Coval

Coval is a testing and evaluation platform for voice and chat AI agents. It runs simulated callers against an agent, including noisy audio, interruptions and policy edge cases, monitors production calls with automated evaluations, and adds human review, so a team can measure quality before launch and catch drift after each change to a prompt, model or vendor. Simulation, monitoring and human review feed one another, so production findings become new tests. It is built for product, QA, operations and compliance teams that deploy voice agents. Plans include unlimited seats, an API, a command line tool, an MCP server and agent skills, and the vendor lists SOC 2 Type II, HIPAA and GDPR compliance on every tier. The company announced a $28 million Series A. The older coval.dev address redirects to coval.ai. Starter is $100 per month, billed monthly, with 100 simulation minutes, 1,000 monitored calls and 30 day trace retention, after a 7 day trial that requires a credit card. Growth is $500 per month with 1,000 simulation minutes, 10,000 monitored calls, 90 day retention and unlimited metrics per call. Overage is $0.40 per simulation minute and $0.10 per monitoring minute on Starter, and $0.25 and $0.05 on Growth. Enterprise starts at $4,500 per month with custom limits, SSO, private deployment options and data residency.