Home / Directory / AI SaaS tooling / Eval & quality / Coval
Coval
Coval is a San Francisco testing and monitoring platform for voice and chat AI agents that runs simulated calls before launch, scores live calls in production, and routes flagged calls to human reviewers.
Teams shipping phone or web voice agents in support, healthcare, insurance, or financial services that want regression tests on every release, and agencies comparing voice platforms before picking one.
Teams that only run text chatbots and already use a general LLM evaluation tool, and buyers who need deep IVR, touch-tone, and voicemail testing on legacy phone trees.
Verdict
Coval brings the simulation approach used for self-driving cars to voice agents. Founder Brooke Hopkins led evaluation infrastructure at Waymo, and the platform runs thousands of simulated callers with different accents, interruptions, and background noise against an agent before it goes live, then keeps scoring real calls afterward. More than 60 organizations use it, including Zoom and Deepgram, and it raised a $28 million Series A led by Norwest in June 2026. It is a young company, but for teams that want voice agent testing with clear pricing, it is an easy one to recommend.
Score Breakdown
These scores cover how well Coval's simulations cover real caller behavior, how well it monitors live calls, how well it fits into developer workflows, and how clear its pricing is, and then how it does as a company on innovation, communication, pricing, and contract terms.
Who Uses Coval, Where It Fits on Price, and Its Competitors
Coval sells to engineering and quality teams that build voice agents on platforms such as LiveKit and Pipecat or on their own stack, and to voice AI vendors such as Deepgram. Teams usually start by running a library of simulated callers against an agent, wire those tests into their release process, and then send production calls through the same metrics. Coval was founded in 2024 by Brooke Hopkins, went through Y Combinator, and has raised $31 million in total.
On price, Coval is easier to start with than most voice testing vendors. Self-serve plans cost $100 and $500 a month with published overage rates, and enterprise plans start at $4,500 a month. Its closest competitors are Hamming AI, Braintrust, and Confident AI.
Pricing
Coval publishes its plans, usage limits, and overage rates. Prices are as of November 2026.
| Plan | Price | What stands out |
|---|---|---|
| Starter | $100 a month | 100 simulation minutes, 3,000 upload minutes, unlimited seats, and a 7-day free trial. |
| Growth | $500 a month | 1,000 simulation minutes, 30,000 upload minutes, 90-day retention, and audit logs. |
| Enterprise | From $4,500 a month | Custom volumes, SAML SSO, SCIM, data residency, and a named support engineer. |
The Field at a Glance
How Coval compares with other AI eval and quality vendors we reviewed, by Overall Score and relative typical engagement cost.
Coval scores 7.6 in eval and quality, behind Braintrust (7.8), Confident AI (7.7), and HoneyHive (7.7), and ahead of Hamming AI (7.5), Orq.ai (7.4), Arga Labs (7.0), Vellum (7.0), and Openlayer (6.5).
Compared with …
- Coval vs Hamming AI 7.6/7.5
How Well Coval Fits Each Agent Testing Use Case
We rated Coval against the jobs teams bring to an agent testing platform, drawing on its pricing page, its Series A announcement, and independent comparisons.
| Use case | Fit | Notes |
|---|---|---|
| Pre-launch voice agent testing | Strong | Simulated callers cover accents, interruptions, and background noise. |
| Release gating in CI | Strong | A REST API, CLI, webhooks, and an MCP server can start runs from a build. |
| Production call monitoring | Strong | Live calls run through the same metrics, with real-time alerts. |
| Human review of flagged calls | Strong | Review queues let people check what the automated metrics flag. |
| Comparing voice vendors | Strong | Teams can run the same scenarios against several voice platforms. |
| Legacy IVR and touch-tone menus | Mixed | Hamming AI covers touch-tone and voicemail behavior in more depth. |
Review Excerpts
These excerpts draw on Coval's pricing page, its Series A coverage, and independent comparisons of voice agent testing tools. Paid reviews and pay-for-play sites were excluded.
“Every prompt change now runs through a few hundred simulated callers before it ships, and the release waits until the scores come back.”
“The published overage rates mattered to us. We could estimate a month of testing before we talked to anyone in sales.”
“Our bank's agent still sits behind an old phone menu, and we wanted more touch-tone and voicemail cases than the scenario library gave us out of the box.”
Where customer reviews couldn't be found, our analysts attempted to give a voice to buyer sentiment, drawing on real information from review sites, forums, and product documentation.
Company Facts
- Founder
- Brooke Hopkins (CEO)
- Founded
- 2024
- Headquarters
- San Francisco, California
- Funding
- $31 million in total, including a $28 million Series A led by Norwest in June 2026
- Customers
- More than 60 organizations, including Zoom and Deepgram
- Website
- coval.ai
Key Leadership
Brooke Hopkins
Founder and CEO
She led the evaluation job infrastructure team at Waymo before starting Coval in 2024.
Methodology
We scored Coval on eight criteria, from simulation and test coverage to contract fairness, and the Overall Score is their simple average on a 0 to 10 scale. The scores draw on reviews from public review sites and forums, with paid reviews and pay-for-play sites excluded, plus a hands-on look at the product, its pricing, and its documentation and training. Scores can change as we add reviews and product checks.
Sources
Facts on this page
- She led the evaluation job infrastructure team at Waymo before starting Coval in 2024: NYU Abu Dhabi
Company facts
- Founder: Coval
- Founded: Coval
- Headquarters: The Next Web
- Funding: Coval
- Customers: The Next Web
Review excerpts
- Excerpts based on: Coval pricing, The Next Web on Coval's Series A, and FAUN's Coval and Hamming comparison