Voice AI testing for healthcare: Coval, Hamming, Cekura, Bluejay
Four funded voice-agent testing vendors, what each does, what they charge where public, and why healthcare call failures still lack independent benchmarks.
Who tests voice agents today?
| Vendor | Notes | Funding or customers | Source quality |
|---|---|---|---|
| Coval | Simulation and monitoring; plans from $100 a month, enterprise from about $4,500 | $28M Series A, 24 June 2026 (Norwest); about $31M total | Press and vendor pricing page |
| Hamming | Voice agent QA; launched on Hacker News on 15 August 2024 | Seed round announced December 2024; case studies name NextDimensionAI, Basata and Bland Labs | Vendor blog |
| Cekura | Testing and monitoring for voice and chat agents | Reported 70+ customers in July 2025; names Confido Health | Vendor and case-study pages |
| Bluejay | Voice and chat agent testing | Seed-stage | Not researched in detail |
What do their healthcare results look like?
Cekura’s Medicare workflow experiment reports agents passing 91.3% (ElevenLabs), 82.6% (Pipecat), 73.9% (LiveKit), 69.6% (Synthflow) and 65.2% (Vapi) of repeated runs. It is one scenario run by a testing vendor, so it is a demonstration, not a benchmark. It does count a workflow as passing only if every repeat passes, which is a good rule.
What can’t they do yet?
Three gaps recur. Attribution: nobody can say whether a failed call came from speech recognition, the model, text-to-speech or telephony. Realism: no public healthcare call recordings were found, and accent, noise and clinical-vocabulary realism are untested. Independence: published pass rates are run by vendors.
Is there anyone outside the US?
SuperBryn, a Kerala-founded start-up, raised a $1.2M pre-seed in December 2025 and names healthcare as a target; its design partners are in the US and the Middle East, and no customers are named. Bolna, an Indian voice-agent platform, raised a $6.3M seed in January 2026 and builds agents rather than testing them.
What did we not verify?
Pricing for Hamming, Cekura and Bluejay; head-to-head claims, which come from vendors’ own blogs; and any healthcare-specific failure rate from an independent source.
Frequently asked questions
Are there healthcare-specific voice agent testing tools?
We did not find a dedicated, funded healthcare voice-agent testing vendor. Coval, Hamming, Cekura and Bluejay are horizontal tools used by healthcare voice companies, and SuperBryn, an Indian start-up, names healthcare as a target.
How much does voice agent testing cost?
Coval's public pricing runs from $100 a month to enterprise plans starting around $4,500 a month, billed by simulation and upload minutes. Prices for the other vendors were not retrieved, and healthcare-specific pricing was not found.
How reliable are published voice agent pass rates?
Treat them as marketing. Cekura's Medicare workflow test reports agents passing between 65.2% and 91.3% of repeated runs, but it is a single scenario run by the vendor that sells testing.