GridWise AI

GridWise AI turns building energy data into clear, explainable actions that cut costs and carbon — in real time.

Comment

🌱 STORY

💡 Inspiration

Buildings consume a massive share of global energy, yet most energy tools stop at dashboards. They answer "What happened?", not "What should we do next?"

During research, we noticed that:

  • Facility managers often know energy is being wasted
  • But lack decision-ready insights
  • And almost never get quantified environmental impact

GridWise AI was inspired by the need to move from monitoring → optimization → measurable impact.

⚡ What it does

GridWise AI is an AI-powered decision support system for sustainable buildings.

It:

  • Analyzes building energy usage data
  • Predicts peak demand and inefficiencies
  • Simulates optimized scheduling scenarios
  • Quantifies energy savings, cost reduction, and CO2 avoided
  • Explains why each recommendation works in plain language

Instead of static charts, users get actionable, explainable recommendations.

🛠️ How we built it

System Architecture (High-Level)

  1. Data Ingestion

    • CSV-based building energy datasets (extensible to IoT streams)
    • Stored in PostgreSQL via Supabase

  2. AI Optimization Engine

    • Time-series demand forecasting (lightweight LSTM / heuristics)
    • Rule-based + data-driven inefficiency detection
    • Scenario simulation for optimized schedules

  3. Impact Analysis

    • Energy savings (%)
    • Cost reduction ($)
    • Carbon emissions avoided (kg CO2)

  4. Explainability Layer

    • LLM generates human-readable reasoning
    • Each recommendation is auditable and transparent

  5. Frontend Dashboard

    • Clean React + TypeScript UI
    • Focused on clarity, not data overload

🚧 Challenges we ran into

  • Designing AI that feels trustworthy, not "magic"
  • Balancing accuracy vs speed for live demo constraints
  • Making sustainability impact numerical, not vague
  • Avoiding overfitting to a single dataset
  • Explaining AI decisions clearly within a 2-minute demo window

Accomplishments that we're proud of

  • Built a fully explainable AI system, not a black box
  • Quantified real-world impact (energy, cost, carbon)
  • Delivered a clear live demo flow judges can understand instantly
  • Focused on decision support, not flashy automation
  • Designed a system extensible to real smart-grid deployments

What we learned

  • Explainability matters more than raw prediction accuracy
  • Sustainability tools must speak the language of operations and finance
  • Small efficiency gains scale into massive environmental impact
  • AI is most powerful when paired with clear decision logic
  • Judges value clarity and realism over buzzwords

What's next for GridWise AI

  • Real-time IoT and smart meter integration
  • Carbon-aware scheduling based on grid emission intensity
  • Multi-building and campus-level optimization
  • Automated ESG and sustainability reporting
  • Integration with demand-response and utility pricing APIs
Share this project:

Updates