Architec

Architec is a next-generation energy audit platform that transforms hours of manual utility bill analysis and building inspections into a 10-minute automated pipeline. By combining high-accuracy OCR with site-specific AI reasoning and interactive 3D visualizations, Architec provides building owners with professional-grade energy insights at zero cost.

The Product

Traditional commercial energy audits cost between $15,000 and $50,000 and take months to complete. Architec bypasses this overhead by allowing users to simply describe their building and upload 12 months of utility bills.

Key Features

• AI Bill Parsing: Automatically extracts kWh, therms, peak demand, and costs from PDF and image-based utility bills using Google Gemini 3.0 Flash.
• Weather-Aware Normalization: Separates heating, cooling, and baseload consumption using statistical regression and local weather data.
• 3D Digital Twin: An interactive Mapbox and Three.js visualization that layers solar heatmaps and volumetric thermal plumes onto building models to show energy loss and upgrade potential.
• AI Analysis Manager: A conversational voice assistant powered by ElevenLabs that allows owners to explore their audit results and conservation measures through natural dialogue.
• Prioritized ECMs: Ranked Energy Conservation Measures (ECMs) with estimated annual savings, simple payback periods, and implementation complexity.

The Tech Stack

AuditAI was built using a modern, agentic development workflow and high-performance AI services.

AI & Reasoning

• Google Gemini 3.0: Acts as the core intelligence layer for high-accuracy bill OCR and for generating expert-level engineering reports (ASHRAE Level II style).
• ElevenLabs: Powers the "AI Analysis Manager" using unbeatable voice quality and low-latency interaction. We used the Conversational AI widget with Dynamic Variables to inject site-specific audit data into the agent's context window.
• Claude / Claude Code: Served as our agentic team-programmer for architecting complex Three.js thermal mapping shaders and orchestrating the multi-stage AI pipeline.
• Appifex AI: Leveraged for rapid UI and backend prototyping directly from natural language prompts, bypassing manual boilerplate setup.

Visualization & Spatial Data

• Mapbox GL & Three.js: Deliver the 3D building shell and spatial rendering.
• Google Solar API: Provides high-fidelity solar radiance data used to generate the precision solar heatmaps on the 3D digital twin.

Infrastructure

• Next.js 16 (Turbopack): Modern, high-velocity frontend framework.
• FastAPI (Python): High-performance backend for energy analytics and data processing.
• DigitalOcean App Platform: Reliable, fast deployment pipeline for both frontend and backend services.

Getting Started

Prerequisites

• Node.js (Latest LTS)
• Python 3.11+

Installation

1. Clone the repository and install dependencies:

   npm ci
   python3 -m venv .venv
   .venv/bin/pip install -r backend/requirements.txt

2. Configure Environment Variables: Create a .env file in the root directory:

   GEMINI_API_KEY=your_key
   ELEVENLABS_AGENT_ID=your_id
   NEXT_PUBLIC_MAPBOX_ACCESS_TOKEN=your_token
   # Add other keys as needed for Solar and NOAA

3. Run the Application: Open two terminals from the repository root:

   Terminal 1 (Backend API):

   npm run backend:dev

   Terminal 2 (Frontend Interface):

   npm run dev

4. Access the App: Visit architec.tech to start your first audit.

---

Built for WildHacks 2026.