How Coval is Making Voice AI Reliable in Production

Coval, building AI evaluation for voice AI, led by Brooke Hopkins and the team, has raised a $28 million Series A, led by Norwest, with participation from Base10 Partners, Twilio Ventures, Y Combinator, MaC Ventures, and Swift Ventures, which brings the total raised to $31 million since the company was founded in 2024.

Building the Infrastructure for Reliable Voice AI.

Before founding Coval, CEO Brooke Hopkins worked at Waymo, where a central challenge was determining when a self-driving system was safe enough for real-world deployment. Achieving that confidence required extensive testing infrastructure, large-scale simulations, regression testing, and production monitoring to ensure reliability in a probabilistic system.

Hopkins believes voice AI is approaching a similar inflection point. As enterprises increasingly deploy AI-powered voice agents, success will depend not only on model capabilities but also on the infrastructure needed to test, evaluate, monitor, and improve those systems in production. Coval is building that infrastructure, helping organizations deploy voice agents with greater confidence and reliability as adoption scales.

Why Coval Exists?

Voice AI is already being deployed across customer support, scheduling, collections, and patient intake, handling millions of real-world conversations every day. Yet many voice agents that perform well in controlled demonstrations struggle once exposed to the unpredictability of production environments.

Factors such as accents, interruptions, background noise, and unexpected user behavior can significantly impact performance. Coval believes this is less a model problem and more an infrastructure challenge. The company is focused on building the testing, evaluation, and monitoring systems needed to ensure voice agents operate reliably, consistently, and compliantly in real-world settings.

Why Now?

Investment in voice AI is accelerating rapidly, with billions of dollars flowing into the sector as enterprises increasingly adopt AI-powered agents. As more voice agents move from pilots to production, the need for reliable testing, evaluation, and monitoring infrastructure becomes increasingly important.

Coval is focused on this emerging infrastructure layer, helping enterprises deploy voice AI systems with confidence. The company’s recent Series A was led by Norwest, whose partner Scott Beechuk believes voice will become a primary interface for human-AI interaction, creating a significant opportunity for the infrastructure platforms that support it.

For founder and CEO Brooke Hopkins, the mission builds on lessons learned at Waymo: trust in AI systems comes not only from model performance, but from the infrastructure that ensures reliability at scale.

What Comes Next?

The new funding will support expansion of Coval’s sales, solutions engineering, and product teams as the company scales its voice AI reliability platform. Product investments will focus on richer simulation environments, broader testing across accents and real-world audio conditions, deeper integrations with enterprise technology stacks, and enhanced monitoring and human review capabilities.

By combining testing, evaluation, and production monitoring, Coval aims to help enterprises deploy voice agents with greater confidence as adoption accelerates. For founder and CEO Brooke Hopkins, the mission remains clear: building the infrastructure layer that makes voice AI reliable at scale.