Salus: what customers signal, and what to settle

Public-opinion evidence for Salus, a product that would collect medications, conditions, vaccinations, appointments, and bills for a person and their family, then suggest what to do. These are not this team's customers. They are the prior to test.

The jobs are familiar. The holder and the advice are not. People already use provider and insurer apps for records, appointments, and prescriptions, and many want that information to reach their clinician. Concern jumps when a private technology company holds the data, and trust falls further when AI reads the record to give advice. The family layer adds people who never agreed to be in the file.

3 in 4

U.S. adults used a health app or site in the past year

75%

Concerned if a private tech company holds their health data

32%

Would trust AI advice drawn from their medical record

3 in 10

Multi-app users who find juggling those apps difficult

Source: KFF Health Tracking Poll, Sept 23–29, 2025, 1,334 U.S. adults. Privacy figure is very or somewhat concerned. AI figure is a great deal or a fair amount of trust. Multi-app difficulty is among people who use more than one health app, website, or account.

What people already do, and who they worry about

Use is broad. About 65% of adults say these tools have made managing care easier, and about 5% say they have made it harder. The opening is specific tasks, especially records, visits, and medications. Consolidation is secondary: half of people who have used health apps use more than one, and most of them say they cope.

Health-app tasks in the past year

Percent of all adults. Asked of people who have ever used a health care app or website, reported on the total adult base. Source: KFF Health Tracking Poll, Sept 23–29, 2025.

Privacy concern by who runs the app

Percent of U.S. adults who would be very or somewhat concerned. A hospital-run app draws the least concern. Salus would sit in the private-technology row. Source: KFF Health Tracking Poll, Sept 23–29, 2025.

Two different AI questions

KFF asked about AI on a health care app. 41% would trust an AI chatbot to make appointments or message their doctor. 32% would trust AI that accesses the medical record to provide personalized information or advice. 8% would trust either use a great deal. Trust stays under half in every age group.

EBRI and Greenwald's 2024 survey of 2,011 privately insured adults ages 21–64 found about three in five comfortable with AI inside smart health technology such as wearables and trackers, and about two-thirds wishing that data were shared more effectively with their doctor. 57% wished it were shared more effectively with their insurer. Comfort with a step tracker is a different claim from advice on a medical record. Use the KFF number for Salus.

Sources: KFF, Sept 23–29, 2025. EBRI Fast Facts #538, July 24, 2025, from the 2024 Consumer Engagement in Health Care Survey.

Risks to settle

Ordered by when they bind. The first group can make the service unsafe or unlawful even if customers like a demo. This is a product brief for counsel, not a legal opinion.

When	Issue	Why it matters	What it implies

Regulatory anchors to hand counsel: HHS guidance on the access right, health apps, and APIs; the FTC mobile health app tool and Health Breach Notification Rule; state consumer-health laws such as Washington's My Health My Data Act; COPPA for children under 13; state minor-consent statutes; FDA clinical-decision-support and device questions if the software diagnoses or directs treatment.

Family sharing is the sharp edge The useful version is a parent tracking a young child's vaccines, or an adult child helping a parent who has agreed. The harmful version is one person seeing a teenager's confidential care, a former partner's appointments, or an aging parent's diagnoses after consent no longer holds. Build the second case out of the data model before inviting the first family.

What to ask your own customers

Talk to people who would be harmed by a wrong assumption, not only people who already track their health. In each interview, put a specific record on the table. A medication list, a teen therapy note, a bill with a diagnosis. Ask what they would do next.

Who to recruit	What to learn	Decision it gates

Patient-community color, not a national sample: The Light Collective, Tangled in the Web (2024), surveyed roughly 300 to 377 people. 74% were moderately to extremely concerned about health information shared without consent. 63% feared hackers, 57% sharing without consent, and 33% an insurer learning their conditions. On AI in health care, 79% were concerned their information could be used in a way that harms them, 76% cited misinformation, 69% incorrect diagnoses, and 52% bias. Use these as prompts, then measure them in your own sample.

Scope that matches the evidence

Start with one adult's medications and appointments, connected through patient-directed access, with the source and date on every item. Family access stays off until each adult accepts a defined proxy, and it stays closed for categories the law treats as confidential for minors. Suggestions are reminders and questions for the next visit.

Generative advice waits until three things exist: a written intended use, a rule for what the product refuses to say when the chart is incomplete, and a counsel review of device status, privacy, and liability. Bills wait until the health record itself is trusted. Advertising, sale, insurer access, employer access, and training on identifiable records stay out of the business model.