Company: Coval
Founder: Brooke Hopkins
Website: https://www.coval.ai/
About Coval
Coval is the testing and evaluation platform that makes voice AI agents safe to deploy in healthcare, holding every call to the standard care demands before a patient is on the line.
The healthcare voice AI market is projected to grow from roughly $468 million in 2024 to more than $3 billion by 2030. But healthcare organisations are adopting voice AI faster than they can prove it is safe for patients, and after years of spam calls and clunky chatbots, many approach automated systems with skepticism. They have a reason to be cautious – a voice agent that shares a diagnosis with an unverified caller, or misroutes someone with an urgent request, creates distrust for the patient and liability for the provider.
Brooke Hopkins started Coval in 2024, drawing on her years of experience at Waymo determining whether a self-driving car could be trusted with a passenger on the road. She saw the same need for evaluation in the emerging voice AI market. Self-driving cars can handle almost every mile of road and still fail on the rare edge case that matters most, and voice AI follows the same curve, performing well in a scripted demo and breaking down the moment a real caller interrupts or uses filler words. Brooke set out to bring this simulation first discipline to voice agents, giving healthcare organizations a way to prove reliability before, during, and after deployment.
Coval tests healthcare voice agents against clinical, patient safety, and HIPAA standards, covering appointment scheduling and medical records requests, as well as billing and insurance verification. It runs tens of thousands of scenarios before a single patient is ever on the line. Once an agent is live, Coval continues monitoring every conversation in real time and routes the hardest calls to human reviewers, so the evaluation system gets sharper with every cycle throughout the life of the agent.
Coval’s early technical challenge was making simulation realistic enough to catch failures across a wide range of scenarios, modeling how generative voices behave against background noise and interruptions, then testing across the full volume of real call traffic. The precision has paid off. Testing across that many scenarios used to take more than two months. Customers now do it in less than a week, with evaluation turnaround dropping from several weeks to roughly one day. Healthcare is now one of Coval’s fastest-growing sectors, with providers and AI vendors like Deepgram relying on advanced audio evaluation metrics to improve agent performance, safeguard patient information, and reduce the risk of clinical errors.
This momentum carried into a $28 million Series A, led by Norwest Venture Partners, which will expand its ability to test and monitor voice agents across more regulated, safety critical environments.
As voice AI becomes the front door to healthcare, every interaction has the power to shape patient trust. Coval is ensuring that trust is earned one call at a time by helping organisations deploy voice AI agents that meet the standards patients deserve.
Want to find out what TechRound can do for your business?
