The SUPERWISE® AMP platform applies configured PHI guardrails at runtime, enforces policy and safety constraints, and keeps an audit trail for every AI-assisted decision. Sentinel separately replaces its documented identifier types on the PC before supported prompts leave. The platform (AMP) is not covered by a BAA. Keep PHI out of it.
Govern AI across patient-facing, operational, and clinical workflows. Sentinel replaces the identifiers it recognizes before a prompt leaves the PC; names and most dates still pass through, so patient work belongs on the hosted model under a custom contract with a BAA.
Patient work goes to a model SUPERWISE hosts on its own equipment, under a BAA with your organization. For the outside tools staff keep using, Sentinel replaces the identifiers it recognizes before a prompt leaves the PC.
A hosted model for patient work
- Hosted by SUPERWISE in a secure cage at 365 Data Centers, Nashville, Tennessee.
- The model is gpt-oss-120b. Subject to change as models advance.
- Any covered entity can sign, including practices, hospitals, and health systems.
- Covered: Chat on the hosted model under a custom contract; Sentinel routing to the hosted model.
- Not covered: Self-serve Chat plans; The platform (AMP); Outside models a customer chooses; Web search.
- If we confirm a security incident that affects your data, we notify you within 72 hours.
- The BAA carries the same 72-hour notice.
- Staff reach it through SUPERWISE Chat, or through Sentinel configured to route a tool to it.
It is available now under a custom contract.
Self-serve SUPERWISE Chat plans don't include a BAA. Keep PHI out of them.
Sentinel for outside tools
SUPERWISE® Sentinel inspects supported prompts on the PC and replaces the identifiers it catches before the request leaves. ChatGPT or Claude receives the transformed prompt without those replaced values. Live activity shows which rule fired.
Sentinel replaces email addresses, phone numbers, Social Security numbers in the 123-45-6789 format, credit card numbers, IBANs, IP and MAC addresses, full-length OpenAI keys, and AWS secret keys when labeled. Bank account numbers, tax IDs and EINs, national IDs, VINs, dates of birth, and medical record numbers are replaced only in the labeled formats the gateway recognizes. It does not catch names, street addresses, passport numbers, bank routing numbers, driver's licenses, Okta tokens, unlabeled numbers, short test keys, or values under a column header in a pasted table. Each value it catches becomes {{REDACTED}}. An admin can change the replacement text, or block the request instead. For each request Sentinel records metadata only: counts, which rule fired, and latency. Prompts and responses are never sent to SUPERWISE.
Sentinel replaces the identifiers it recognizes before a prompt leaves the PC. Names and most dates still pass through, so patient work belongs on the hosted model under a custom contract with a BAA. Sentinel does not make any tool HIPAA compliant. It also does not determine whether an organization complies with HIPAA. Read the answers for ChatGPT and Claude, and check the PII redaction documentation before relying on a format.