Useful, honest and careful with your health
PeakRoutine sits close to a medical condition, so we hold ourselves to clear rules about what we say, what we show and what we do with your data.
Safety rules the coach can't break
The coach uses AI to write its replies. Around it, PeakRoutine runs fixed rules in ordinary code that don't depend on what the AI writes. They check your question before it reaches the AI, and check the reply before you see it.
- No diagnosis. Questions like "Do I have diabetes?" or "Have I reversed it?" are answered by pointing you to your clinician, who can test and diagnose.
- No dosing. Starting, stopping or changing a medicine or supplement is a question for your pharmacist or doctor. The coach can help you write the question down.
- No promises. If a reply promises an outcome, gives a diagnosis or mentions a dose, it is replaced with a fixed, safe reply.
- Urgent symptoms such as fainting, confusion or chest pain get a clear instruction to seek help now.
- Pregnancy gets redirected to your pregnancy care team, because targets differ.
When you mention a reading at one of these levels, the coach tells you it's worth raising with your clinician:
| Reading | Why it matters |
|---|---|
| Fasting glucose 126 mg/dL or higher | A level clinicians use when testing for diabetes |
| Any glucose 200 mg/dL or higher | A level clinicians use when testing for diabetes |
| Glucose under 70 mg/dL | Low blood sugar |
| HbA1c 6.5% (48 mmol/mol) or higher | In the diabetes range |
These are prompts to see a clinician, never a diagnosis. Home readings and sensors can be wrong, and only your clinician can confirm a result.
How we talk about your data
GMI is not your A1c. The glucose management indicator is an estimate calculated from at least 10 days of sensor readings. HbA1c is a blood test that reflects roughly the past three months. They often differ, because of things like how long your red blood cells live, anemia, and how you wear your sensor. We always label GMI as an estimate and show your lab A1c beside it.
Patterns are observations, not causes. PeakRoutine looks at three comparisons in your last 28 days: short nights and your morning glucose, a walk after a meal and the rise, and stressful days and your daily average. Each shows medians and how many days or meals it's based on. Confidence never goes above "moderate", and we say "not enough data yet" when that's the truth.
One meal comparison is an anecdote. Sensor accuracy varies, and a lot changes between two days. Meal compare shows what was different, not what caused the difference.
The morning value isn't a fasting test. It's the middle sensor reading between 5:00 and 8:00 AM, which can include a late snack or the dawn rise. We don't call it fasting glucose.
Lab flags come from your lab. When we show a result as high or low, that's the flag printed on your report. The app doesn't judge your values.
Your data, in plain language
- Never sold, never used for ads. We don't share your health data with advertisers or data brokers.
- Encrypted. Your data is encrypted in transit and at rest.
- Lab report files stay on your iPhone. When you add a report, the file is read once to fill in the values and isn't stored on our servers. Only the values you confirm are saved.
- We name our AI provider. AI features, like the coach, lab report reading and meal photo recognition, send the data they need to Anthropic, our AI provider. The coach and lab report reading only work after you turn on AI personalization, and you can turn it off at any time.
- Delete any time. You can delete your health background in the app, or your whole account and its data.
PeakRoutine sells directly to you, not through a clinic or insurer, so we don't describe ourselves as "HIPAA compliant". Instead, we've told you above exactly what we do. The full details are in our privacy policy.
What we don't claim
We don't say PeakRoutine reverses prediabetes, prevents diabetes, lowers your A1c by any amount, or is clinically proven. We don't have studies showing that, and we won't use testimonials that imply results until we do.
What the research does show is that lifestyle change helps. In the Diabetes Prevention Program trial, people with prediabetes in the lifestyle group were 58% less likely to develop type 2 diabetes over about three years. That's why clinicians recommend lifestyle change, and it's background for why we built PeakRoutine. It is not a result of using our app.
PeakRoutine doesn't predict glucose, doesn't include human clinicians or dietitians, and isn't a CDC-recognized diabetes prevention program. If you'd like a structured program, ask your clinician about one near you.
Sources
- American Diabetes Association, Standards of Care in Diabetes: diagnostic criteria for prediabetes and diabetes.
- Centers for Disease Control and Prevention, preventing type 2 diabetes and the National Diabetes Prevention Program.
- Diabetes Prevention Program Research Group, New England Journal of Medicine, 2002: lifestyle intervention or metformin and the incidence of type 2 diabetes.
- Bergenstal RM et al., Diabetes Care, 2018: the glucose management indicator (GMI).
PeakRoutine is a wellness app. It does not diagnose, treat, cure or prevent any disease, and it is not a substitute for medical advice. Never start, stop or change a medication without your clinician. If a reading worries you, contact your clinician; in an emergency, call 911.
Last updated October 27, 2026. Questions or corrections: info@peakroutinehealth.com.