Thryv Turns Directory Assistance Into a 24/7 Channel with Voice AI

Overview

Thryv (NASDAQ: THRY) powers marketing and customer operations for more than 100,000 small businesses — with a large call center footprint behind it. As the company pushed to modernize customer interactions with AI, the challenge wasn’t capability. It was deployment: how to introduce Voice AI into production without risk.

The answer was a controlled starting point. Thryv launched a Voice AI agent on a New Zealand directory assistance service — a deterministic, high-volume use case where performance could be measured from day one.

What began as a contained test quickly became a live production system.

Problem

Thryv’s directory assistance service depended on human agents operating within standard business hours. Demand, however, didn’t.

This created a structural gap: callers outside those hours couldn’t be served, but extending coverage to 24/7 wasn’t economically viable with a traditional staffing model at existing volumes. The result was missed calls, inconsistent availability, and untapped demand.

At the same time, any move to Voice AI had to meet production standards — not just answer calls, but reliably retrieve information, integrate with internal systems, support protected caller groups, and deliver a consistent customer experience.

Solution

After evaluating multiple vendors, Thryv selected Synthflow based on three factors: platform capability, predictable cost structure, and the ability to move quickly from development to production.

Rather than starting with a broad or experimental use case, Thryv deployed a Voice AI agent on a deterministic directory assistance workflow — a controlled environment where accuracy, integrations, and performance could be validated with confidence.

The agent was designed to retrieve and deliver information in real time via API integrations, while maintaining a hybrid model: complex or edge-case requests could seamlessly escalate to human agents when needed. This ensured reliability without sacrificing coverage.

Just as importantly, the platform enabled Thryv’s internal team to move fast. Engineers — without specialized telephony expertise — were able to build, test, and iterate in a development environment, then push updates into production without long cycles or external dependencies.

The Results

Thryv’s first production Voice AI deployment delivered value quickly — and created momentum for broader enterprise expansion.

20K+ Calls per Month

The AI agent handles over 20,000 directory assistance calls monthly, operating at consistent, production-level volume without capacity constraints or performance degradation.

100% Call Answer Rate

All inbound calls are now answered, including after-hours and periods of concurrent demand — eliminating missed calls that were previously unavoidable under a human-only model.

+4.4% More Calls Answered

Calls that previously went unanswered outside business hours are now fully served, converting previously lost demand into active customer interactions.

~1 Month to Production

From initial development to live deployment, the system was implemented and validated in just over a month — accelerating time-to-value without extended integration cycles.

Conclusion

Thryv’s directory assistance deployment shows how Voice AI moves from pilot to production: start with a controlled, deterministic use case, prove reliability at scale, and expand from there.

What began as a contained rollout is now a foundation for broader transformation. With performance validated, Thryv is extending Voice AI into intelligent IVR and knowledge-driven automation to improve resolution and modernize customer interactions across the enterprise.