आहार (āhāra) · Sanskrit: food, anything consumed. OS · operating system — the layer that schedules, enforces, and adapts.
AharaOS runs a daily protocol for everything you ingest — food, water, supplements, medicines — generated by an LLM that reads your conditions, your recovery, and your real body. One app. One loop. Not another calorie tracker.
Android + Health ConnectRazorpay subscriptions (₹299/₹999)Web dashboardApp Store + Play Store release
The intelligence inside
Nutrition is a reasoning problem, not a lookup problem. AharaOS treats it that way — every feature here is powered by an LLM grounded in your profile and a research-backed condition filter.
🧠
Gemini 2.5 Flash
LLM-generated daily protocol
Every morning the model reads your goal, conditions, sleep, HRV, yesterday's adherence, and macro gaps, then returns a structured plan — 10-15 items, timed, rationaled, condition-safe.
🛡️
Hard-gated filter
Condition-aware food safety
A research doc grounded in clinical guidelines is inlined into every prompt. Foods are gated by condition before ranking by macro. If safety is unknown, it's flagged — never guessed.
🔌
Model Context Protocol
MCP server for Claude Desktop
AharaOS ships a native MCP server so you can talk to your protocol from Claude Desktop or any MCP-compatible agent. "Reset today's checklist," "log 100g paneer at 11am" — all through one tool call.
⌚
Apple Health + Fitdays
Live signal ingestion
Sleep, HRV, SpO₂, resting HR, steps, active kcal, body-composition — all flow in automatically and feed the next LLM call. Your body rewrites your plan every morning.
📖
Structured outputs
Pydantic-typed LLM responses
Every protocol is a typed schema — no free-form text parsing, no hallucinated macros. Foods come from the DB; timings + rationales come from the model. Never both from the model.
💬
Coming in Phase 3
LLM coach · weekly review
The coach reviews seven days of adherence + body trends, writes a reflection in your voice, and course-corrects next week's protocol — surfaced only when adherence actually slips.
Why AharaOS is different
Every nutrition app in India treats you like a healthy athlete. You aren't. You have a thyroid, a knee, a cholesterol number, a family history. AharaOS treats those as inputs, not footnotes.
Generic trackers
Count calories. Ignore why you're eating.
Same plan for a 20-year-old athlete and a 55-year-old diabetic.
Food database is Western — no idli, no ragi, no sambar.
"Log your meal." That's the whole feature.
Rewards that reinforce restriction and guilt.
AharaOS
Filters every food through your conditions first. Macros come second.
Daily protocol rebuilt from last night's sleep, HRV, and adherence.
Indian vegetarian DB first — 1,145+ foods with condition tags.
LLM coach reviews your week and course-corrects tomorrow.
You've used one of these. They count calories well. None of them know your thyroid, your ACL, or your thali.
Feature
HealthifyMe
MyFitnessPal
Cronometer
AharaOS
Indian food database
✓
Partial
✕
✓ IFCT + INDB
Condition-aware filtering
✕
✕
✕
✓ Hard gate
LLM-generated daily plan
✕
✕
✕
✓ Rebuilt daily
MCP-native (talk via Claude)
✕
✕
✕
✓
Apple Health + Health Connect
✓
✓
✓
✓
Reacts to sleep / HRV
✕
✕
Display only
✓ Shapes today
India-native payments (UPI)
✓
✕
✕
✓ Razorpay
Starting price
₹500+/mo
$10/mo
$9/mo
Free in beta
Public information as of April 2026. None of the three have an LLM that gates food candidates by your research-backed condition profile before macro ranking — that is the gap AharaOS fills.
Six-month body-composition arc
The trajectory AharaOS plans for — month by month — starting from a real Fitdays baseline. Solid bars are actuals logged; dashed bars are the projected trajectory the protocol is working toward.
Actual (Fitdays fasted reading)Projected trajectory
Founder baseline, Apr 12 2026: 67.3 kg · 15.8% body fat · 52.2% body water · visceral 8.8 · body age 26. Target by Oct 2026: 11% body fat, body water 58%+, visceral below 7.
Transformation speed is individual. Dietary guidance — not medical advice.
Where you'll run AharaOS
Native per platform — no webview wrapper, no cross-platform tax. Deep HealthKit + Health Connect integration is the whole point. Here's what's live and what's on the way.
📱
iOS
SwiftUI · Swift 6 · HealthKit
Private beta
Native SwiftUI app pulling body, heart, sleep, activity, gait, workouts, mindful, hydration directly from HealthKit — 30+ types in one parallel sync. Currently in TestFlight dogfooding with the founder. Invite-only access for early private-beta users.
TestFlight · invite-only
🤖
Android
Kotlin · Compose · Health Connect
Coming soon
Native Kotlin + Jetpack Compose app with Health Connect integration for the same signal depth as iOS. Feature parity at public launch — both stores the same week, not iOS-first. Queued for Phase 3.
Play Store · launching with iOS
💻
Web
Next.js 15 · Tailwind · shadcn/ui
Coming soon
In-browser dashboard — today's protocol, body trends, adherence, LLM coach reviews. Shares the same REST contract as the mobile apps. Built for desk-based routines and admin review. Queued for Phase 3.
aharaos.com · dashboard
🔌
MCP server
Model Context Protocol · Python
Live today
Talk to your protocol from Claude Desktop or any MCP-compatible agent. "Reset today's checklist," "log 100g paneer at 11am," "what's my protein gap?" — all through native tool calls. Open source, self-hostable.
Open source · GitHub
Conditions it understands today
The food filter is a hard gate, not a ranking signal. If a food isn't safe for your condition, it never enters the protocol — no matter how well it hits your macros.
AharaOS is dogfooded daily by its founder before it ever reaches a user.
I tore my ACL and realised every nutrition app was recommending deep-fried samosas to "hit my carb target." I built AharaOS so that never happens to anyone again — not to my mother, not to my father, not to the 70 million Indians with thyroid disorders.
Sumanth Malipeddi · Founder · Tirupati, India · Dogfooding since April 2026
How to get in
Private beta. Manually approved, one at a time — so every early user gets a real onboarding.