HugoScore hugoscore.org

Full review

Neatly Health CAIHL draft report

Evidence-linked HugoScore draft report for a health AI tool that affects patients.

Back to profile Markdown source

HugoScore CAIHL Draft Report: Neatly Health

Status: Draft for human review Last reviewed: 2026-06-28 Review method: Deep public-source review refreshed 2026-06-28 using official product, FAQ, about, privacy, terms, Apple App Store, and Google Play materials; no hands-on testing, vendor interview, security audit, or independent model evaluation. Service: Neatly Health Vendor: Neatly Health, Inc. Category: Patient visit recording and health companion AI

Summary

Neatly Health is a patient-facing mobile app for recording medical visits, producing transcripts and plain-language summaries, and asking AI follow-up questions before, during, and after care. Public materials describe a patient-side tool for people who want to remember what was said, prepare better questions, share information with caregivers, understand next steps, and compare their experience with aggregated "patients like you" patterns.

From a CAIHL perspective, Neatly is potentially agency-expanding because it gives patients a usable record of the conversation and helps them process care on their own terms. The main caution is that the app is free because Neatly sells aggregated, de-identified, cohort-based insights to pharmaceutical and medical-device companies. That model is disclosed, but it still creates a standing patient-agency question: can users understand and control secondary use of sensitive health conversations?

Evidence Reviewed

CAIHL Profile

  • Who does this AI serve? Patient-directed, with a commercialization caveat. Neatly is built for patients and caregivers, but aggregated insight customers are also part of the business model.
  • Can patients tell AI is involved? Yes. Recording, transcripts, summaries, chat, suggestions, reminders, and pattern recognition are visible product claims.
  • Can patients meaningfully choose? Mostly yes. Use is voluntary and free, but users carry the burden of recording consent and must understand sensitive data-sharing, cohort-insight, and optional third-party opportunity implications.
  • Can patients correct or challenge what the AI produces? Partial. Public materials describe correction and deletion rights, but app-level editing and source-challenge workflows need verification.
  • Does it help patients understand or act? Yes. Plain-language summaries, next steps, question prep, caregiver sharing, reminders, and health-history chat are action-oriented.

Agency Interpretation

Neatly's clearest agency value is preserving the visit record for the person whose care is being discussed. That can reduce anxiety, improve recall, support caregiver coordination, and help patients ask better questions.

The unresolved issue is secondary data use. Neatly discloses de-identified, aggregated insight sales, cohort-based "patients like you" features, and optional third-party opportunities, and says identifiable data is not shared without permission. For CAIHL, the question is whether patients can see, understand, refuse, or shape those data uses as clearly as they can use the recording and summary features.

Key Unknowns

  • Whether users can edit transcripts, summaries, action items, reminders, and health-history facts.
  • Whether AI answers are source-linked to the original visit transcript.
  • Which AI vendors or subprocessors handle audio, transcripts, photos, records, and derived insights.
  • Whether aggregated insight sharing is opt-out, opt-in, or governed only through broad terms.
  • How users should reconcile app-store privacy labels with policy-level disclosures about de-identified aggregated insight sharing.
  • Whether clinical-trial or patient-support opportunities stay clearly separate from neutral health support.
  • Whether the app supports languages beyond English and meaningful accessibility accommodations.
  • Whether independent usability, accessibility, safety, privacy, or patient-outcome evidence exists.

Publication Recommendation

Ready for human review as a draft profile. Keep the confidence at medium until correction controls, AI processing details, secondary-use controls, accessibility, app-store privacy-label interpretation, and independent evaluation evidence are verified.