Inspiration

People increasingly use general-purpose AI chatbots for nutrition and health-related questions. Conversational AI can make dietary self-monitoring easier and more accessible, but existing approaches often either require detailed logging or overwhelm users with too many nutritional metrics.

Zdravko AI was built around a simpler idea: for everyday nutrition, users do not need to track everything. A small number of carefully chosen indicators can capture a large part of what matters in practice.

The project therefore focuses on four daily pillars — calories, protein, fruit & vegetables, and water — while also asking a second question: how can a conversational nutrition assistant remain useful while having explicit boundaries for situations in which it should not simply answer the user’s request?

What it does

Zdravko AI is an evidence-informed conversational assistant designed to help adults monitor nutrition, hydration, physical activity and dietary balance in a practical, low-pressure way.

Its core principle is deliberate simplification. Rather than asking users to track dozens of nutrients, Zdravko focuses primarily on four daily pillars:

  • Calories provide a practical check on overall energy intake.
  • Protein helps support satiety, muscle maintenance and training goals, while encouraging a more nutritionally substantial food pattern.
  • Fruit & vegetables act as a simple proxy for dietary variety and contribute fiber, vitamins and minerals.
  • Water provides a straightforward hydration target.

Together, these pillars do not measure every aspect of diet quality, but consistently getting them broadly right provides a strong nutritional foundation without requiring obsessive or highly precise tracking.

The assistant is designed to work with imperfect conversational input rather than requiring users to weigh or numerically log every food item.

Zdravko AI also incorporates explicit safeguards governing how the assistant should respond to potentially unsafe situations. These include age-based functionality, calorie and weight-loss safety limits, handling of possible disordered-eating patterns, restrictions on diagnosis and supplement recommendations, escalation to appropriate professionals, and explicit communication of uncertainty.

Built With

  • chatgpt
  • claude
  • consensus
  • conversational-ai
  • elicit
  • genai
  • llm
  • prompt-engineering
Share this project:

Updates

Submission history