Zero unsanctioned outputs in 90 days.
Bridge API in front of every customer-facing LLM call. Ninety days in production with zero unsanctioned outputs. Audit trail reviewed and approved by the credit union's primary regulator at day 60.
Affinity Federal Credit Union operates across multiple states with a member base concentrated in regulated industries. The credit union had piloted AI in member services in 2023 with poor results: three incidents of unsanctioned model outputs in the first 60 days, triggering an internal AI deployment moratorium.
By early 2024, the credit union's risk team scoped a re-deployment. The constraint was absolute: zero unsanctioned outputs. The regulator's expectations had hardened in the intervening months, and any further incident would terminate the AI program entirely.
Bridge API was selected as the routing primitive. Every customer-facing AI interaction would route through Bridge with strict behavioral policy enforcement, full audit trail, and regulator-reviewed configuration.
Replacing routing FTEs without breaking clinical trust.
Affinity needed an AI deployment with a zero-incident track record in 90 days, with the audit trail required for regulator approval. The deployment had to integrate with the credit union's existing member services platform without breaking the existing user experience for service representatives.
Eight weeks from scope to scale.
Weeks 1-3 covered integration with Affinity's member services platform and Bridge API policy configuration. Weeks 4-12 ran the 90-day production window with 100% of customer-facing AI calls routed through Bridge. The audit trail was reviewed and approved by Affinity's primary regulator at day 60. Day 90 closed the window with zero unsanctioned outputs.
Eight-week measured outcomes.
“The audit trail did the work. Our regulators saw it once and the conversation was over. That's not what we've ever experienced with AI deployments before.”