Platform, configuration and devices
AI, ambient documentation and automation
Draft-only AI.
How to read status labels: In the pilot (v0.2.0, synthetic data) Simulated in the pilot Coming (Wave 2, rolling out) Designed, not yet built Open decision Not offered / no claim made Planned partner integration
C0870Is AI switched on by default, and who decides to enable it?
No. The specification describes it, but the pilot does not include it. For reference, REQ-2401 (a top-priority requirement, the medium and large tiers) says AI features must be off by default, enabled per facility with model inventory entry and validation evidence. Check: feature toggle fails without inventory entry.
C0871Could an AI feature file or sign anything in a chart on its own?
AI outputs must be labeled, cited to source elements, and never auto-filed or auto-signed. That is REQ-2402, a top-priority requirement for the medium and large tiers. Not yet. It is designed in the specification and not built in the pilot. Test in the specification: filing requires human accept.
C0872Where would AI inference run, and could patient data be used to train models?
Designed, not built: there is no code for this in the pilot. REQ-2403 says AI inference must run on-site or in a customer-approved region; PHI must not be used to train models without separate written approval. Its acceptance check: egress check blocks unapproved model endpoint.
C0873Does AuroraMed log AI inputs/outputs by reference hash and model version?
REQ-2404 is a high-priority requirement for the medium and large tiers: the system must log AI inputs/outputs by reference hash and model version. This is on the design side of the line. Nothing in v0.2.0 does it. To verify it, the specification says inference log entry exists for each use.