People will compare Kin Health and Heidi Health because both involve AI processing consultations and both produce summaries. But they should not be compared as if they are the same product — because they serve different users, produce different outputs, operate under different governance frameworks, and carry different risk profiles.
Kin Health: Patient Memory and Follow-Through
Kin is patient-initiated. The patient records their consultation on their own phone using a free app. The system produces a plain-language summary with action items and next steps. The patient shares it with family and caregivers through a care circle. The output does not enter the medical record. Kin is free, monetised via downstream referrals, labs, and prescriptions. Founded by GoodRx alumni and physician brothers. $9M seed, May 2026. Not HIPAA-certified (patient-facing), claims comparable privacy standards.
The user is the patient. The primary value: remembering what the doctor said and acting on it.
Heidi Health: Clinician Workflow and Documentation
Heidi is clinician-initiated. The clinician activates the ambient scribe during the consultation. The system generates clinical documentation — notes, letters, coding suggestions, referral content — for the clinician to review and save to the patient record. Heidi reports 15 million+ UK sessions, use by one in two UK GPs, 1.8 million appointments per month, and a 61% decrease in primary care out-of-hours documentation time. $65M Series B at $465M valuation. MHRA Class I registered, DTAC assessed, ISO 27001 and SOC 2 Type II certified.
The user is the clinician. The primary value: faster, more complete clinical documentation with more eye contact during the consultation.
Different Users, Different Governance, Different Risks
Kin is downloaded by patients. Heidi is procured by clinicians or healthcare organisations. Kin operates under consumer data privacy frameworks (patient-initiated personal use). Heidi operates under medical device regulation, NHS information governance, clinical safety standards, and DPIA requirements. Kin's output stays on the patient's phone. Heidi's output enters the permanent medical record.
The risk profiles differ accordingly. A wrong Heidi note — if uncorrected by the clinician during review — becomes part of the permanent record, potentially affecting future prescribing, referrals, insurance, and medico-legal review. A wrong Kin summary may cause the patient to misunderstand their care plan, miss a safety-netting point, or share incorrect information with caregivers — without any clinical review step catching the error.
Both risks are real. Neither is trivial. But they require different mitigation strategies: clinician review for Heidi, clear verbal communication for Kin.
Where the Categories May Converge
Over time, patient-side and clinician-side ambient AI may converge: the same consultation captured once, producing both a clinical record (for the clinician) and a patient summary (for the patient) from the same audio source. This convergence would require shared consent frameworks, data-sharing agreements, and quality assurance for both outputs — but it would eliminate the duplication of two separate recording systems operating on the same conversation.
Which Tools Complement iatroX?
Heidi documents the consultation. Kin remembers it for the patient. iatroX grounds the clinical answer behind the consultation — with cited evidence, calculators, exam learning, and CPD. Documentation is one layer. Patient memory is another. Clinical reasoning support is a third. All three serve the same consultation. None replaces the others.