EarthWise – AI-personalized eco & wellness coach

🌍 About the Project

EarthWise is an AI-powered sustainability and wellness coach that tailors eco-friendly habits based on your actual purchases and lifestyle. By linking your shopping behaviour, we turn everyday actions into meaningful impact.

💡 Inspiration

We noticed that most habit-tracker apps give generic advice like “drink more water” or “walk more”. But behind every person’s purchases and routines lies a unique ecosystem of behaviour.
We asked: What if we could use purchase data + AI to create truly personalized tasks?
This is where the integration of the Knot SDK and our AI planner comes in.

🛠 How We Built It

Frontend: Next.js (App Router) + TypeScript + Tailwind + shadcn/ui
Backend/Data: Supabase (auth + Postgres)
Transaction Insight: We built a Next.js API route (/api/knot/sync) that proxies to Knot’s dev endpoint and transforms SKU-level transactions into personalized insights and task suggestions.
AI Planning: Our endpoint is structured to accept {goals, purchases} and output a multi-step plan, ready to plug into providers like Google Gemini or OpenAI.
Impact Metrics: We compute a user’s “points → real world impact” using a formula like
$$ \text{kgCO}_2 = 0.45 \times \text{points} $$
and show them as trees planted or liters of water saved.

🔍 Challenges We Faced

Integrating the Knot SDK in a hackathon timeframe: we hit domain allow-listing / CORS issues, so we created a server-side fallback path to ensure the demo would always show insights.
Keeping task completions and user state synced across pages in Next.js. We moved persistence into Supabase instead of only component state.
Balancing realism and time economy: our AI planning endpoint returns mock steps for now, but we built the architecture so it can later call a real LLM provider.

🎯 What’s Next

Enable the Knot web-modal link flow for real account connection.
Swap mock AI plan with a live model from Google Gemini and let users choose.
Deploy to Vercel + support real task history so the “Predictive Intelligence” view becomes dynamically derived.

🧠 Why It Matters

Creativity: We bridge transaction data and personalized eco-tasks, something not many sustainability apps offer.
Utility: Users get tailored suggestions based on what they actually buy, not just what they say.
Charity & Impact: Points convert into measurable impact (kg CO₂, trees, water), which resonates with the “societal need” dimension of judging.
Avidity: We built a polished UI, a data-driven backend, and integrated sponsors (Knot) in under 36 hours—demonstrating our team’s drive and execution speed.

Thank you for reviewing EarthWise. We can’t wait for you to test it and see how much impact can come from everyday actions.