WhatsApp AI Meal Planner Prototype for Quick Commerce
How we turned health intent into a one-tap grocery cart with a working prototype built in 2-3 days
Executive Summary
From business hypothesis to working, cost-modeled prototype
Personalized 3-day meal plan generation
Ingredients mapped to real catalog products with prices
AI cost per generated meal plan
The Challenge
Quick-commerce and online grocery platforms face a gap between customer health intent and actual grocery purchase. A customer may want to eat healthier, lose weight, gain muscle, or follow a more balanced diet — but they still have to plan meals, calculate calories, estimate macros, write a grocery list, search for each item in the grocery app, and add products one by one. That friction causes most users to drop off before purchase. Nutrition apps and grocery apps are disconnected, so advice never translates into a purchasable cart. For the platform, the result is smaller baskets — 4-5 items on average — weaker retention hooks, and a missed opportunity to serve health-conscious customers.
Key Pain Points
- ✕Health intent never converts because meal planning and grocery shopping are disconnected
- ✕Manual recipe search, calorie math, and item-by-item cart building cause drop-off
- ✕Small 4-5 item average baskets and weak retention hooks
- ✕AI product ideas pitched as abstract concepts instead of working demos
Our Solution
Amasa built a working WhatsApp AI meal planner prototype for the Indian quick-commerce market — a speculative, demo-led prototype rather than a live client deployment. The bot collects a user's health profile on WhatsApp, including age, weight, goals, and dietary preferences, then generates a calorie-targeted 3-day Indian meal plan with real meals such as dal, roti, and sabzi. Meal generation runs on the Claude API, calorie targets use the Mifflin-St Jeor equation with macro splits based on user goals, and nutrition is grounded in NIN (National Institute of Nutrition) data. Each ingredient is mapped to an actual catalog product with price across 200 mapped SKUs, and the user taps once to load the full grocery list into their basket. The prototype was built in 2-3 days as a complete demo package: live SKU mapping, nutrition grounding, a recorded demo video, pitch deck, UI mockups, proposal PDF, and modeled unit economics of about ₹3 per generated plan.
Implementation Approach
Hypothesis & Flow Design
Framed the intent-to-cart gap and designed a WhatsApp-native health profile and meal plan flow
Nutrition & Generation Engine
Combined Claude API generation with NIN nutrition data, Mifflin-St Jeor calorie targeting, and goal-based macro splits
Catalog Mapping & Cart
Mapped meal-plan ingredients to 200 real catalog SKUs with prices and built one-tap basket loading
Demo Package
Delivered a recorded demo video, pitch deck, UI mockups, proposal PDF, and per-plan unit economics
The Results
The implementation delivered transformative results across all key metrics, with immediate impact on operational efficiency, accuracy, and customer satisfaction.
Impact Metrics
| Metric | Before | After | Improvement |
|---|---|---|---|
| Meal planning | Manual planning and item-by-item ordering | Under 15 seconds plan generation | Near-instant |
| Intent-to-cart bridge | No link between advice and SKUs | 200 mapped SKUs with one-tap cart | Fully connected |
| Basket size | 4-5 items average (per pitch) | 12+ items projected | 2-3x projected |
| Sales conversation | Abstract AI concept deck | Working demo with real unit economics | Demo-led |
Key Takeaways
- A working prototype with real APIs, catalog data, and unit economics beats a concept deck in AI product sales
- WhatsApp-native flows remove app-download friction for the Indian quick-commerce market
- Grounding meal generation in NIN data and the Mifflin-St Jeor equation keeps AI output nutritionally credible
- Basket-size and revenue figures are pitch projections — the measured proof point is speed: hypothesis to working product in a weekend
Inside the Prototype
The Journey Before
For a customer who decides to eat healthier, the path from intent to a filled grocery cart looked like this:
- Decide to eat healthier.
- Search for recipes.
- Estimate calories and macros.
- Write a grocery list.
- Open the grocery app.
- Search every ingredient manually.
- Add items one by one.
- Adjust quantities and checkout.
Nutrition apps and grocery apps were disconnected – advice never translated into a purchasable cart. Most users dropped off before purchase, leaving platforms with 4–5 item baskets and weak retention hooks.
The Prototype Flow
The WhatsApp bot collects a user’s health profile – age, weight, goals, and dietary preferences – then generates a calorie-targeted 3-day Indian meal plan with real meals such as dal, roti, and sabzi. Each ingredient is mapped to an actual catalog product with price, and the user taps once to load the full grocery list into the basket.
- Claude API for meal-plan generation
- WhatsApp as the user interface
- 200 mapped SKUs in the prototype catalog
- NIN (National Institute of Nutrition) data for nutritional grounding
- Mifflin-St Jeor equation for calorie targeting
- Macro splits based on user goals
Prototype Results
| Metric | Before | After |
|---|---|---|
| Meal planning | Manual planning and item-by-item ordering | Under 15 seconds plan generation |
| Catalog mapping | No bridge from intent to SKUs | 200 mapped SKUs |
| AI cost | n/a | ₹3 per generated plan |
| Basket size | 4–5 items average, per pitch | 12+ items projected |
| Revenue impact | n/a | 2–3x basket-size increase projected |
The Impact
This was not a live client deployment, so the basket-size and revenue figures are pitch projections, not measured outcomes. The real proof point is Amasa’s speed and product thinking: the team turned a business hypothesis into a working, cost-modeled AI product in a weekend.
Instead of pitching an abstract AI idea, Amasa created a functional WhatsApp demo using real APIs, real catalog data, real nutrition logic, and real unit economics – delivered alongside a recorded demo video, pitch deck, UI mockups, and proposal PDF.
The Takeaway
Amasa showed how a quick-commerce platform could turn health intent into a filled grocery cart through a WhatsApp-native AI meal planner. The prototype generated personalized meal plans in under 15 seconds, mapped ingredients to 200 catalog SKUs, and demonstrated a projected path from 4–5 item baskets to 12+ item baskets.
Quick Facts
Industry
E-commerce
Solution Type
Generative AI
Published
August 24, 2026
Technologies Used
Related Resources
Ready to Achieve Similar Results?
Let's discuss how we can transform your business with AI solutions tailored to your specific needs