🏚️ AI Home Renovation Planner Agent

Meet your friendly home makeover helper! This agent looks at your room photos, plans the renovation, and creates beautiful “after” pictures. Built with Google ADK and Gemini 2.5 Flash.

Features

• � Smart Image Analysis
• 🎨 Photorealistic Rendering
• 💰 Budget-Aware Planning
• 📊 Actionable Roadmap
• 🤖 Real Multi-Agent Orchestration
• ✏️ Easy Iterative Edits

What this means in kid-speak: Show the app your room → it understands → it designs → it shows the “after” picture → you say “make it cream” → it changes it. Magic!

How It Works

The system uses a Coordinator/Dispatcher pattern with three friendly helpers:

1. Visual Assessor 📸

   • Looks at your room and inspo photos
   • Understands style, colors, layout
   • Spots problems and quick wins

2. Design Planner 🎨

   • Makes the plan that fits your budget
   • Chooses colors, materials, lights, and more
   • Lists what to do first

3. Project Coordinator 🏗️

   • Shows the “after” picture
   • Gives timeline, budget, and to‑do list

🚀 Quick Start

1. Get the code

   bash
   git clone 
   cd ai_home_renovation_agent
   

2. Install Python packages

   bash
   pip install -r requirements.txt
   

3. Add your Gemini API key

   • Windows (PowerShell, current window):

     powershell
     $env:GOOGLE_API_KEY="YOUR_API_KEY"
     

   • Windows (permanent, reopen terminal):

     powershell
     setx GOOGLE_API_KEY "YOUR_API_KEY"
     

   • macOS/Linux:

     bash
     export GOOGLE_API_KEY="YOUR_API_KEY"
     

   • Or create a .env file:

     env
     GOOGLE_API_KEY=YOUR_API_KEY
     

4. Launch ADK Web

   bash
   adk web
   

5. Open your browser → pick the app named ai_home_renovation_agent

🧸 One‑Minute Setup

• Step 1: Put your Google Gemini key in .env

  env
  GOOGLE_API_KEY=YOUR_API_KEY
  

• Step 2: Install packages

  bash
  pip install -r requirements.txt
  

• Step 3: Start the app

  bash
  adk web
  

• Step 4: Click ai_home_renovation_agent in the browser and follow the prompts

That’s it. If you can open a browser and paste a key, you’re done.

🎮 How to Use

• Upload a photo of your room. Optional: add an inspo image.
• Tell your budget (e.g., $5,000) and the style you like.
• Press go. The agent thinks, plans, and shows your new space.
• Change anything with simple words: “make cabinets cream” or “lighter floor”.

Usage Examples

Scenario 1: Current Room + Budget

python
[Upload photo of your kitchen]
"What can I improve here with a $5k budget?"

→ Agent analyzes your space, suggests budget-friendly improvements, generates rendering.

Scenario 2: Room + Inspiration

python
[Upload photo 1: your kitchen]
[Upload photo 2: Pinterest inspiration]
"Transform my kitchen to look like this. What's the cost?"

→ Agent extracts style from inspiration, applies to your room, provides budget + rendering.

Scenario 3: Text Only

python
"Renovate my 10x12 kitchen with oak cabinets and laminate counters. 
Want modern farmhouse style with white shaker cabinets. Budget: $30k"

→ Agent creates design plan and generates rendering from description.

Scenario 4: Iterative Refinement

python
[After initial rendering]
"Make the cabinets cream instead of white"
"Add pendant lights over the island"
"Change flooring to lighter oak"

→ Agent refines the rendering with your feedback.

Sample Prompts

• "I want to renovate my small galley kitchen. It's 8x12 feet, has oak cabinets from the 90s. I love modern farmhouse style. Budget: $25k"
• "My master bathroom is tiny (5x8) with a cramped tub. I want a spa-like retreat with walk-in shower. Budget: $15k"
• "Transform my boring bedroom into a cozy retreat. Thinking accent wall, new flooring. Budget: $12k"

Tools & Capabilities

• google_search: Finds renovation costs, materials, and trends
• estimate_renovation_cost: Calculates costs by room type and scope
• calculate_timeline: Estimates project duration
• generate_renovation_rendering: Creates photorealistic renderings
• edit_renovation_rendering: Refines renderings based on feedback
• Versioned artifacts: Automatic version tracking for all renderings

Multi-Agent Pattern

Demonstrates Coordinator/Dispatcher + Sequential Pipeline:

Coordinator (Root Agent)
    ├── Info Agent (quick Q&A)
    └── Planning Pipeline (Sequential)
          ├── Visual Assessor (image analysis)
          ├── Design Planner (specifications)
          └── Project Coordinator (rendering + roadmap)

Why this pattern?

• Efficient: Only runs workflows that are needed
• Modular: Each agent has clear responsibilities
• Scalable: Easy to add new features
• Production-ready: Real-world agentic system pattern

🧩 Requirements

• Python 3.10+
• Packages from requirements.txt:
  • google-adk>=1.15.0
  • google-generativeai>=0.8.3
  • python-dotenv>=1.0.0
• A Google Gemini API key in GOOGLE_API_KEY

🛠️ Troubleshooting

• "adk: command not found"
  • Install ADK: pip install google-adk
• "Invalid API key"
  • Make sure GOOGLE_API_KEY is set in your current terminal or .env
• App not visible in browser
  • Restart adk web
  • Make sure you ran it in the project folder
• Images not changing after edit
  • ADK Web may have image-edit limitations; use the “generate” step for fresh renders

❓ FAQ

• Do I need a GPU? No. This uses Gemini’s API.
• Can I use only text? Yes. Photos are optional.
• Windows support? Yes. Use PowerShell examples above.

📂 Project Structure

python
ai_home_renovation_agent/
├─ agent.py              # All agent definitions (root + sub‑agents)
├─ tools.py              # Rendering tools and utilities
├─ requirements.txt      # Python dependencies
├─ __init__.py           # Package init
└─ README.md             # This file

📜 License

MIT License. Use freely. Have fun renovating!