Case Study
How a UK fintech cut first-response time from 14 hours to 4 minutes
Before Voxys Connect, this neobank's support team was drowning in a queue split across three tools: a help-desk for email, a separate chat product for the app, and a Twilio-based SMS console for fraud alerts. Each had its own inbox, its own metrics, and no shared customer history.
"We had agents jumping between four tabs to answer a single ticket. Median first-response was 14 hours. Anything sub-4-hour felt impossible." — Head of Customer Operations
The setup
The team migrated all three inbound channels (email, in-app chat, SMS) into a single Voxys workspace, added WhatsApp as a fourth channel for fraud notifications, and turned on Captain AI for tier-1 triage.
What changed
Hour 1: Captain AI handles password resets, card-freeze requests, balance lookups, and basic transaction questions — directly, without involving an agent. About 62% of incoming volume falls into these categories.
Hour 2 onward: For anything Captain AI can't resolve (~38% of volume), the conversation reaches a human agent with a one-paragraph summary, the customer's account context, and the last 90 days of interaction history pre-loaded. No tab-switching.
The numbers, six weeks in
- Median first-response time: 14 hours → 4 minutes (99% reduction)
- Tier-1 deflection: 62% of conversations fully resolved by Captain AI
- CSAT: 4.1 → 4.6 out of 5
- Cost per ticket: down 41%, primarily because the average tier-1 ticket no longer needs an agent
The detail that surprised them
The biggest CSAT lift came not from speed, but from the fact that customers no longer had to repeat themselves. When an agent finally jumped in, they had the full conversation history. "Stop making me re-explain my situation" was the #1 complaint before — now it's basically gone.
What's next
The team is rolling out Voxys for their SMB onboarding flow next. The goal: convert 20% more sign-ups by replying to first messages in seconds, not days.
