A financial services company wanted its customers to be able to pay their bills by phone at any hour, without tying up agents and without compromising card data security. Voxivra designed and deployed an AI voicebot for secure payments with multi-factor authentication and direct integration with the billing and CRM systems.
Context
Phone payments made up a significant share of the calls reaching customer service. Every call followed the same pattern: identity verification, balance lookup, payment entry, confirmation. Simple on paper, but slow, costly and limited to business hours.
Challenge
Several constraints had to be reconciled:
- Payment security: card data could not be exposed to agents, recordings or logs, in line with PCI-DSS requirements.
- Reliable authentication: verify customer identity without making the experience painful, while limiting fraud risk.
- Up-to-date data: balances and account status had to come from the billing system in real time.
- 24/7 availability, including month-end peaks.
- Continuity with agents: when a call was transferred, the context had to follow it.
Solution
The approach relied on a modular architecture built around the telephony platform and voice AI components.
Conversational flow
The voicebot greets customers in natural language, understands their intent (pay a bill, check a balance, something else) and guides them step by step. Speech-to-text (STT) and neural text-to-speech (TTS) keep the exchange natural, while a language understanding engine handles rephrasing and unexpected requests.
Multi-factor authentication
Identification combines several factors: caller ID matching, account-related verification details and a one-time code. The verification level is matched to the requested operation.
Secure payment capture
Card details are entered on the phone keypad (DTMF). This sequence is isolated from the rest of the conversation: tones are neither transcribed nor recorded, and the data goes straight to the payment processor. This design reduces the scope exposed to PCI-DSS requirements.
Billing and CRM integration
The voicebot queries the billing system for balances and due dates, posts the payment, then logs the interaction in the CRM automatically. When a transfer to an agent is needed, the agent receives the verified identity and the reason for the call.
Telephony hardening
Encrypted signaling and media (TLS, SRTP), protections against SIP trunk abuse and operation logging round out the setup. These practices are covered on our VoIP security page.
Results
- 65% lower costs.
- 4-minute handling time, divided by 3.
- 92% customer satisfaction.
- 24/7 payments, independent of business hours.
- PCI-DSS compliance for the phone payment process.
- Full CRM integration: every interaction is logged with no manual data entry.
Key takeaways
- Security starts with the architecture. Isolating DTMF capture and keeping card data out of recordings is simpler and more reliable than masking it after the fact.
- Authentication should be proportionate. Matching controls to the risk of each operation protects both security and customer experience.
- Integration is where the value lies. A voicebot that reads from and writes to billing and CRM systems resolves requests instead of just qualifying them.
- Transfers still matter. A smooth hand-off to an agent, with context, keeps satisfaction high for cases outside the flow.
Working in the same industry? Explore our approach to voice solutions for financial services or our custom AI voicebot service.