Havenly.ai

AI home concierge that transforms your space with zero effort - upload a photo to get decoration suggestions or a video to automatically sell your clutter

Comment 1

Havenly.ai - Your AI-Powered Home Concierge

Inspiration

Living in cramped spaces while juggling busy schedules, we noticed how overwhelming it can be to both make your space feel like home AND deal with clutter. Students and young professionals often struggle with two opposing needs: wanting to decorate and personalize their space, but also needing to declutter and potentially make money from unused items.

We were inspired by the idea of creating an AI assistant that could handle both sides of this equation - helping people discover what would make their space more cozy while also identifying what they could sell or rent out. The vision was to create a "home concierge" that takes the mental load off users and handles the entire process autonomously.

What it does

Havenly.ai is an intelligent home assistant with two powerful modes:

🛋️ Buy Mode - Make Your Room Cozy

  • Upload a photo of your room and get AI-powered decoration suggestions
  • Intelligent chat assistant that remembers your style preferences across sessions
  • Real-time product search that finds specific furniture and decor items
  • Personalized recommendations that improve over time using memory AI

📦 Sell Mode - Declutter Your Space

  • Upload a video walkthrough of your room for automatic item detection
  • AI identifies sellable items with timestamps, descriptions, and price estimates
  • Automatically generates compelling marketplace listings with optimized images
  • Posts items to student rental platforms like UseThis with one click
  • Handles the entire selling process from detection to listing

The app learns from your interactions, remembers your preferences, and provides increasingly personalized suggestions. It's like having a personal interior designer and marketplace assistant rolled into one.

How we built it

Frontend (React + TypeScript)

  • Built with React 19 and TypeScript for type safety
  • Styled with Tailwind CSS and custom design system featuring glass morphism
  • Smooth animations using Framer Motion for premium user experience
  • Responsive design that works seamlessly across all devices

Backend (Python FastAPI)

  • High-performance FastAPI server handling AI processing and marketplace integration
  • Nebius AI (DeepSeek-V3 and Qwen/Qwen2-VL-72B-Instruct) for image analysis, object detection, and natural language processing
  • Tavily API for real-time product search and market data
  • Mem0 for user preference learning and conversation memory
  • OpenCV for video processing and frame extraction
  • Appwrite for database storage and file management

AI Pipeline

  • Room analysis using computer vision to identify room type and suggest improvements
  • Video processing that extracts frames and identifies sellable objects
  • Smart pricing using market data and AI estimation
  • Natural language generation for creating compelling product listings
  • Memory system that learns user preferences and provides personalized recommendations

Integrations

  • UseThis student rental marketplace for automated posting
  • Real-time product search across multiple retailers
  • Image processing and storage for marketplace listings

Challenges we ran into

Deploying with ML libraries: While the entire Havenly.ai pipeline works smoothly in our local development environment, we weren't able to deploy the full backend due to compatibility issues with machine learning libraries on platforms like Render or appwrite. Specifically, the OpenCV-based video processing and some AI dependencies (used for frame extraction and Nebius model integration) exceeded the resource or compatibility limits of most free-tier hosting services. As a result, we opted to focus on a seamless local demo and documented deployment steps for future scaling.

AI Model Integration: Getting multiple AI services to work together seamlessly was complex. We had to carefully orchestrate calls between Nebius AI for vision tasks, Tavily for search, and Mem0 for memory, ensuring they all contributed to a cohesive user experience.

Video Processing Performance: Processing video files to extract frames and identify objects in real-time was computationally intensive. We optimized by implementing background processing with progress tracking and smart frame sampling.

Memory and Personalization: Building a system that actually learns and remembers user preferences across sessions required careful prompt engineering and data structure design with Mem0.

Marketplace Automation: Integrating with external platforms like UseThis required reverse-engineering their APIs and handling various edge cases in the posting process.

Real-time UI Updates: Coordinating between video processing, AI analysis, and frontend updates while maintaining smooth user experience required careful state management and WebSocket-like polling.

Accomplishments that we're proud of

🤖 Truly Autonomous AI: We built an AI that doesn't just give suggestions - it actually takes action. From analyzing your room to posting marketplace listings, Havenly handles the entire workflow autonomously.

🧠 Memory-Powered Personalization: The app genuinely learns from your interactions and gets better over time. It remembers your style preferences, room types you've worked on, and products you've liked.

🎨 Premium User Experience: We created a beautiful, responsive interface with smooth animations and micro-interactions that feels like a premium consumer app, not a hackathon project.

⚡ Real-time Processing: Successfully implemented complex video processing and AI analysis that works in real-time with progress tracking and user feedback.

🔗 End-to-End Integration: Built a complete pipeline from room analysis to marketplace posting, including real product search and actual platform integration.

📱 Production-Ready Architecture: Designed with scalability in mind using modern tech stack, proper error handling, and deployment-ready configuration.

What we learned

AI Orchestration: We learned how to effectively combine multiple AI services to create a cohesive experience. Each AI has its strengths, and the magic happens in how you orchestrate them together.

User Experience in AI Apps: Building AI-powered apps requires different UX considerations - users need to understand what's happening, see progress, and feel in control even when AI is doing the work.

Memory Systems: Implementing persistent memory that actually improves user experience is harder than it seems. It requires careful prompt engineering and understanding of how to structure learned information.

Computer Vision for Real Applications: Working with real user-generated content (photos and videos) taught us about the challenges of computer vision in uncontrolled environments.

Marketplace Integration: We learned about the complexities of integrating with external platforms and the importance of robust error handling and user feedback.

Full-Stack AI Development: Balancing AI processing on the backend with responsive frontend experiences requires careful architecture and state management.

What's next for Havenly.ai

🏪 Real Marketplace Integration: Connect to actual selling platforms like Facebook Marketplace, eBay ( I attemped it this time and got my account banned :(( ) , and Craigslist for broader reach and real transactions.

📱 Mobile App: Native iOS and Android apps with camera integration for seamless photo and video capture.

🥽 AR Visualization: Implement AR features so users can see furniture and decor in their actual space before buying.

🤝 Social Features: Allow users to share room transformations, get feedback from friends, and discover trending decoration ideas.

🏢 Professional Services: Connect users with interior designers, handymen, and other home improvement professionals.

🏠 Smart Home Integration: Integrate with IoT devices to provide smart home recommendations and automation suggestions.

💰 Revenue Sharing: Implement commission-based revenue model with partner retailers and marketplaces.

🌍 Global Expansion: Expand to international markets with localized product search and marketplace integrations.

🎯 Advanced Personalization: Implement more sophisticated AI that understands lifestyle, budget constraints, and long-term decoration goals.

📊 Analytics Dashboard: Provide users with insights about their space, spending patterns, and decoration journey over time.

Havenly.ai represents the future of home management - where AI handles the complexity so you can focus on enjoying your space. We're excited to continue building toward a world where everyone can have a beautiful, organized home with zero effort.

Built With

Share this project:

Updates