The trigger — end of session
Heidi session complete
AI transcript and structured summary ready · M. Harrison · Lower back · 48 min
Patient profile
Patient-owned · Portable · Consent-native
Full AI session summary added to patient's portable health record
Patient can present this at any future clinic — no login, no transfer request
Patient controls consent — they choose what the next clinic sees
Aligns with MyHealthRecord (AU) and global patient-held record direction
Raw notes never leave without explicit patient consent
Kinetic network
Abstracted · Anonymous · Structured
Heidi AI extracts only structured fields — no raw text enters the network
Fields shared: session count, protocol category, ICD code, outcome score
Contribution score increments — unlocking access tiers and improving search rank
Clinic identity is never attached to the contributed record
Clinician opinions, notes & identifiers never transmitted
Core design principle
Both fears are addressed by what the system structurally cannot do — not what it promises. The patient profile model eliminates switching fear because no clinic identity is ever surfaced. The abstraction layer eliminates judgment fear because no clinical opinion ever enters the network. Clinics don't need to trust Heidi. The architecture makes the risk impossible.
What stays protected at every step
No clinic discovery
The receiving clinic never learns who previously treated the patient. History is patient-owned, not clinic-linked. Switching fear: eliminated.
No raw notes shared
Clinician opinions, subjective assessments and session notes never leave the originating clinic. Only structured outcome fields are extracted. Judgment fear: eliminated.
Patient controls consent
The patient decides what the next clinic sees when they present their profile. Consent is active and revocable — not assumed or buried in a policy.