JPMorgan plans longer-running AI agents for corporate workflows

JPMorgan Chase has prepared a rollout of longer-running AI agents for use later this year. According to Chief Analytics Officer Derek Waldron, the bank expects agents to handle tasks for much longer periods.

The plan comes as large companies test AI systems that can manage more steps without constant human input.

JPMorgan prepares longer-running AI agents

In a report by CNBC, Waldron said AI agents now move beyond single tasks and short instructions. He said newer systems can manage workflows across several software programs and business steps.

“We’ve entered now the era of long-running autonomous agents,” Waldron said. He said the agents can run for one or two hours instead of minutes. Waldron said older agents usually completed a goal in two or three minutes. The new systems can continue a workflow and manage more actions before human review.

JPMorgan runs under CEO Jamie Dimon and remains the largest U.S. bank by assets. The bank also spends nearly $20 billion each year on technology. Longer-running agents still face corporate security and governance checks before full deployment. Waldron said the bank expects to have those systems in 2026.

AI systems take on longer workflows

Waldron said technology leaders now ask how long AI can work before needing human help. He called that ability “intellectual coherence” during his comments on agent performance. Model improvements have helped agents reason through longer tasks and manage related workstreams. Waldron said the technology can act more like a team manager than one worker.

“Team managers can parse out a problem and delegate activities,” Waldron said. He said teams can then run longer and finish more complex work. Recent agent tools can write code, control browsers, and work with desktop software. Waldron said those abilities help AI systems take on more complete business tasks.

He said agents could later remain coherent for multiple hours, then days, and then weeks. However, he tied that timeline to future improvements and enterprise controls.

Private banking tools lift sales

JPMorgan has already used AI tools in private banking and back-office operations. Waldron said the strongest early productivity gains appeared in software development and operations teams. The bank now sees AI helping revenue-focused roles, including private banking. In that area, systems review market activity, client positions, and research overnight. Waldron argued that those tools help bankers spend more time on client conversations. He said the bank has seen a 20% increase in gross sales from the tools. 

JPMorgan believes individual bankers could eventually cover up to 50% more clients. Waldron tied that estimate to tools that prepare information before client meetings. Dimon has said AI will displace some workers across the company. He also said JPMorgan plans to train and redeploy employees affected by those changes. Waldron said many companies first approached AI as a cost-cutting tool.

However, he said enterprises also use AI to create durable business advantages. “For enterprises to win with AI, it’s not about cutting the maximum number of jobs,” Waldron said. He said the focus should remain on sustainable competitive advantage. JPMorgan has also changed how it reviews outside software vendors. Waldron said the bank now studies whether it can build more capabilities in-house. He said some software vendors may face more pressure as companies develop internal AI tools. “The moat around certain types of software companies is most certainly diminished,” Waldron said.