PenFed Credit Union's deployment of Salesforce Agentforce has delivered measurable operational gains that extend beyond typical pilot metrics: $1.6 million in annual savings, 10% reduction in call handle time, 50% reduction in after-call work, and 40% fewer held calls. The credit union's approach centres on Agent Wingman, a real-time transcription tool that keeps human agents in control by requiring approval before updating customer records. This design choice matters operationally because it creates a compliance checkpoint whilst solving the core problem agentic AI addresses—the work around the conversation rather than the conversation itself. PenFed's CEO appearing on Salesforce's earnings call signals this is production CX infrastructure, not a proof-of-concept, and reflects how the credit union consolidated from 400 platforms to 12 strategic partners, running its entire operation—call centre, mobile, web, and branches—on Salesforce. The consolidation itself drove efficiency gains; Schenck framed tech sprawl as "a tax on innovation, it's a tax on speed, and it's a tax on security."
The broader context reveals Agentforce has crossed into mainstream adoption at scale. Salesforce reported $1 billion in ARR, 4 million autonomously handled inquiries over 15 months, and 205% year-over-year revenue growth. Customers like Florida Prepaid handling 75% of business-hour calls autonomously and Vivino running support for 74 million users with 37 human reps demonstrate the technology works across verticals. Yet PenFed's model raises a critical question for teams already running Agentforce or considering deployment: does your platform architecture support the 360-degree customer record that makes agentic AI operationally valuable across channels, or are you building isolated agent capabilities that create new silos? The answer determines whether agentic AI becomes a contact centre efficiency tool or a strategic asset that strengthens relationship intelligence over time.
For CX leaders evaluating agentic AI investments, PenFed's framework offers a pragmatic template. The deployment separates meaningful outcomes—reduced after-call work, faster resolution—from vanity metrics. It treats human approval as a feature, not a limitation, particularly in regulated industries where accuracy and compliance cannot be delegated. The consolidation narrative also matters: PenFed's gains came partly from eliminating platform friction, suggesting that teams running multiple best-of-breed tools may see diminishing returns from agentic AI until they address underlying integration debt. The question becomes whether your organisation has the platform discipline to make a unified customer record operationally usable, because without it, even sophisticated AI agents cannot deliver the continuity that justifies their deployment cost.
Salesforce used its most recent earnings call to spotlight PenFed Credit Union’s Agentforce deployment in its contact center. James Schenck, President and CEO at PenFed Credit Union, said the program will save $1.6 million annually and reduce call handle time by 10%. Schenck also reported a 50% redu
Agentforce has surpassed $1 billion in annual recurring revenue as adoption of the company’s agentic AI platform continues to accelerate across enterprise customers. Since going live on help.salesforce.com and its 1-800-NO-SOFTWARE support line 15 months ago, Agentforce has autonomously handle