GridDaddy

Inspiration

Arizona has massive solar generation during the day and the 2ed largest clean nuclear baseload in the country (Palo Verde) running 24/7. Yet, from 5 PM to 9 PM, the sun sets, everyone comes home, and HVACs kick into overdrive—forcing the grid to spin up dirty "peaker" fossil fuel plants.

We realized that if users just shifted their appliance usage (pool pumps, dishwashers, EVs) by a few hours, they could run their homes on nearly 100% clean energy. The concept of "Demand Response" exists for large factories, but we wanted to bring it to the individual. We built GridDaddy to make load-shifting personal, gamified, and AI-driven.

What it does

GridDaddy is a personal carbon footprint tracker and AI-powered energy coach. Users log the daily usage of their home's heavy appliances, and the app cross-references it against real-time grid carbon intensity.

It calculates your exact CO₂ emissions, visualizes your 7-day trends, and pinpoints your "worst energy habits." Using our Gemini-powered AI chatbot, GridDaddy analyzes your historical data and local grid mix to provide conversational, highly actionable advice—like telling you exactly when to run your dryer tomorrow to maximize your use of the nuclear baseload and avoid gas peaker plants.

How we built it

Frontend: We built a highly polished, responsive dashboard using React, TypeScript, and Vite. We used Tailwind CSS and shadcn/ui to create a premium dark-mode aesthetic, and Recharts for beautiful data visualization.

Backend & Data: We used Supabase with Auth0 to handle user authentication, profiles, and historical check-in data. Vercel Serverless Functions power our backend API routes.

AI Integration: We integrated the Gemini API to act as our Smart Energy Coach. We pass the user's specific home size, appliance history, and live grid data into the prompt context so the LLM can generate highly personalized scheduling insights.

Challenges we ran into

Data Translation: Translating complex grid metrics (gCO₂eq/kWh) into understandable, personalized metrics (e.g., "lbs of CO₂ emitted by your specific HVAC system") required careful normalization and math.

AI Context Management: Feeding the AI the user's historical appliance data alongside live grid intensity in a way that produced accurate, non-hallucinated advice took significant prompt engineering.

UI/UX Design: Energy data is historically boring. Designing an interface that looked like a modern, premium SaaS product rather than a utility bill required a lot of focus on micro-animations, color gradients, and brand consistency.

Accomplishments that we're proud of

We successfully gamified a complex environmental problem (grid load shifting) into a consumer-friendly app that users actually want to check daily.

We built a working AI interface that truly understands the user's local energy mix and personal habits, providing real value rather than generic advice.

We successfully highlighted the environmental importance of nuclear baseload energy within the app's core mechanics.

What we learned

We learned an incredible amount about how the electrical grid actually works—specifically the critical role of nuclear baseload and why evening "peak hours" are so environmentally damaging. We learned how to securely pass structured database contexts into an LLM to generate personalized, data-driven insights.

We realized that tiny behavioral shifts (like delaying a dishwasher by 2 hours) can have a massive, measurable impact on carbon emissions if scaled across a community.

What's next for GridDaddy

Smart Home Integration: We want to connect GridDaddy directly to smart home APIs (like Nest or smart plugs) so the AI can automatically delay appliances without requiring manual check-ins. Predictive Grid Forecasting: Implementing time-series ML models to forecast grid intensity 48 hours in advance, allowing the app to send proactive push notifications.

Community Leaderboards: Adding social features that allow neighborhoods to compete to see who can save the most CO₂ by flattening the curve together.