JPMorgan Chief Analytics Officer Says Bank’s Next AI “Agents” Aim to Run for Hours Without Human Help

JPMorgan Chase plans to roll out next-generation artificial intelligence agents later this year that can keep working autonomously for far longer than today’s typical AI assistants, according to reporting that quotes the bank’s chief analytics officer.

In an interview described by CNBC, Derek Waldron said the shift marks “the era of long-running autonomous agents.” He contrasted these systems with short-lived automation, saying the newer agents are designed to run for an hour or two to pursue a goal or instructions from people, rather than carrying out a task only briefly before waiting for oversight.

AI agents, sometimes called “digital workers,” are an evolution from chatbots and single-purpose automation. Instead of completing one narrow task, they are intended to manage workflows across multiple steps and, crucially, across disparate software systems that firms rely on to run operations. The long-running capability is the practical hurdle: organizations need confidence the systems will keep functioning reliably and within rules while operating without constant human direction.

CNBC also reports that JPMorgan is already seeing measurable business impact from its broader AI push, including a 20% increase in private banking gross sales. The bank’s leadership sees AI as a lever not only for productivity, but for expanding the reach of services to clients, with the potential, as described in the report, to allow bankers to cover up to 50% more clients through improved workflow support.

Long-running agents are drawing attention partly because they reintroduce risks that shorter tools can minimize. As MIT’s Sloan School of Management explains in its educational overview, “agentic AI” refers to systems that can take actions toward goals, which changes how organizations must think about controls, monitoring, and the boundary between assistance and execution. JPMorgan’s reported focus on longer autonomic runs suggests the bank believes those controls are close enough to workable for real deployment.

JPMorgan, the largest U.S. bank by assets, typically moves cautiously on technology that touches regulated data and customer-facing processes. The reported mention of governance and security hurdles points to a recurring theme in corporate AI rollouts: companies may be able to use AI for advice or drafting quickly, but letting AI execute tasks over time inside core systems requires additional safeguards and testing.

Still, JPMorgan has not, at least in the reporting available here, disclosed specifics on where the agents will be used first, what kinds of permissions they will have, or how performance will be measured beyond the executive’s high-level statements. The bank also did not describe in this material whether the agents will be limited by industry rules, what audit logs or human-approval steps will be required, or how often teams will intervene during longer runs.

What to watch next is how JPMorgan structures the launch: which business lines get first access, what governance model accompanies autonomy, and whether the bank ties agent deployments to operational metrics such as cycle times, error rates, or sales productivity. As other financial institutions experiment with AI access to platforms and workflows, the industry will be watching to see whether “long-running” becomes routine or remains confined to controlled pilot environments.