic chatbots and true digital assistants.
Bank of America will allocate 4,000 million dollars in 2025 to AI and new technologies, which represents a third of its total technological budget of 13,000 million dollars, he reported Pymnts. Aditya BhasinBank’s CTO, reports that developers obtain improvements of 20% in efficiency when using coding assistants with AI, while employees save “tens of thousands of per year” in preparation of meetings for meetings with business clients, according to official data published by the institution.
The Bank avoids unnecessary risks with generative while perfecting systems that already work perfectly.
Goldman Sachs launched in January 2025 GS AI Assistant to 10,000 employees, with plans to extend it to the entire workforce during the year, as reported CNBC. Marco ArgentiBank’s CIO and former Vice President of Amazon Web Services, has an ambitious vision that described in an exclusive interview: create an assistant that absorbs the Goldman culture and eventually “Do things like a Goldman employee, not only say things like a Goldman employee.”
The system has access to multiple models: GPT-4O, Gemini 2.0 flash, Claude 3.7 Sonnet and Open Source models, which allows you to select the optimal tool for each task, as detailed Fox Business. The functions include summarizing complex documents, translating code between programming languages, generating initial content and performing data, confirmed Goldman in an internal memo seen by Reuters.
Argenti predicts that in three to five years, the line between humans and bots will blur so much that interact with ia will be “How to talk to another GS employee”he declared CNBC. The long -term vision includes agentic behavior: the wizard will complete tasks in the name of employees following multiple steps, and generate detailed plans “in the way an experienced employee of Goldman would do,” he explained in his corporate blog Goldman Sachs.
Citigroup It deployed in December 2024 two main tools to 140,000 employees in eight countries: Citi Assist and Citi stylusas reported Reuters. Tim RyanHead of Technology and Business Enablement, describes Citi Assist as “having a superintelligent partner at hand to navigate policies and procedures commonly used in HR, risk, compliance and finance,” according to an internal memo cited by the agency.
Citi Stylus allows to summarize, compare and search for multiple documents simultaneously, with translation capabilities for analysts and Wealth Managers to translate research to specific languages preferred by customers, HR Grapevine reported. David GriffithsCTO of Citi, added in April 2025 web plugin functionality to summarize content during navigation and integration with existing chat applications, he explained Fortune.
The bank also launched Citi Squad For 30,000 developers and reports that it completed approximately 220,000 automated code reviews during the first 2025 quarter, according to Cio dive. Jane Fraser, CEO of Citi, confirmed during the call of quarterly results that these tools generate “Productivity profits”Measured while the bank invests 11.8 billion dollars annually in technology.
Wells Fargo Perhaps the most impressive achievement has not achieved without much fanfare: its assistant Fargo He managed 245.4 million interactions in 2024 – more than twice the original projections – without ever exposing delicate data of clients to language models, according to Venturebeat.
Chintan MehtaChief Data and Analytics Officer, explains in the specialized publication that they use a “privacy-first” architecture where speech is transcribed locally, the text is purified and tokeniza with internal systems before processing. Gemini Flash 2.0 drives Fargo, but smaller models as they are used internally according to specific needs, he detailed in the same interview.
The adoption of the Spanish language grew dramatically, since it represents more than 80% of the use since its launch in September 2023, according to internal data cited by Venturebeat. The system averages multiple interactions per session, indicating genuine user engagement beyond specific consultations.
Bank of New York Mellon It represents the most advanced case, according to the report of The Wall Street Journal. Leight-Ann RussellCIO of the Bank, confirmed to the WSJ that they use “dozen” of digital employees with corporate logins that directly report to human managers, working autonomously in coding and validation of payment instructions.
The bank developed two digital “people” in three months through their AI Hub: one to clean vulnerabilities in code and another to validate payment instructions, Russell detailed to the Wall Street Journal. Each person can exist in multiple instances, assigned to specific equipment to limit access to delicate information.
Russell anticipates that they will soon have their own email accounts and can communicate with colleagues through Microsoft Teams, he told the newspaper. “This is the next level,” he says according to WSJpredicting that “in six months it will be very, very prevalent.”
Mass implementation generates obvious labor tensions. Jeremy BarnumCFO of JPMorgan, instructed in May 2025 to managers “to resist the growth of personnel where possible and increase the focus on efficiency”, which suggests 10% reductions in areas such as accounts, processing and fraud services, according to Vision Times.
But the data of Evident ai index They show a nuanced reality: the hiring of AI in banking grew 13% in the last six months, with one in every 50 bank employees now working on specific roles of AI, he published Tech Report. JPMorgan, Wells Fargo and Citigroup lead the hiring of the IA talentindicating that technology creates new jobs while transforming others.
Goldman Sachs officially maintains that AI “will enhance employees to make more, It will not necessarily result in the need for less humans”, Said Argenti to CNBC. The Executive emphasizes that “The importance of having a phenomenal human workforce will be amplified.”
Alexandra MousavizadehEvident co-zo, says according to Cio Dive that “the characteristic of Top banks is that they began preparing five or six years ago with restructuring, hiring and implementation of research laboratories.” Banks that started late face a “gigantic company” to determine the right route while making cuts in other areas.
American banking giants have established a potentially insurmountable competitive advantage in digital workers. With technological budgets that exceed 10,000-18,000 million dollars annually per institution, 2,000+ specialists and platforms tested with hundreds of millions of interactions, have created digital employee ecosystems that will be extremely difficult to replicate.
For the rest of the global financial industry, the message is clear, according to analysts consulted by The Wall Street Journal: Digital workers are not the future of US banking. They are your present operation.
