Perfect Plate
PerfectPlate is an AI nutrition app built for India, a country where health-conscious professionals, fitness enthusiasts, and people managing conditions like diabetes are all let down by the same problem: nutrition apps were not built for them.
During our IIT Roorkee Product Management Certificate programme, our study group identified this gap together. Mohammed Mujeeb shaped the early product vision and designed the initial UI concepts. Prakruti Kothari, Praveen K S, Sowmya Krishna, and Rashmi Nagaraj contributed as co-ideators. It was guided by our mentor Kapil Dongare.
When the programme ended, We kept in touch and I decided took the idea forward as a product concept, which is ready to be implimented. This is the full case study covering how the problem was defined, how the product was designed, the decisions made along the way, and what a PM can build with structured thinking and the right tools.
PerfectPlate AI nutrition app is a product I designed for a problem that affects 77 million people in India and gets worse every year. Indian food is among the most nutritionally complex in the world. Yet every calorie tracker on the market was built on Western food databases that do not recognise dal, roti, or idli sambar. Combine that with the reality that most Indian diabetics manage their condition without consistent access to a dietitian, and you have a clear product gap. This case study covers how I designed PerfectPlate to fill it.
What Is PerfectPlate?
PerfectPlate is an AI-powered nutrition and health tracking mobile app for iOS and Android. It is built specifically for the Indian market, using the Indian Food Composition Tables from the National Institute of Nutrition as its nutritional database, supplemented by AI food recognition via Claude Vision API.
The app serves three broad user groups: health-conscious individuals who want smarter eating, fitness enthusiasts tracking macros for performance, and people managing specific health conditions like diabetes, PCOD, or hypertension through diet.
Three capabilities define the product. First, an AI Food Analyser that identifies any Indian meal from a photo and returns full nutritional data in under 8 seconds. Second, a Diet Contradiction Engine that flags foods that clash with your personal health goal in real time. Third, Medical Document AI that reads your prescription or blood report at onboarding and uses that data to personalise every recommendation the app makes. No other nutrition app in India currently combines all three.
The Problem PerfectPlate Solves
India has a growing health crisis and no intelligent digital tool built around it. The global health and wellness app market is projected to reach $100 billion by 2030, growing at 20% annually, yet the apps available to Indian users today were built for entirely different contexts.
The calorie estimates for Indian regional dishes vary by up to 160 percent across major apps. A fitness enthusiast in Bengaluru tracking his macros for muscle gain gets the same generic interface as a diabetic in Lucknow trying to manage blood sugar. That is not personalisation. That is a product that was never designed for either of them.
Manual logging is the first failure point. Studies on nutrition app retention consistently show that logging-dependent apps see drop-off within two weeks. Urban professionals with full schedules simply do not have time to open an app and manually enter every ingredient in a home-cooked meal. The second failure point is context.
An app that tells you that you consumed 144 grams of carbohydrates today means nothing without knowing your goal, your health condition, your medication, or your activity level. The third failure point is Indian food recognition.
The biggest apps on the market, including Healthify Me and MyFitnessPal, have incomplete or inaccurate databases for Indian regional foods. Chole bhature from a Delhi dhaba, poha from a Pune breakfast stall, rasam rice from a Chennai home kitchen: these meals exist largely outside the datasets that power mainstream nutrition AI.
PerfectPlate: AI Nutrition App Build for India
You can click on respected Tabs Below to see some innovative info and features!
PerfectPlate was designed around all three failure points. The idea originated during a product management assignment in our IIT Roorkee PM Certificate programme, where the study group identified this gap collectively. When the programme ended, I took the concept forward as a full product, building every deliverable from PRD to GTM strategy to live interactive prototypes.
PerfectPlate is an AI nutrition app built specifically for India. It uses Claude Vision to identify any Indian meal from a photo, pulls nutritional data from the Indian Food Composition Tables, and returns a health score in under 8 seconds. It reads uploaded prescriptions and blood reports to extract HbA1c values, medications, and conditions. It then personalises every recommendation around the user’s specific health profile.
The design principle throughout was zero friction. A diabetic in Lucknow who is already managing a complex condition should not have to learn a new app workflow. Snap, analyse, act. Three steps, under 10 seconds, zero manual entry.
Ethan, 41, diabetic. Opens MyFitnessPal to log his dal tadka and roti lunch. Searches for 5 minutes. Finds “Dal, generic Indian” with no regional variation. Gives up. Does not log. Has no idea if his lunch was safe for his HbA1c.
Ethan snaps a photo of his lunch. PerfectPlate identifies dal tadka, 2 rotis, and a side of salad in 8 seconds. Health score 7.8 out of 10. Diabetic-friendly badge shown. His Metformin dosage and HbA1c from last month’s report are already factored in. One tap to log.
Process Intro
The build had four phases. The biggest challenge throughout was not the AI — it was the data. Indian food nutrition data is fragmented, inconsistently formatted, and largely unavailable in machine-readable form.
Results
The investor one-pager and prototype have been validated through product feedback sessions. Revenue projections are based on comparable Indian health app benchmarks.
PM Decisions And Tradeoffs
The most consequential product decision in PerfectPlate was the phased launch structure. The original vision included smartwatch integration, Food Finder, dietitian tie-ups, and AI meal plans all at launch. After mapping the development complexity and team requirements, it became clear that attempting everything in Sprint 1 would either collapse the timeline or produce a low-quality first version of every feature.
The decision was to launch Phase 1 with four focused capabilities: AI Food Recognition, Calorie and Nutritional Tracking, Diet Contradiction Detection, and User Feedback. This reduced the MVP scope by roughly 40 percent but gave the team full control over every Phase 1 deliverable and a realistic path to 10,000 user sign-ups before adding complexity.
The data problem is the product problem
I underestimated how fragmented Indian food nutrition data is. IFCT is the best source available but it has gaps for regional dishes. Sourcing, cleaning, and structuring this data is as much a product challenge as the AI layer on top of it.
Prescription AI is the real differentiator, not food AI
Every competitor can eventually add Indian food recognition. No one has built prescription-aware nutrition yet. That is the moat. I should have led with it in every conversation instead of leading with the food analyser, which sounds more impressive but is more easily replicated.
Build for the caregiver, not just the patient
In India, health decisions are often made by family members, not just the patient. A daughter managing her father’s diabetes remotely is as valuable a user as the diabetic himself. The Family Plan at Rs.199 per month is not just a pricing tier. It is a different product for a different user who I nearly overlooked entirely.
Pricing Decisions
The second major tradeoff was pricing. The Initail GTM documents set the launch price at Rs.750 per month with a 14-day free trial. After reviewing the target user income levels, Rs.10 LPA for Ethan and Rs.80 LPA for Kapil, and competitive pricing from Healthify Me and similar apps, I kept the Premium tier at Rs.299 per month for the portfolio and investor positioning.
The reasoning is that Rs.299 is psychologically below the Rs.300 threshold, sits comfortably within a Rs.10 LPA budget, and is a price point that converts trial users to paid at a higher rate than Rs.750.
The decision I would make differently is the Medical Document AI feature placement. Currently it is in onboarding, which adds friction to the first-use experience. A leaner approach would be to make document upload optional at onboarding and prompt it contextually after the user’s first food log, when they have already experienced the product’s core value and are more likely to engage with the deeper personalisation.
Metrics and Success
The north star metric for PerfectPlate is daily active users who log at least one meal per day. This single metric captures both acquisition and the core behaviour the product is built around. A user who downloads the app but never logs a meal has not experienced the product. Supporting metrics across three time horizons:
- Day 60. 10,000 Phase 1 sign-ups. 40 percent of users return on Day 7. Average of 2.1 food logs per active user per day. AI food analyser used by at least 60 percent of active users.
- Day 120. 25,000 total downloads. 30-day retention above 35 percent. Free to Premium conversion rate above 8 percent. Net Promoter Score above 40.
- Month 09. 50,000 downloads. Rs.7.5 lakh MRR. Phase 2 features (smartwatch, Food Finder) live and contributing to a measurable retention uplift. At least 100 verified food reviews submitted via the Perfect Points system.
The single metric I would watch most closely in the first 60 days is the drop-off point in the onboarding flow, specifically how many users complete the medical document upload step. High completion means the Medical Document AI is landing as a differentiator. Low completion means the feature needs to be repositioned, either simplified or moved post-onboarding as discussed in the PM Decisions section above.
India has 77 million adults living with diabetes, the second highest count globally. IDF Diabetes Atlas, 2021.
Another health tech PM case study — how I designed an AI tool that translates clinical blood reports into plain language.