Every day, businesses deploy AI agents to handle critical conversations with customers and partners—but without the right infrastructure to ensure their reliability, they're flying blind. To solve this challenge, we're excited to announce that Coval has raised $3.3M led by MaC Ventures, with participation from General Catalyst, Y Combinator, Fortitude Ventures, Pioneer Fund, Lombard Street Ventures, and other great Angels, to build the foundation for trustworthy AI agents.

📣 Check out our Announcement on TechCrunch here 📣

Our Why

The parallels between self-driving cars and conversational AI agents are striking. Both are autonomous systems navigating complex, multi-step journeys with countless possible paths between start and end points. Just as a self-driving car must navigate from pickup to destination through varying conditions, a voice agent must guide a conversation from initial greeting to final resolution through diverse user interactions.

Our founder and CEO, Brooke Hopkins, brings unique insight to this challenge. At Waymo, she led the evaluation infrastructure team, building systems that could simulate millions of possible journey variations. Her experience revealed a crucial insight: a self-driving car is essentially an agent on wheels, and the same principles for evaluating its performance apply remarkably well to conversational AI.

When people say they don't want a future with voice-enabled workflows, what they really mean is they don't want to interact with unreliable voice AI. By applying the rigorous simulation and evaluation principles from autonomous vehicles, we're building the infrastructure needed to make voice agents as reliable as self-driving cars.

Where We Are Today

Through conversations with hundreds of engineering teams, we've identified a clear pattern: while companies are rapidly developing AI agents, they lack the infrastructure to deploy them confidently at scale. Engineers spend countless hours on manual testing, yet still miss critical edge cases that could damage customer trust.

Coval is changing this paradigm. We're bringing battle-tested simulation techniques from autonomous vehicles to ensure AI agents perform reliably at enterprise scale. Our platform enables companies to rigorously evaluate their AI systems through automated testing, similar to how each code change at Waymo was tested in a virtual environment before deployment.

What We're Building

At Coval, we're creating more than just another AI testing tool—we're building the foundation for enterprise-grade voice and chat AI reliability. Our platform applies probabilistic testing approaches proven in robotics to the unique challenges of voice and chat interactions.

Comprehensive Testing Framework

Our testing suite supports multiple input types including natural language scenarios, transcripts, workflow graphs, and audio files, all enriched with expected results and run across different personas and voice options to ensure consistent performance.

Advanced Conversation Analytics & Observability

Our platform provides comprehensive analytics across your entire agent lifecycle. We track everything from technical performance (latency, audio-text sync, speech rate) to conversation quality (topic adherence, workflow compliance, tool call accuracy), with support for custom metrics. Our production monitoring system enables automated evaluation of every live conversation against your performance benchmarks, with real-time alerts when your agents deviate from expected behavior.

End-to-End Development Support

From pre-production automation through GitHub Actions to production monitoring with real-time metrics, we integrate seamlessly into your development workflow, complete with automatic tracking and Slack notifications.

Trust and Transparency

Our shareable trust center helps you demonstrate agent reliability, with human-in-the-loop review capabilities and comprehensive API access for custom integrations.

Instead of simple input-output testing, we evaluate agents across the full spectrum of possible conversations, identifying edge cases and ensuring natural, consistent performance. Whether you're building customer service agents, sales assistants, or any other voice-enabled application, Coval provides the infrastructure you need to deploy with confidence.

Join The Future

While we're starting with voice and chat agents, our vision extends far beyond conversational AI. We believe the principles we're developing for evaluating autonomous conversations will become fundamental to testing all agentic systems. Just as our evaluation methods evolved from self-driving cars to voice AI, we're building infrastructure that will support the next generation of autonomous agents across industries.

We're using this funding to expand our team with top talent who share our vision of reliable, trustworthy AI agents. The stakes are high—autonomous systems are poised to generate trillions in economic value, but only if users can trust them. With our unique expertise in evaluation infrastructure, Coval is positioned to become the foundation for testing and validating agentic systems of all kinds.

If you're passionate about shaping the future of autonomous systems and building the evaluation infrastructure that will power the next wave of AI, join us. We’re growing our engineering team. 

Apply here to reach out to brooke@coval.dev.