Dudils

The Problem

Architects assessing house plans for passive solar orientation had no centralised tool. Plans were scattered across dozens of builder websites, each with different formats, and there was no way to compare them side-by-side or filter by orientation suitability.

Dudils is a search engine for Australian house plans, purpose-built for passive solar assessment. It crawls 26+ builder websites, extracts floor plan images and specifications, and presents them in a unified interface where architects can filter by bedrooms, dimensions, and builder.

Key Product Decisions

The admin curation workflow was the most important design decision. Rather than automatically publishing every crawled plan, I built a review queue where the architect manually inspects each plan, sets the north direction, identifies the garage/road side, and rates the plan's passive solar potential. This human-in-the-loop approach ensures quality over quantity.

The crawler pipeline handles the messiest part — each builder site has a different tech stack, different HTML structure, and different ways of presenting floor plan data. I built a generic extractor that handles ~80% of sites, plus custom extractors for builders with unusual data formats (JSON attributes, inline SVGs, mirrored plan variants).

Technical Highlights

• •Crawled 1,600+ floor plans across 13 builders
• •Custom extractors for Clarendon (inline SVGs with double encoding), JG King (JSON data attributes), and others
• •Incremental crawling with polite rate limiting and robots.txt respect
• •Admin dashboard with draft review, approval scoring, and one-click promotion to production
• •Greyscale CSS filter on all floor plan images for visual consistency across builders

The site is live at dudils.com with 1,600+ plans ready for curation. The admin architect is actively reviewing and promoting plans, with the goal of becoming the definitive resource for passive-solar-assessed house plans in Australia.