Adverse-event capture
Potential adverse-event mentions are flagged as they occur and routed into a pharmacovigilance review queue, with per-tenant configuration for where those flags go and who receives them.
Safety & governance
Five rules that run on every response.
Guardrails are validation stages, not prompt instructions. A response that breaches one is not softened or re-tried into acceptability — it is blocked before it reaches a patient.
The guardrails
Enforced after generation, not requested before it.
A prompt asking a model to behave is a preference. These run on the generated text itself, which is why they hold when a question is phrased in a way nobody anticipated.
- G1
No directive language
The system does not tell a patient what to do. It explains what sources say and routes the decision back to a clinician.
- G2
No diagnosis or assessment
Responses do not diagnose, stage, or assess an individual’s condition, however the question is phrased.
- G3
No unattributed superlatives
Comparative and superlative claims about a therapy must carry an attributed source, or they do not ship.
- G4
No personal health information
Outbound text is scanned for personal health information before it is shown, in addition to redaction on the way in.
- G5
Mandatory citations
A substantive medical claim without a source reference is treated as a validation failure, not a stylistic one.
Review workflows
What happens around the answer.
Human escalation
Questions the system should not answer become tasks in a human-response queue rather than approximate answers. The handoff is recorded as an event, so the path from question to human response is reconstructable.
Medical, legal and regulatory review
Content changes move through pull-request review, which produces a reviewer, a timestamp and a diff for every published claim — the evidence trail an MLR process needs.
Auditability
Operational logs cover system activity, usage and deployment behaviour, including guardrail outcomes, so a review can ask what the system did and get an answer.
Questions we get asked
Straight answers.
- Does OpenSupport give medical advice?
- No. It provides health information with citations and directs clinical decisions back to a healthcare professional. Validator rules block directive language and any diagnosis or assessment of an individual.
- How are adverse events handled?
- Potential adverse-event mentions are flagged and routed into a pharmacovigilance review workflow, configurable per tenant. The platform supports capture and routing; it does not replace your pharmacovigilance obligations or your reporting timelines.
- Does the platform store personal health information?
- It is built not to. Inbound text is redacted for personal identifiers before retrieval or generation, and outbound text is scanned again before it is shown. The patient experience does not require a stored medical history to answer a general information question.
- Can a sponsor influence which sources an answer uses?
- No. Retrieval and ranking logic is identical for every tenant. A deployment can be scoped to a defined portfolio or to its own private content, and that scope is enforced during retrieval — but within scope, nothing can be paid into a higher position.
- What happens when the system does not know?
- It says so, and where appropriate the question is escalated to a human-response queue. An answer that cannot be grounded in retrieved sources fails validation and is not shown.
Bring your hardest compliance question.
The fastest way to evaluate the boundaries is to try to cross them. We will give you a deployment to do exactly that.