Practitioner-led AI governance
Karunyx builds AI evaluation and governance tools for healthcare and biotech — designed by the people who've run the Architecture Review Boards and written the SOPs, not around them.
Regulated industries are adopting AI under pressure — from clinical decision support to drug development to payer automation — but the tooling hasn't caught up. Audit logs get bolted on. Risk classification is treated as a footnote. PHI redaction is left to the buyer to sort out.
The result: organizations deploying AI in high-stakes environments carry risk that neither their vendors nor their own teams can fully see, measure, or defend to a regulator.
Each enters a different regulated vertical with its own practitioner credibility. Both are built on the same conviction about how governance should work.
Biotech & pharma
GxP risk-based classification, built around the SOP your team already follows — not a rewrite of it. Encodes governing rules like the Severity Override Rule directly, and outputs structured data straight into your validation workflow. (Working name.)
Stage: working single-user proof-of-concept
Read more →Healthcare payers & providers
Kagetora, our AI security testing platform, brings tiered evaluation and prompt-injection detection to AI systems operating near PHI — deterministic and semantic judges, audit-grade logging, built for the payer/provider security review process.
Stage: under active development
Read more →Neither product started from a market map. Each started from a problem its founder hit personally, inside a regulated organization, with an auditor or a security review on the other side of it.
Biotech & pharma
Matthew Teli brings 30+ years in biotech and pharma quality and validation leadership, with inspection experience spanning FDA, MHRA, TGA, Health Canada, ANVISA and PIC/S.
Healthcare
Samuel Beera brings 25+ years in enterprise architecture, including building production AI infrastructure and zero-trust AI security architecture inside a large healthcare payer organization.
Where this is going
Both products are designed to feed a shared Sovereign Learning Loop — governance infrastructure regulated organizations can eventually run on their own infrastructure, without sending sensitive data to a vendor.
This is the direction we're building toward, not a shipped product today.
If you're evaluating AI inside a regulated environment and the governance question is landing on your desk, we'd like to hear how you're handling it.