Senior executives from America’s largest banks gathered at a Goldman Sachs financial services conference in New York, delivering a unified message: artificial intelligence has graduated from experimental projects to become embedded in their daily workflows. Leaders from JPMorgan Chase to Citigroup shared concrete productivity metrics, particularly in software development, customer service, and operational processes.
JPMorgan Reports Doubling of Productivity Rates
However, these same executives didn’t shy away from addressing the elephant in the room. As AI tools enable teams to accomplish more with fewer resources, workforce adjustments appear inevitable once productivity gains stabilize and workflows reach maturity.
JPMorgan Chase stands out as one of the clearest success stories in early AI adoption. Marianne Lake, who leads the bank’s consumer and community banking division, revealed that productivity in AI-enabled units has climbed to approximately 6%, double the 3% rate observed before deployment. Looking ahead, Lake suggested that operational roles could see productivity improvements reaching 40% to 50% once AI becomes fully integrated into routine workflows.
Rather than pursuing broad experimentation, the bank has adopted a measured approach. Its internally developed large language model platform allows employees to draft and summarize content within a secure environment, with strict data access controls. Management emphasizes that workflow redesign, not merely tool access, has driven these improvements.
Wells Fargo Acknowledges Coming Workforce Changes
Wells Fargo CEO Charlie Scharf stated that while the bank hasn’t reduced headcount directly due to AI implementation, productivity has increased substantially. “We’re accomplishing significantly more work,” Scharf noted, adding that management anticipates identifying areas where fewer employees will be needed as efficiency continues climbing.
Scharf also pointed to internal budget projections for 2026 that already indicate a smaller workforce, even before accounting for AI’s full impact. Rising severance costs, he observed, signal that preparations for future workforce transitions are already underway.
Citigroup and PNC: Acceleration Rather Than Disruption
At Citigroup, incoming CFO Gonzalo Luchetti reported a 9% productivity improvement in software development, reflecting increased use of AI copilots for coding, testing, and documentation. He also highlighted gains in customer service, where AI enhances self-service options and supports agents during live interactions.
PNC CEO Bill Demchak framed AI as an accelerator of existing trends rather than a disruptive force. The bank’s headcount has remained relatively flat for roughly a decade despite business growth, driven by automation and branch optimization. According to Demchak, AI will likely push this long-running shift further.
Goldman Sachs Links AI to Workflow Redesign
Public reporting indicates that Goldman Sachs has been deploying AI through its internal “OneGS 3.0” program, focusing on sales processes, client onboarding, and process-intensive functions such as lending workflows, regulatory reporting, and vendor management. These initiatives are unfolding alongside job cuts and slower hiring, directly connecting workflow redesign to staffing decisions.
Where Early Results Are Materializing
Across Wall Street, the strongest initial productivity gains are appearing in roles that depend heavily on documentation, follow repeatable procedures, and operate within defined parameters. In these areas, generative AI can dramatically reduce the time required for information retrieval, material summarization, content drafting, and moving work through approval chains.
Common areas experiencing early impact include operations, software development, customer service, sales support, client onboarding, and regulatory reporting. AI outputs in these functions typically undergo human review before finalization.
Governance Remains the Primary Constraint
For banks, enthusiasm isn’t the limiting factor, control is. U.S. regulators demand robust oversight of models, and these expectations extend to AI systems. Existing model risk management frameworks already apply, pushing banks toward AI architectures that can be audited, monitored, and explained.
In practice, this means AI rarely operates independently. Prompts and outputs are logged, performance is monitored for drift, and humans remain accountable for high-stakes decisions such as lending approvals, dispute resolution, and official reporting.
Productivity First, Employment Questions Later
Executive comments suggest a phased transition is underway. During the initial phase, headcount remains relatively stable while output increases as AI tools spread across teams. The second phase begins once those gains become consistent enough to influence staffing plans through attrition, role modifications, or targeted reductions.
Signals from institutions like Wells Fargo suggest that some banks are approaching this second stage.
