Back to Work

AI-Powered Design Intelligence Meets Seamless Product Discovery

Interior VisionProduct Development

Interior Vision is a reimagined software platform for interior designers and architects featuring intelligent project management, computer vision, and AI capabilities. When a leading interior design SaaS startup needed to bridge the gap between design inspiration and product procurement, they turned to our team to build something the industry had never seen: an AI-powered platform that instantly identifies furniture in any image and finds purchasable matches across the web. The result? Interior Vision, a full-stack platform combining Claude Vision AI, real-time collaboration, and browser-native workflows that transforms how designers move from mood board to spec sheet.

Industry

Interior Design Technology, Home Décor, PropTech

Timeline

Phased development (MVP to full platform)

Interior Vision

Professional interior designers and home decorators needed a centralized workspace that could handle the entire design-to-purchase workflow, from saving inspiration to identifying products to managing budgets.

The interior design industry has long struggled with fragmented tools: Pinterest for inspiration, spreadsheets for budgeting, screenshots for product tracking, and manual searches to find where items can be purchased. Designers waste hours reverse-searching images, tracking down products across dozens of retailer sites, and manually updating specification sheets when products go out of stock. Interior Vision saw an opportunity to consolidate this workflow into a single intelligent platform powered by modern AI.

Solving Multiple Pain Points

The Product Discovery Gap

Designers see furniture and décor they love in photos but have no efficient way to identify and purchase those exact items, or close matches. Manual Google searches and reverse image lookups are time-consuming and often inaccurate.

Workflow Fragmentation

The typical designer workflow involves juggling 5+ tools: Pinterest for inspiration, Excel for budgeting, email for client communication, web browsers for product searches, and manual screenshots for documentation. Nothing talks to each other.

Budget & Specification Management

Tracking product costs, availability, dimensions, and vendors across multiple rooms and projects becomes a spreadsheet nightmare. When items go out of stock or prices change, designers have no way to know without manually checking every link.

Style Consistency

Maintaining design coherence across a mood board is subjective and time-consuming. Designers need objective analysis of color palettes, style trends, and aesthetic cohesion.

Browser-Native Workflows

Designers spend hours browsing e-commerce sites. Saving products required screenshotting, copying URLs, and manually entering data, breaking their browsing flow.

User-First, AI-Native Development

Discovery & Research

  • Interviewed interior designers to understand pain points in their daily workflows
  • Mapped the end-to-end journey from inspiration gathering to final specification delivery
  • Identified AI as the key differentiator for visual product recognition

Architecture Design

  • Selected Next.js 15 for a modern, performant full-stack framework
  • Chose Supabase for real-time collaboration, authentication, and PostgreSQL database
  • Integrated Anthropic's Claude Vision API for state-of-the-art furniture object detection
  • Used SerpAPI (Google Shopping) for intelligent product matching via reverse image search
  • Built with Tailwind CSS and Radix UI for accessible, polished interfaces

Core Technical Decisions

1. AI-Powered Visual Recognition

Instead of simple image classification, we implemented Claude Vision to detect individual furniture objects within interior photos, generate bounding boxes, and provide contextual descriptions. This enables pinpoint-accurate product searches.

2. Browser Extension as First-Class Feature

We built a Chrome extension (Manifest V3) that allows designers to save products and images from any website with one click, integrating directly into their natural browsing workflow.

3. Real-Time Collaboration

Using Supabase Realtime subscriptions, we enabled instant updates across users, browser tabs, and devices, critical for design teams working on shared projects.

4. Resilient Product Data Pipeline

Built a containerized product enrichment pipeline that normalizes inconsistent retailer page structures into a unified product schema, handling diverse e-commerce architectures with appropriate rate limiting and respect for site policies.

5. Intelligent Caching Strategy

Object detection results: cached permanently (images don't change). Product search results: cached 1-24 hours (prices fluctuate). Style assessments: cached and regenerable on demand.

Iterative Development Phases

  1. 1.Authentication and project management
  2. 2.Mood board creation and image uploads
  3. 3.AI object detection and visual search
  4. 4.Product specification sheets with budget tracking
  5. 5.Browser extension for seamless product/image saving
  6. 6.Pinterest integration for inspiration imports (via official Pinterest API with user authorization)
  7. 7.AI-powered style assessment and reporting
  8. 8.Visual planning with floor plan layouts

Key Features Delivered

AI-Powered Visual Product Search

  • Designers upload interior design photos
  • Claude Vision analyzes images and detects furniture/décor objects with bounding boxes
  • Users select which objects to search for
  • System auto-crops objects and runs reverse image searches via Google Shopping
  • Results show matching products with prices, retailers, and availability
  • One-click save to spec sheets or product cache

Smart Specification Sheets

  • Track products across multiple rooms per project
  • Editable table with inline editing for all product details
  • Real-time budget tracking with totals and variance
  • Bulk operations (assign rooms, delete multiple items)
  • Automatic availability refresh with scheduling
  • AI-generated product descriptions

Collaborative Mood Boards

  • Drag-and-drop image uploads
  • Organize by room or view all images flat
  • Pinterest integration via official API with user authorization for inspiration imports
  • Source URL tracking and metadata preservation
  • Real-time sync across team members

AI Style Assessment & Reporting

  • Analyze mood board images for design style identification
  • Extract color palettes with dominant and accent colors
  • Generate professional design narratives and recommendations
  • Confidence scores for style classifications
  • Batch processing for multiple images

Browser Extension (Chrome)

  • One-click product saving: Right-click any product on any website → auto-screenshot, OCR, scrape details, save to Interior Vision
  • Image saving to mood boards: Right-click images → save directly to active project
  • Pinterest integration: Authenticated board imports with room assignments via Pinterest API
  • Visual search queue: Queue images for later AI processing
  • Context menu integration throughout

Visual Planning & Floor Plans

  • Template-based floor plan creation
  • Drag-and-drop item placement on layouts
  • Wall and space measurement tools
  • Interactive canvas for room planning

Budget & Project Management

  • Set project-level budgets
  • Track spending by room and category
  • Real-time cost totals with shipping, installation, and tax
  • Remaining budget calculations
  • Multi-room organization

Technical Highlights

Frontend

Next.js 15, React 19, Tailwind CSS 4.1, Radix UI

Backend

Next.js API Routes, Supabase (PostgreSQL + Storage + Realtime)

AI & Machine Learning

Claude Vision (Anthropic), Google Generative AI, Replicate

APIs & Integration

SerpAPI (Google Shopping), Pinterest API

DevOps

Docker (for Camoufox scraping), TypeScript, ESLint

Production-Ready Success

Semantic Visual Understanding: Leveraged Claude's multimodal reasoning to identify and contextualize furniture in design photos (e.g., "mid-century walnut sideboard" rather than just "furniture"), trading raw detection speed for richer descriptive output that drives more accurate downstream product matching

Unified Workflow: Designers move from inspiration → detection → shopping → specification in a single platform, replacing the previous patchwork of Pinterest, spreadsheets, screenshots, and manual searches

Browser-Native Adoption: The Chrome extension keeps designers in their natural browsing flow, eliminating the screenshot-and-paste cycle that previously broke their workflow

Real-Time Collaboration: Teams work simultaneously on shared projects with instant cross-device sync via Supabase Realtime

Scalable, Modern Stack: Built on Next.js 15 and Supabase for performance, real-time capabilities, and straightforward horizontal scaling

Ready to Transform Your Industry?

If you're building a product where AI is core to the experience, we'd love to help. From multi-agent orchestration to real-time collaborative platforms, we deliver polished, production-ready software.