Security & Trust
How we protect the data our clients trust us with
Plain answers for security reviewers: what we do today, what is still in progress, and how to reach us. We mark gaps honestly rather than overstate.
Last updated: 2026-11-16 (DRAFT)
Your data
How we handle client data
- Client data is used only to provide the service. It is never sold, and our contracts exclude it from AI model training. An internal export tool that builds training data from agent activity must exclude client data before any client is onboarded.
- We do not accept patient health information (PHI) unless a Business Associate Agreement is signed and our own vendors are covered.
- Selected Guardian/Pantheon data-export paths invoke content filtering; coverage must be checked for each integration. This is not a guarantee that every sync blocks every identifier. Website analytics and error reporting use the separate controls described below.
Controls
Security controls
- Encryption: TLS 1.2+ in transit. Data held in our hosted database is encrypted at rest by the provider; some products currently keep data in local files that are not encrypted at rest. Ask for the product-specific summary.
- Secrets: platform secrets are kept in encrypted vaults, not in source code; a few known exceptions are being retired.
- Access: controls differ by product. GoodCreds sign-in requires multi-factor authentication; some products have no sign-in yet and run only locally. Administrative-access controls are being documented. Ask for the product-specific summary.
- Audit trail: our core platform keeps an append-only, hash-chained audit log; its latest verification found forks in the chain, and a fix is pending. Not every product writes to it yet, and coverage differs by product; ask for the product-specific summary.
- Risk analysis: HIPAA risk-analysis interview completed November 2026; the analysis remains a draft pending attorney review and owner approval.
In progress
What is not done yet
- SOC 2: monthly evidence collection started November 2026; no audit report yet. We answer security questionnaires in the meantime.
- Single sign-on (SAML/OIDC) for client accounts: planned.
- Backup restore drills and backup-encryption verification: planned before any uptime commitment.
Incidents
Incident response
We follow a written incident-response plan. Clients are notified of a confirmed personal-data breach within 72 hours, and of a breach of PHI under a BAA within 5 business days.
Subprocessors
Subprocessors
This is a non-exhaustive summary covering our services, website and store; the subprocessors for a client engagement are listed in the DPA. The Privacy policy describes website third-party processing, including optional services, scheduling and fonts.
- Supabase — database and storage (U.S.)
- Anthropic — AI model inference (U.S.)
- Hostinger — application hosting
- Cloudflare — edge network
- Vercel — web hosting
- Stripe — payments
- Resend — contact-form notification emails, if enabled; submissions may be forwarded for delivery
- Google — optional product-buyer sign-in at api.jtmdai.com
- PostHog — optional website visit analytics, loaded only after Accept and subject to Do Not Track and Global Privacy Control
- Sentry — website error reports, separate from the consent banner and stopped by Do Not Track or Global Privacy Control; see the Privacy policy for collection scope
We give 30 days’ notice before adding or replacing a subprocessor that handles client data.
Contact
Report a vulnerability or request documents
Email [email protected] with “Security” in the subject. You can request our security summary and questionnaire answers. DPA and BAA templates may be shared only after a licensed healthcare attorney has reviewed and approved them.