The biggest bank in America is preparing to unleash a new generation of artificial intelligence, digital workers that don't just answer questions or sort data but manage entire workflows on their own, for hours at a stretch, across multiple software systems. JPMorgan Chase plans to deploy these AI agents later this year, a move that signals how fast Wall Street's largest institution is moving to replace human labor with machine autonomy.
Derek Waldron, JPMorgan's chief analytics officer, laid out the bank's ambitions in an exclusive interview with CNBC. The shift, as Waldron described it, marks a leap from AI tools that handle a single task to something closer to a digital employee, one that can parse problems, delegate sub-tasks, and maintain what he called "intellectual coherence" over sustained periods.
Waldron drew a sharp line between where AI stood recently and where JPMorgan is headed. Existing AI tools typically run for a few minutes, completing a narrow assignment before handing control back to a human. The new agents are different.
"We've entered now the era of long-running autonomous agents. That means that agents don't just run for two or three minutes to carry out a goal or some instructions of a human, they can run for an hour or two."
That's the near-term target. Waldron said the trajectory points further still, agents that remain coherent for "multiple hours, then days, then weeks." And he confirmed a concrete timeline: "We will have those in 2026."
The analogy Waldron offered is worth noting. He compared the next-generation agents not to individual workers but to team managers.
"Just like how people function, team managers can parse out a problem and delegate activities, and teams can run for a lot longer to do more complex things."
In plain terms, JPMorgan is building AI systems that don't just do a job, they organize jobs. They break a problem into parts, assign sub-tasks, and coordinate the results. That's a qualitative change from the chatbot era, and it carries real consequences for the people currently doing that coordination work.
JPMorgan Chase is not a company that dabbles. The bank, the largest in the United States by assets, carries a nearly $20 billion annual technology budget. Under CEO Jamie Dimon, who has run the firm since 2006, JPMorgan has consistently positioned itself at the front edge of financial technology adoption.
Waldron told CNBC that the bank now looks more closely at whether it can build capabilities in-house rather than buying from outside software vendors. That's a telling shift. For years, banks licensed enterprise software from specialized firms. Now, armed with massive budgets and increasingly capable AI platforms, JPMorgan appears ready to cut out the middleman.
Waldron was blunt about what that means for the software industry.
"The moat around certain types of software companies is most certainly diminished versus where it was in the past."
When the nation's biggest bank says your competitive moat is shrinking, that's not an abstract observation. It's a warning. Enterprise software vendors that once counted on long-term contracts with major financial institutions should be paying close attention.
This isn't all projection and promise. Waldron pointed to measurable results from AI tools already in use. In private banking, the bank has seen a 20 percent increase in gross sales because of AI-driven capabilities. He said AI tools could eventually allow individual bankers to expand their client coverage by as much as 50 percent.
Those are not small numbers. A 20 percent sales bump in private banking, a high-margin business line, translates directly to the bottom line. And if a single banker can handle 50 percent more clients, the math on headcount gets uncomfortable fast.
Waldron framed the bank's approach carefully, though. He insisted the goal isn't mass layoffs.
"For enterprises to win with AI, it's not about cutting the maximum number of jobs. It's all about trying to create a sustainable competitive advantage."
That's a diplomatic way to put it. But Dimon himself has acknowledged what everyone in the industry already understands: some workers will be displaced. CNBC reported that JPMorgan is preparing to train and redeploy employees impacted by the changes. How many employees, and into what roles, remains unclear.
JPMorgan isn't operating in a vacuum. Over the past year, long-running AI agents have emerged from several corners of the technology world. Anthropic's Claude Code and a tool called OpenClaw, which reportedly went viral, represent early examples of the kind of autonomous, sustained AI work that Waldron described. The technology is maturing quickly, and major institutions are racing to harness it before competitors do.
What makes JPMorgan's move significant is scale. Plenty of startups and mid-size firms experiment with AI agents. But when a bank with nearly $20 billion in annual tech spending commits to deploying autonomous agents across its operations, the ripple effects reach far beyond one company. Vendors, regulators, competitors, and employees across the financial sector all face a changed landscape.
For all the detail Waldron provided, key questions remain open. JPMorgan has not disclosed which departments or business units will receive the new agents first. The security and governance protocols required before deployment, a non-trivial concern when autonomous systems operate inside a systemically important bank, have not been publicly detailed. And the specific number of employees who may be displaced has not been disclosed.
Those gaps matter. Autonomous AI agents that run for hours across multiple software systems will inevitably touch sensitive customer data, trading systems, and compliance processes. The governance framework around those agents will determine whether this is a well-managed transformation or a high-speed train with no brakes.
Waldron's comments about building in-house deserve a second look. For decades, enterprise software companies built their businesses on the assumption that big banks would always need specialized vendors. JPMorgan's shift toward internal development, powered by AI, threatens that assumption at its foundation. If the biggest bank in America decides it can build what it used to buy, smaller software firms face an existential question.
And JPMorgan won't be the last to reach that conclusion. Every major financial institution with a serious technology budget will be watching what Waldron's team deploys this year. If the results match the private banking numbers, 20 percent sales growth, 50 percent more client coverage per banker, the pressure to follow will be intense.
Waldron says this isn't about cutting jobs. Dimon says some workers will be displaced. Both statements can be true at the same time, but the tension between them is real. When AI agents can manage workflows autonomously for hours, coordinating tasks across systems the way a mid-level manager does today, the organizational chart gets shorter. That's not a prediction. It's arithmetic.
JPMorgan says it will retrain and redeploy affected workers. That's the right instinct, and the bank deserves credit for stating it publicly. But retraining programs are only as good as the jobs waiting on the other end. If the same AI that displaced the first round of workers can also handle the roles those workers were retrained into, the cycle repeats.
None of this means the technology is wrong or that JPMorgan is wrong to pursue it. Competitive markets reward efficiency, and a 20 percent sales increase is hard to argue with. But the people who built their careers doing the work these agents are about to absorb deserve straight talk, not corporate euphemism.
When the nation's largest bank builds digital team managers that work around the clock and never ask for a raise, the question isn't whether the workforce changes. It's whether anyone in charge is honest enough to say how much.