AI-enabled virtual primary care
K Health
K Health’s current enterprise offering includes dynamic intake, PatientGPT record questions, and clinician care. Patient questioning and institutional routing are distinct functions. The old direct consumer service is historical, and partner-specific privacy and refusal controls remain unverified.
Checklist assessment dated October 30, 2026. Independent confirmation is incomplete. Hugo has not personally verified this service.
Agency checklist · October 30, 2026
Constrains 1 of 12 yes
Document-based draft. Independent confirmation is incomplete. Hugo has not personally verified this service.
Assessed function: K Health enterprise AI for health systems: dynamic intake, PatientGPT medical-record conversation (for example inside Hartford HealthCare's portal) and virtual care. The discontinued direct consumer service is historical context.
A no can mean missing documentation or institution-dependent control; it does not by itself establish harmful behavior. These categories are not clinical safety or product-quality ratings.
- Public documents do not establish actual product behavior.
- The independent review screened rules and selected evidence; it did not replicate all twelve answers for this tool.
See all twelve answers and their evidence
| Question | Answer | Evidence / review | Source and reasoning |
|---|---|---|---|
| 1. Told it is AI | No | not-documented Not independently replicated | Certain of our Services may be provided using our artificial intelligence models and tools, such as our artificial intelligence-enabled Symptom Checker.Source K Health's privacy policy discloses AI use generally. No in-product notice or consent step for health-system intake or PatientGPT is documented in K Health's own pages. Press describes PatientGPT as 'an AI tool'. |
| 2. Operator and model maker named | No | not-documented Not independently replicated | These may include, for example, hosting and server services, customer support, chatbot, analytics providers, and payment processors.Source K Health, Inc. is named. The policy lists an unnamed 'chatbot' service provider. The maker of the language model behind PatientGPT is not named in the sources read. |
| 3. Patient chooses to use it | No | deployer-dependent Not independently replicated | Our dynamic intake and PatientGPT integrate directly into EHR systems, acting as a secure digital front door for both virtual and in-clinic care.Source Health systems deploy it and intake is part of the front door. For PatientGPT alone, the AHA says it is 'a safe option for patients who choose to consult AI', so Q3 and Q5 would likely be yes for PatientGPT by itself. |
| 4. Patient sets the task | No | documented Not independently replicated | ones that draw directly from their existing medical records and can funnel patients toward care in their own systemSource STAT describes health-system chatbots as a way to funnel patients to the system's own care. The patient asks the questions but the operator steers toward its own service line. |
| 5. Patient can decline or stop | No | deployer-dependent Not independently replicated | HHC’s AI solution offers a safe option for patients who choose to consult AI about medical matters.Source PatientGPT appears optional in Hartford HealthCare's portal (AHA). Enterprise intake is not shown to be optional, and the service is institution-deployed. No decline right documented by K Health. |
| 6. Patient sees the AI output | No | not-documented Not independently replicated | Quotation not recorded; evidence repair pending. Source PatientGPT answers are seen by the patient (HIT Consultant: it can generate a summary of their concerns), but the sources do not show that the patient sees the AI output of the intake or virtual-care components. Answer is for the whole function. |
| 7. Patient can correct or challenge | No | not-documented Not independently replicated | Quotation not recorded; evidence repair pending. Source The Notice of Privacy Practices lets patients ask to correct their medical record, but no route to contest AI output is documented. |
| 8. Patient can get a copy | Yes | documented Not independently replicated | You can ask to see or get an electronic or paper copy of your medical record and other health information we have about you.Source K Health's Joint Notice of Privacy Practices (updated 2026-02-16). Applies to health information K Health holds as a provider. Whether PatientGPT chat logs are included is not stated. |
| 9. Patient can delete | No | not-documented Corrected after independent review; accepted by editor | For residents covered under the State Privacy LawsSource The cited deletion right is in the geographically limited state-law supplement. |
| 10. Training or sale does not override patient choice | No | documented Not independently replicated | Specifically, we may use your personal information in order to improve our artificial intelligence models and toolsSource Also: de-identified data may be used for 'any purpose not prohibited by law, including training our artificial intelligence models'. No patient opt-out for that. HIT Consultant reports Hartford HealthCare says patient data will not train external models, which is a deployer assurance, not K Health's policy. |
| 11. Patient can run it or take it elsewhere | No | not-documented Not independently replicated | Quotation not recorded; evidence repair pending. Source Vendor-hosted, deployed by health systems. No patient route to run it or move the function. |
| 12. No advertising or data business | No | documented Not independently replicated | We may "sell" your sensitive personal data.Source K Health's privacy policy documents advertising and sale or sharing of personal data with ad partners, plus sharing of de-identified data for any purpose. That policy was written for the consumer service. Enterprise deployments may be governed by health-system and BAA terms that were not public. |
Initial coding: Claude Sonnet helpers, compiled by Claude Opus 5.5 (Anthropic). Selective review and corrections: GPT-6 Astra (OpenAI). Editor: Hugo Campos. Acceptance of corrections is not product verification.
Checklist rules · Independent review and limitations · Later evidence follow-up
Earlier v1.3 assessment and control finding
Historical assessment under different rules. It does not set the current checklist zone.
Agency finding, before method v1.3
Mixed, institution-led
Legacy editorial placement, pre-method: 46
This profile has not yet been reviewed under the published method. Its finding and any number are editorial placements made before the method existed.
- Evidence
- Documentary review. Public sources only, no hands-on testing by HugoScore.
- Review status
- AI draft. No person has settled the findings.
Earlier evidence and assessment
The sections below preserve the earlier review. Its classifications and provenance are historical; the checklist above gives the current draft classification.
Earlier finding · Legacy editorial placement, pre-method: 46
Method v1.3. The control line is evidence for how the agency finding arose, not a quality score. Legacy editorial placement, pre-method: 46. Not comparable with v1.3 positions.
Mixed, institution-led
K Health’s current enterprise offering includes dynamic intake, PatientGPT record questions, and clinician care. Patient questioning and institutional routing are distinct functions. The old direct consumer service is historical, and partner-specific privacy and refusal controls remain unverified.
Patient agency
How this tool changes agency
Record conversation may help patients understand history and prepare questions. The marketing does not establish source citations, correction of extracted context or alternatives to AI intake.
Partner-specific refusal, records correction, provenance and data-use controls remain unverified
Patient agency assessment
Who sets and changes the goal?
PatientGPT offers direct questioning of records, while intake and care routing remain institutionally configured. These should not be collapsed into a single clinician-only function.
What can the patient understand, question, or do?
Record conversation may help patients understand history and prepare questions. The marketing does not establish source citations, correction of extracted context or alternatives to AI intake.
Can the patient evaluate the conditions of use?
Enterprise branding confirms the current deployment model. It does not verify notice within each partner app, sponsor reporting or the consequences of refusal.
Text findings
Conditions of use
Published controls and their limits
Enterprise branding confirms the current deployment model. It does not verify notice within each partner app, sponsor reporting or the consequences of refusal.
What remains unknown?
Not tested or not established
Partner-specific refusal, records correction, provenance and data-use controls remain unverified
Who evaluated this?
AI-assisted public-source draft
Vendor statements describe published conditions, not independently verified behavior. No clinical, security, accessibility, or legal validation is claimed. Earlier evidence remains dated in the report and history.
Sources checked
Source-specific findings and retrieval limitations are recorded in the full report.
Earlier review provenance
Criteria
CAIHL-derived HugoScore framework and September 7 qualitative review priorities. Draft v1.2 numerical anchors remain unadopted.
Reviewer
AI-assisted public-source reassessment prepared in OpenAI Codex.
AI / model
OpenAI Codex / GPT-6
Human review
Hugo Campos authorized publication of these AI-assisted draft reassessments on September 8, 2026. This does not claim comprehensive human verification of every finding.
Review date
2026-09-08
Limitations
Partner-specific refusal, records correction, provenance and data-use controls remain unverified No live product use, patient-data upload, account creation, code audit, clinical evaluation, or independent implementation validation.
Earlier review method
Focused public-source reassessment using CAIHL: patient authority, critical capacity, and informed control. Existing evidence plus one focused primary-source pass and at most one targeted follow-up. No live product testing. Numerical scores remain provisional editorial placements, not a new calculation.
Document-based draft · Low to moderate, product claims with blocked governance follow-up (AI-assisted draft)