Coval

HQ: United States
Founded Year: 2024
Company Type: Privately Held
Category: QA
This is CXSpy's primary classification of the Coval software. The software may fit other categories as well.
Type: tool
CXSpy classifies CX software in one of two types (a platform or a tool). This is our view of the Coval software.

Description:Coval is a San Francisco startup that tests and monitors voice and chat AI agents before and after they reach customers. Brooke Hopkins founded it in 2024 after leading evaluation infrastructure at Waymo, where self-driving software was tested in large-scale simulation, and the company went through Y Combinator in the summer 2024 batch. Coval applies a similar idea to conversational AI. It simulates thousands of customer conversations with different personas, accents, and background noise, scores how the agent performs against defined metrics, and flags regressions when a prompt or model changes. In production it monitors live calls and chats so teams can catch failures and trace them back to their cause. In June 2026 Coval raised a $28 million Series A led by Norwest, with Base10, Twilio Ventures, Y Combinator, MaC Venture Capital, and Swift Ventures, bringing total funding to $31 million. It said it had more than 60 customers, including Zoom and Deepgram. Self-serve plans start at $100 per month.

Coval Call Center Quality Assurance ("QA") Software

About Coval

Coval was started in 2024 by Brooke Hopkins, who previously led the team building evaluation infrastructure at Waymo. Self-driving cars are tested against huge numbers of simulated scenarios before they reach the road, and Hopkins built Coval on the view that AI agents talking to customers need the same kind of testing. The company joined Y Combinator in the summer 2024 batch.

The product covers two stages. Before launch, Coval generates simulated conversations against a voice or chat agent, using test sets and personas that vary intent, accent, interruptions, and background noise. Each run is scored with built-in and custom metrics such as task completion, latency, and policy adherence, so teams can compare versions and catch regressions when they change a prompt, model, or tool. After launch, Coval monitors production conversations, scores them, and alerts teams when quality drops.

This sits close to classic contact center quality assurance, but the subject is the AI agent rather than a human one. Coval's customers include companies building AI agents for their own support lines and platform vendors that sell voice AI to others.

In June 2026 Coval announced a $28 million Series A led by Norwest, with participation from Base10, Twilio Ventures, Y Combinator, MaC Venture Capital, and Swift Ventures, for $31 million raised in total. The company said it had more than 60 customers at the time, naming Zoom and Deepgram. Its site also shows Chime, Perplexity, ServiceNow, StubHub, Toast, and Upstart.

Coval Features

1. Simulated conversations that test voice and chat AI agents at scale

2. Personas that vary intent, accent, interruptions, and background noise

3. Built-in and custom metrics for task completion, latency, and policy adherence

4. Regression testing when prompts, models, or tools change

5. Production monitoring and scoring of live conversations

6. Alerts and tracing to find the cause of agent failures

7. Works with voice agent platforms and in-house stacks built on LiveKit or Pipecat, via API, CLI, and MCP

8. Dashboards to compare agent versions and track quality over time

Coval Pricing

Coval publishes plan pricing. Starter costs $100 per month and includes 100 simulation minutes and 1,000 monitored calls a month with 30 days of trace retention. Growth costs $500 per month and includes 1,000 simulation minutes and 10,000 monitored calls a month with 90 days of retention. Both come with unlimited seats and a seven day free trial. Enterprise starts at $4,500 per month with custom volumes, SAML single sign on, private cloud deployment, data residency, and a named support engineer.