• 

  Sanis is an AI-driven preventive nutrition platform for pets

Sanis: AI-Driven Preventive Nutrition Platform

Market Gap & Objective

Existing pet-tech focuses on reactive remedies after symptoms appear. Sanis shifts the paradigm toward preventive nutrition. Our proof-of-concept demonstrates a pet-centralized system that transforms smartphone photos of raw food into fridge-ready feeding insights. By identifying nutritional deficits before they manifest as chronic illness, we remove the "kitchen-counter friction" of manual calculations and empower owners to manage pet longevity proactively.

What it does (Pet-Centralized Intelligence)

Sanis acts as a proactive health hub. By capturing a photo of raw ingredients, the platform cross-references the identified components with a unique pet profile (breed, age, weight, and health risks). A real-time dashboard tracks macronutrients (kcal/protein/fat) and provides instant alerts for nutritional deficits or excesses, ensuring every meal is balanced for that specific pet's needs.

How we built it (Gemini AI Vision)

The core engine is built on Gemini AI, which we use to automate ingredient recognition and volume estimation. We integrated a database of over 2,000 nutrients and developed a custom algorithm to translate pixel data into "household portions." To close the loop, we built a Smart Toppers Engine that generates specific, actionable suggestions like adding "10g of pumpkin" to bridge a fiber gap to allow owners to act on AI insights using simple household staples.

Challenges we ran into (Zero-Shot Spatial Reasoning)

A primary technical hurdle was marker-less volume estimation. To ensure a seamless user experience, we moved away from requiring physical reference objects. Instead, we leveraged the spatial reasoning capabilities of Gemini AI to infer portion sizes directly from the environment in the photo. Mapping diverse "raw visuals" to precise nutritional data required extensive prompt engineering to maintain accuracy across various preparation styles.

Accomplishments that we're proud of

We achieved near-instant processing speeds, allowing a user to move from photo to insight at the point of preparation. We successfully bridged the gap between "high-level AI analysis" and "practical kitchen advice," creating a tool that is fast enough to be used during a busy feeding routine.

What we learned

We discovered that in preventive health, latency is the enemy of adoption. A tool must be faster than the user's intuition to be effective. We also learned how much "hidden" health data exists in a simple bowl or chopping board of food that multimodal AI can extract but humans typically overlook.

What's next for Sanis

Our roadmap focuses on longitudinal health tracking. By aggregating daily nutritional data, we will build a predictive model to correlate diet with long-term health outcomes. Our goal is to alert owners to potential health shifts months before symptoms appear, positioning Sanis as the definitive digital hub for proactive pet care.

Built With

• gemini
• github
• netlify
• react
• typescript
• vite