TruStone Financial Credit Union

TruStone Financial Credit Union, a Minnesota-based credit union, was drowning in member communications at a scale that manual staffing could not absorb. The member services team processed roughly 5,500 conversations per week, with backlogs swelling to 1,000 unanswered messages every Monday morning because the team could not suspend the service over weekends. As Jill Shaffler, leader of TruStone's enterprise applications team, noted, the team would catch up Saturday only to face another wave by Monday. Missed inquiries reached 1,300 over a five-month window. Without a scalable solution, member experience deteriorated, staff faced chronic overload, and the credit union had no systematic way to analyze what members were actually asking.

TruStone deployed Ruth, an AI-powered conversational chatbot built on Eltropy's unified communications platform. The name was chosen through an internal employee contest, with a deliberate backstory tying Ruth to the credit union's first bookkeeper — a design decision that prioritized member familiarity over pure functionality. Jill Shaffler's team described the approach plainly: 'The technology comes second to us.' The rollout was phased: Ruth launched on the public website in July, giving staff time to tune responses and monitor containment rates before a broader November integration into the digital banking platform. That sequencing proved prescient — conversation volume doubled within the first week of the digital banking launch, validating the phased approach. Eltropy's platform provided the underlying conversational AI infrastructure, with generative AI enhancements identified as a near-term roadmap item.

Ruth delivered measurable impact across every tracked dimension within months of full deployment:

• 20% reduction in contact center call volume
• 46% of member conversations handled independently by Ruth, without agent escalation
• 9.4 / 10 member satisfaction score
• Average wait time dropped to 34 seconds
• Missed inquiries fell from 1,300 to 72 over five months
• After-hours overflow support now limited exclusively to fraud reporting

The containment rate climbed from near-zero at launch to 46%, demonstrating that AI chatbot value compounds as the model is tuned and members grow accustomed to self-service. Staff were freed from repetitive inquiry volume and able to focus on complex member needs.

• Brand the bot before you deploy it. TruStone's naming contest and historical backstory drove member adoption faster than a generic chatbot rollout would have.
• Phase rollouts to reduce launch risk. Starting on the website before digital banking let the team refine Ruth's responses at lower volume before exposure scaled.
• Containment rates grow over time. Starting near zero and reaching 46% self-service shows that early numbers understate long-term value — measure trajectory, not just launch-day performance.
• Audit your overflow triggers. Reducing after-hours support to fraud-only is a meaningful operational boundary — define yours before go-live.
• Align internal teams first. Cross-functional ownership (enterprise applications + unified communications) was central to TruStone's smooth implementation.