Banking, Member Exclusive

The Big Five banks are putting AI agents to work, mostly out of customers’ sight

  • Top U.S. banks deploy AI agents mainly for employees and back-office work, with humans reviewing output and customer-facing autonomy rare.
  • JPMorgan bars third-party agents from customer accounts; Citi, Wells Fargo and Goldman roll out internal agents to tens of thousands of staff.
close

Email a Friend

The Big Five banks are putting AI agents to work, mostly out of customers’ sight

For two years, large banks talked about AI assistants that answer questions. In 2026 the vocabulary has shifted to agents: software that carries out multi-step work with some independence. At the five largest U.S. banks by assets, the pattern is consistent. Agents go to employees and back-office teams first, humans still review the output, and autonomous software touching customer accounts remains the exception.

JPMorgan Chase: longer-running agents, a locked front door

JPMorgan told CNBC in June that it plans to roll out agents later this year that can work on their own for hours rather than minutes. Derek Waldron, the bank’s Chief Analytics Officer, said the industry has entered the era of long-running autonomous agents, and expects them to eventually stay coherent for days and weeks, acting more like team managers who delegate than individual workers. The bank has also tied AI tools to a 20% lift in private banking gross sales.

JPMC’s researchers are testing the investing side too. In July, a team reported building eight agents, powered by models from OpenAI and Anthropic, that moved between asset classes based on market regimes. All eight beat a 60/40 portfolio on a risk-adjusted basis in backtests, though the authors stressed these were simulations and said they remain wary of handing allocation decisions to an agent.

Meanwhile outside agents face a firmer boundary, with CEO Jamie Dimon saying in October  that JPMorgan won’t let third-party agents operate customer accounts, warning of a “liability transfer” for customers who authorize them. The bank is piloting its own tool, Smart Cash, instead.

Bank of America: Erica scales, with people in the loop

…


 

0 comments on “The Big Five banks are putting AI agents to work, mostly out of customers’ sight”

10-Q, Member Exclusive

Wall Street wants tokenized stocks. Some of its biggest companies don’t want them traded this way.

  • JPM, Wells Fargo, Goldman Sachs, and dozens of other public companies are pushing back against having their shares traded on OKXICE’s proposed tokenized stock platform.
  • The resistance is about who controls the trading, whether prices and liquidity remain reliable, and what issuers gain when third parties turn their shares into tradable tokens.
Sara Khairi | November 16, 2026
Member Exclusive, Opinion

The financial services industry is getting better at selling to itself. Its customers still have problems, though.

  • Financial services is getting better at building and selling technology to itself, but that doesn’t necessarily mean customers are getting better experiences.
  • Innovation through partnerships and new products should help customers get things done faster, easier, or more effectively.
Sara Khairi | November 13, 2026
5 questions, Member Exclusive, Payments

How Gen Z’s expectations are pushing treasury toward real-time finance feat. Citizens

  • Gen Z is raising the bar for treasury, expecting payment visibility, transparency, and control to keep pace with real-time money movement.
  • Citizens is modernizing banking connectivity, but instant payments also demand earlier fraud checks and better information-sharing across banks, fintechs, and payment providers.
Sara Khairi | November 12, 2026
Business of Fintech, Member Exclusive

How Block is turning Square and Cash App into one local growth network

  • Block has connected Square’s seller network with Cash App’s consumer base via Neighborhoods, giving local businesses a way to turn discovery into direct orders, customer relationships, and potentially repeat visits.
  • Neighborhoods is designed to bypass the traditional marketplace model. Its growth will depend on whether Cash App can generate meaningful customer demand and repeat spending without adding another layer of cost for sellers.
Javarya Kamran | November 10, 2026
10-Q, Member Exclusive

Can Block turn an agentic voice layer into a merchant growth tool?

  • Block is betting voice AI could help Square become more valuable to merchants by improving the workflows that drive sales and service.
  • Turning this ambition into merchant value will depend on execution: accurate orders, useful recommendations, and measurable returns that justify the added cost.
Sara Khairi | November 09, 2026
More Articles