Banks Power Up with AI

BNY and Goldman Deploy AI Workers With Login Credentials

BNY has gone further than any rival. The bank now runs more than 140 "digital employees" — AI agents with actual corporate directory listings, login credentials, and human managers who conduct performance reviews. BNY CEO Robin Vince described the setup plainly: "The digital employee has a login, it can actually operate in the systems, and it actually has a human manager that's responsible for training it... like a performance review, if you will." AI also writes 40% of BNY's software code, according to News 8000.

Goldman Sachs took a different path. CIO Marco Argenti spent six months co-developing autonomous agents alongside engineers from Anthropic, the AI company behind the Claude model. The agents now handle back-office compliance and accounting tasks. Argenti called it "a digital co-worker for many of the professions within the firm that are scaled, are complex and very process intensive."

1 source

UBS and Morgan Stanley Free Advisors From Paperwork

At UBS, AI agents send financial advisers daily alerts about client accounts — flagging needed trades, expiring positions, or required transfers. Richard James, UBS's head of AI product, said: "Once a financial advisor makes a decision about a transaction with the client, AI agents can trade and complete money transfers." The result: advisers have freed up 70% of their administrative desk time to spend directly with clients, according to Lee Newspapers.

Morgan Stanley is preparing to go further. The bank oversees $1.2 trillion in assets through its Morgan Stanley at Work platform. It has opened that platform to customer-built AI agents via a new technical standard called the Model Context Protocol. Head of AI for Wealth Management Koren Maranca said the bank is now "preparing these agents to start pushing reminders or recommendations to the financial advisors regarding their clients," with trials set for late summer 2026.

1 source

Bank of America's Erica Quietly Powers a 10% Income Jump

Bank of America's AI push is the oldest — and now the most profitable. Its Erica assistant, launched in 2018, has grown to more than 24 million active users. During the bank's Q2 2026 earnings call, CFO Alastair Borthwick credited "Ask Merrill" — an AI tool for financial advisers — with helping drive a 10% year-over-year increase in consumer banking net income. The bank said AI allowed staff to focus on "higher-value client interactions," according to Lee Newspapers.

1 source

Big Productivity Promises, But Fragile Foundations Underneath

McKinsey estimates agentic AI could add between $200 billion and $340 billion annually to the banking sector. But experts warn that the race is moving faster than the plumbing can handle. Monica Hovsepian of OpenText points out that only 9% of banking workflows have been rebuilt to support autonomous AI agents. Most banks still run fragmented legacy databases that are too broken to support safe, reliable AI execution — a situation she calls "AI theater."

Legal liability is also unresolved. If an autonomous agent makes a catastrophic transaction error, it is unclear who is responsible — the human manager, the compliance team, or the third-party AI provider like Anthropic. KPMG's U.S. banking sector leader Peter Torrente says firms are now redesigning every role from scratch, deciding which jobs stay human, which become hybrid, and which get handed entirely to an agent. The workforce, he said, is being "looked at end to end."