Zero Mile Meals

Zero-Mile Meals solves both the technical and social barriers preventing HGSF implementation through AI-powered adaptive optimization that respects community food cultures.

​

How It Works:

Community Learning: Community cooks, parents, and school staff apply their collective knowledge, creating meals children actually eat. Volunteers document meals naturally via WhatsApp using photographs, local language text, or voice messages. GPT-4o processes all modalities simultaneously (100+ languages supported), building a data-driven baseline capturing embedded community expertise about what works in practice.

 

Seasonal Integration: The platform integrates district-specific prices and availability from government agricultural bulletins weekly, tracking seasonal fluctuations automatically.

​

Cultural-Adaptive Optimization: A custom genetic algorithm evaluates 10,000+ meal combinations per optimization cycle, treating the community-optimized baseline as foundational wisdom to preserve. The cultural distance metric is heavily penalized in the fitness function—the algorithm explicitly minimizes disruption to meals the community knows children will eat while simultaneously balancing nutritional adequacy (protein, fiber, micronutrients) and cost constraints (less than 2% budget increase tolerance).

 

Continuous Adaptation: Weekly recalculation adapts recommendations as seasonal availability changes, maintaining nutrition standards through local sourcing across agricultural cycles.

 

Unlike static nutritionist-designed meal plans that ignore local knowledge, Zero-Mile Meals respects community expertise while adding computational nutritional optimization.