Voice-agent QA vendor Cekura calls HIPAA, SOC 2 and GDPR compliance 'defaults,' offers adversarial red-teaming

Cekura, a vendor that tests voice and chat AI agents before they reach production, describes itself on its site — still hosted at vocera.ai — as "the reliability layer for voice agents."

The company says compliance is built into its own infrastructure rather than optional: "The security and compliance infrastructure voice AI demands: HIPAA, SOC 2, and GDPR, not as checkboxes but as defaults," Cekura states. The site names no independent auditor, certification date or audit scope behind that claim — a self-description, not a published, certified compliance schedule.

Cekura's core product runs synthetic conversations against a customer's agent before it goes live: "Run thousands of synthetic conversations before go-live," the company says. That testing is Cekura's own service sold to its customers, not an independent third party assessing Cekura itself.

The company also offers adversarial testing aimed at surfacing failure modes before a customer's agent ships: "Adversarial red-teaming Probe for jailbreaks, leaks, and off-script behavior," per the site.

One named use case is healthcare. "We test every release against thousands of synthetic patients before it ever reaches a real one," a customer testimonial from Twin Health on Cekura's site states, describing use in a patient-onboarding agent.

Another customer, identified as Kastle, says in a testimonial on the same page: "Cekura has become a critical part of our development pipeline. Now we don't ship any agents to production without first aggressively testing them on Cekura."

Cekura offers self-serve signup rather than a closed pilot: the site presents both "Book a Demo" and "Start Free Trial" as immediate options, with no waitlist language on the page checked.