Banks introduce more powerful agents this year

JP Morgan Chase It plans to introduce an artificially intelligent agent later this year that can operate autonomously for much longer than existing versions, marking a new milestone in enterprise AI adoption, CNBC exclusively reported.

Derek Waldron, JPMorgan’s chief analytics officer, told CNBC in an interview that AI agents are evolving from tools that complete single tasks to digital workers that manage multi-step workflows and different software programs.

“We have entered a long-term era of autonomous agents,” Waldron said. This means that “instead of just running for 2-3 minutes to carry out human goals and instructions, agents can run for 1-2 hours.”

Long-lived agents have already emerged over the past year, with examples like Anthropic’s Claude Code and OpenClaw going viral. But JPMorgan’s implementation plans suggest the technology is close to clearing the security and governance hurdles that have slowed adoption within large corporations.

JPMorgan, run by CEO Jamie Dimon since 2006, is the largest bank in the United States by assets and has an annual technology budget of nearly $20 billion.

While much of the conversation around generative AI focuses on model intelligence, Waldron said technology leaders are increasingly looking at other questions. In other words, how long can an AI system operate effectively before human intervention is required?

This concept, which Waldron called “intellectual coherence,” is aided by improvements in how AI models reason, allowing AI to become “more of a team manager than an individual worker,” he said.

“Just as humans function, team managers can analyze problems and delegate activities, allowing teams to work longer hours to perform more complex tasks,” Waldron said.

Recent advances that help agents do more complex jobs include the ability to write code, control web browsers, and interact directly with desktop software, he said.

The long-running agent is not yet ready for enterprise use due to security concerns, but Waldron said its arrival is not far off. “It will be introduced in 2026.”

Eventually, he said, AI agents will maintain coherence over “hours, then days, then weeks.”

While the productivity gains from AI are most evident in software development and back-office tasks, Waldron said AI is increasingly playing a role in generating revenue.

For example, in private banking, AI systems scrutinize market activity, customer positions, and research overnight, allowing bankers to focus on interacting with customers.

He believes these tools will increase the bank’s total revenue by 20% and ultimately allow individual bankers to increase their customer reach by up to 50%.

Dimon has made clear that some of the workforce will be replaced by AI, and said the company is preparing to train and redeploy employees affected by the changes.

But Waldron added that while many companies initially approached AI as a cost-cutting tool, they are increasingly recognizing its revenue-enhancing potential.

“For companies to win with AI, it’s not just about cutting the maximum number of jobs,” he said. “The key is to try to create a sustainable competitive advantage.”

Waldron said the way banks think about building software and buying from outside vendors has also changed. JPMorgan is now looking more closely at whether it can build capabilities in-house, which could put pressure on some traditional vendors, he said.

“There’s definitely less of a bunker around certain types of software companies than there has been in the past,” he said.

—CNBC’s Gabriel Fonrouge contributed to this report.