Sign in
Private Beta · India-first

The food
operating system.

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.

🇮🇳 Built for India · Mumbai data residency
Apple Health + Fitdays scale
🔒 DPDP compliant
Today
🔥 4 day streak
5 of 9 done · 120g protein target
06:30
Turmeric + warm lemon water
Anti-inflammatory · ACL recovery
07:15
Banana + Pintola PB toast
24g protein · morning gap closed
08:00
Creatine 5g (Wellcore)
Muscle retention on cut
13:00
Whey scoop + vitamin C
24g protein · iron absorption
17:30
Trainer session · upper-body
ACL-safe · no loaded knees
Body
fasted · today
67.3 kg BASELINE
15.8%
body fat
53.2 kg
muscle
8.8
visceral
52.2%
body water
1603
bmr
26
body age
❤️
Import from Apple Health
body · heart · sleep · activity
📸
Import from Fitdays screenshot
visceral fat · body water · BMR · body age
NEW
Coach
daily review
Yesterday
You closed protein at 118g, water at 2.6L — great recovery day. Sleep dipped to 6.4h, HRV down 12%.
Today's tweak
Bumping your morning protein to curd rice + sprouts and holding creatine. Cardio swapped for upper-body only — ACL load stays low.
This week
Body fat trend: 15.8% → 15.1%. Body water +1.6%. On track for month-1 target.
Food search
condition filter on
Active: ACL recovery · Hypothyroid · High LDL
Samosa (deep-fried) Avoid
Raw cabbage salad Reduce
Sambar + moong dal Prefer
Ragi mudde Prefer
Paneer butter masala Reduce
Amla + turmeric shot Prefer

Where we are right now

AharaOS is being built in public. Here's what's shipped, what's live today, and what's next — honest, not marketing.

Phase 1
Shipped
MCP foundation + LLM protocols
FastMCP server, Supabase (Mumbai), Gemini 2.5 Flash wrapper with structured outputs, checklist MVP running on the founder's phone daily since April.
FastMCP server + 5 core tools Supabase schema + RLS Gemini protocol generator Checklist dogfooded 14+ days
Phase 2 · now
In progress
iOS app + condition filter + native HealthKit
Onboarding, reward loop, condition-aware recommendations, native HealthKit firehose (body + heart + sleep + activity + gait + workouts), 1,145-row Indian food DB with inflammation/goitrogen tags.
Conditional onboarding agent Native HealthKit firehose (30+ types) Indian food DB v2 tagging iOS app v1 — private dogfood Reward loop + streaks
Phase 3 · soon
Queued
Public launch · iOS + Android + Web
Native Android (Health Connect), Razorpay UPI subscriptions, Next.js web dashboard, LLM coach weekly reviews, nutritionist marketplace.
Android + Health Connect Razorpay subscriptions (₹299/₹999) Web dashboard App 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.
  • Positive reinforcement only. Missed items stay silent.

AharaOS vs the apps you've already tried

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 databasePartial✓ 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 / HRVDisplay only✓ Shapes today
India-native payments (UPI)✓ Razorpay
Starting price₹500+/mo$10/mo$9/moFree 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.

Inflammation / ACL recovery High cholesterol (LDL) Hypothyroidism Hyperthyroidism Hypertension Type 2 diabetes Insulin resistance PCOS / PCOD CKD (advisory — consult doctor)

Built by someone living it

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.

How to get in

Private beta. Manually approved, one at a time — so every early user gets a real onboarding.

1
Request access
Fill the short signup — name, email, goal, diet. Under a minute.
2
Get your credentials
Username, password, and a personal Apple Health ingest token — emailed to you after manual review.
3
Finish onboarding
Nine conditional questions — goal, conditions, routine, training. First protocol drops immediately.
4
Connect Apple Health
Paste the token once into Health Auto Export. Sleep, HRV, activity, body composition sync automatically after.

Ready to stop counting calories
and start healing?

Invite-only beta. Indian vegetarians with real conditions welcome.

Request Access →