Version: v1.2Effective Date: September 24, 2026Company: Viking Media Labs Inc., a Canadian corporation operating Doserly
Educational output
An AI report is an informational summary of recorded data, not medical advice, diagnosis, treatment, or a safety clearance. It can invent or miss facts and can be outdated or inappropriate for your situation. A trend or possible association does not prove cause. Have a qualified clinician or pharmacist check relevant records and references before any medication, preparation, or dosing decision. Do not use a report in an emergency. Pet information and any animal-care decision require a veterinarian.
Offline interaction checks use a limited reference set and are separate from AI reports. They may miss products or personal risk factors; no alert does not establish safety. Discuss potential interactions with a pharmacist or healthcare practitioner before starting, stopping, or changing anything. Calculators show arithmetic from inputs, not a prescription; verify the values and result against official instructions and a qualified professional.
What is sent and when
You choose to request a report and tap Allow & Generate after a pre-send disclosure. Doserly sends OpenAI a report-specific set of medication/compound, dose/adherence, symptom, lab, protocol, correlation, and report-window facts, plus Doserly-curated references. The current path substitutes private aliases for names and IDs, relative day tokens for calendar dates, generic labels for custom items, and a random submission token. It excludes notes and other free text; retrieval uses library canonical names only. An outbound guard blocks detected identifiers or leaked source text. These safeguards reduce direct identification, but the person-level health summary remains personal information and may identify someone by context. Doserly restores display details when saving the report to your account.
Wearable summaries are excluded while their server option is disabled. If offered later, Include wearable summaries is off by default for each report and requires a separate choice. The possible transfer is current and prior 28-day personal averages, observed-day counts, coverage, source labels, and screened possible dose/outcome associations with dates and uncertainty. Raw samples, sample timestamps, and health-store record IDs are excluded from that summary request. These summaries remain personal health data. Declining AI processing leaves other app functions available.
OpenAI retention and reporting a problem
The current OpenAI Responses request uses background mode and does not set store: false. Under standard API controls, response application state is retained for at least 30 days by default. Separate abuse-monitoring logs may contain prompts, outputs, or derived metadata and are kept up to 30 days by default, with legal or safety exceptions. OpenAI does not train its general models on API data by default unless its customer opts in; Doserly has not opted in for user reports. A provider-held request cannot be immediately erased by deleting a Doserly report. The actual hosted project controls remain to be verified.
In a saved report, tap "Report a problem with this content" to flag concerning AI output in-app, or email support@doserly.com. Check an exported report before sharing it and protect the copy you choose to save or send.
Rights and limits
Subject to non-excludable consumer rights and remedies, Viking makes no additional voluntary warranty that AI output is error-free or suitable for a medical decision. Nothing here limits statutory quality, privacy, health, safety, or other duties and remedies that applicable law preserves. Contact privacy@doserly.com for a provider-data or deletion request. Mail: Viking Media Labs Inc., 72 Victoria St S, Kitchener, ON N2G 4Y9, Canada.