Methodology
- Choose sources. We prefer national and international guidelines (WHO, ICMR–NIN, AHA/ACC, ADA, NICE, ACR, Ministry of Ayush) and large systematic reviews. Patient-education sites such as the NHS, Mayo Clinic and Harvard's Nutrition Source help us explain ideas plainly.
- Write rules. Each recommendation becomes a rule with clear inputs (age, conditions, measurements, goals) and outputs (targets, food choices, exercise doses, removed poses). Each rule carries tags pointing to its sources.
- Safety first. Safety rules run before everything else: emergency symptoms, alert thresholds, contraindications and groups who need extra care, such as teens, pregnant users and people with heart conditions.
- Indian defaults. We use Asian BMI and waist cut-offs, ICMR–NIN protein needs and Indian portion units, and build menus from regional cuisines.
- Personalise and adapt. Your measurements and logs set targets. A weekly review adjusts the plan based on how much of it you completed and how your readings moved.
Clinical review workflow
Every rule, recipe, exercise, pose and article moves through four states: draft, in review, approved and retired. A qualified reviewer for that area (physician, registered dietitian, physiotherapist or certified yoga therapist) must approve it before it reaches users. Reviews are logged with the reviewer, date and source version. We re-review the full rule set every quarter and whenever a major guideline changes.
Current status: rules and content in this release are drafts awaiting clinical sign-off.
Content principles
- All text, recipes, plans and illustrations are created by Ayush.
- We paraphrase guidance and cite it; we do not copy text, plans, images or branded programmes.
- We describe benefits in cautious language ("may help", "is linked with") unless strong evidence supports more.
- Projected timelines are shown as estimates, with the basis explained.
Use of AI
When you switch on AI coaching, Ayush uses a large language model (Anthropic's Claude) to explain your plan in plain language, reword it in your chosen language and fit schedules around your day. The AI works inside fixed limits:
- It cannot add, remove or change safety rules, targets or contraindications.
- It receives plan details without your name, contact details or Health ID.
- Its replies are checked for red-flag topics, which are routed to the safety screens.
- It never comments on medicines beyond advising you to ask your doctor.
- AI coaching is off unless you switch it on, and you can switch it off at any time.
Limitations
- Ayush relies on information you enter. It cannot examine you or verify readings.
- Nutrition values are approximate for typical home recipes.
- Guidelines are written for populations; your doctor may set different targets for you.
- Ayush does not cover type 1 diabetes, pregnancy-specific plans, kidney disease diets, eating disorders or post-surgical rehabilitation. Please follow specialist advice for these.