Perplexity Health released an AI platform that reads wearable data and electronic medical records to build personalized workout and nutrition plans. The service connects with Apple Health, Fitbit, Ultrahuman, and Withings devices and pulls lab results from more than 1.7 million health institutions across the United States. It delivers recommendations based on individual biomarkers rather than generic search results.

Why it matters: Consumers now face AI platforms that promise to translate their body's data into daily action plans. Without peer‑reviewed validation, these tools function as research assistants, not clinical decision makers.

The system reads resting heart rate, step count, sleep duration, and recent bloodwork to generate daily activity and meal suggestions. Each plan lists specific exercise sets, calorie targets, and nutrient ratios. The platform updates weekly as new data arrive.

Current guidelines recommend using validated decision‑support tools. Perplexity's AI has not yet published peer‑reviewed validation studies. Early trials suggest AI can improve alignment of advice with personal physiology, but measurable impact on outcomes like blood pressure or sleep quality remains unproven.

How the data flows: Users connect wearable streams and authorize access to electronic health records via a secure API. The platform extracts lab values such as hemoglobin A1c, lipid panels, and vitamin D levels. It then matches them to peer‑reviewed articles indexed in PubMed. Data are encrypted with AES‑256 and stored on secure cloud servers before any algorithmic processing.

Encryption standards meet HIPAA requirements. The platform excludes raw records from model training. A Medical Advisory Council of board‑certified physicians and data scientists reviews output for safety.

Only subscribers with Pro or Max plans may activate the service. The feature is limited to U.S. residents. Subscription fees are disclosed on the company website and vary by tier. The FDA classifies AI health platforms that deliver treatment recommendations as medical devices, requiring validation of clinical association, analytical performance, and real‑world outcomes.

The bottom line: Consumers should treat the AI output as a supplemental research tool and discuss any changes with a healthcare provider. Without peer‑reviewed evidence, the platform cannot replace professional diagnosis or therapy. Ongoing regulatory review and future clinical trials will determine whether algorithmic plans can reliably improve health metrics compared with standard care.