Patient care navigation and health copilot AI
VisitRecall
VisitRecall is a patient and family visit-recall app that emphasizes on-device transcription, no stored audio, encrypted transcript/summary storage, patient-controlled sharing, and no model training under its Anthropic contract. It is agency-aligned in concept, but the breadth of longitudinal records, family sharing, lab/bill explanation, and values-based guidance makes correction and consent controls especially important.
Public-source research has been drafted; final human publication review and change-log detail are still required.
Summary judgment · 83% toward patient-directed
Potentially agency-expanding
A personal visit-recall tool may expand patient and caregiver agency by preserving what happened in the room, but broad longitudinal and billing/lab features require careful privacy, evidence, and boundary review.
Patient agency
How this tool changes agency
Plain-English summaries, next steps, family sharing, visit prep, cross-doctor context, and values-based question support are directly action-oriented.
Deletion/export rights are public, but transcript, summary, profile, lab, bill, and privacy-classification correction workflows need app-level verification.
Patient-facing signals
Who does this AI serve?
Public materials explicitly say the tool is for patients and families, not doctors, insurers, or employers.
Can patients tell AI is involved?
AI summaries, chat, privacy classification, question generation, and longitudinal insights are visible product features.
Can patients meaningfully choose?
Use is voluntary; privacy materials describe sharing controls, export/deletion rights, and explicit consent before sending transcripts or documents to Anthropic.
Can patients correct or challenge what the AI produces?
Deletion/export rights are public, but transcript, summary, profile, lab, bill, and privacy-classification correction workflows need app-level verification.
Does it help patients understand or act?
Plain-English summaries, next steps, family sharing, visit prep, cross-doctor context, and values-based question support are directly action-oriented.
Text findings
Who is left out or burdened?
Evidence incomplete
The tool targets family/caregiver burden, but public evidence on language, disability access, device requirements, cost, and family power dynamics is limited.
What happens to patient data?
Meaningful public detail
Privacy materials describe on-device transcription, no stored audio, encrypted Firebase/Google Cloud storage, Anthropic transcript processing with consent, deletion/export rights, and no model training under contract.
Are the clinical boundaries clear?
Clear in wording
Terms emphasize no medical advice and user responsibility for recording consent; high-stakes lab, bill, and value guidance still require careful presentation.
Who defined what good looks like?
Mostly vendor-defined
No independent clinical, equity, or patient-outcome evaluation was found in this pass.
Review method
Deep public-source review of official product page, privacy policy, terms, app listings, and credible news context; no vendor interview, app walkthrough, or independent model evaluation.
Draft profile · Medium draft