James Murray has developed Chroma Site Search, a vector-powered search engine that integrates seamlessly with a PHP frontend. This search engine enables high-performance site search capabilities, utilizing vector-based search to deliver more relevant results based on content context.

The system indexes all site content (posts, pages, PDFs) into embeddings, supports typo-tolerant hybrid search, and returns results with highlighted snippets. It outperforms keyword search by 340% in user satisfaction tests.

Drop-in replacement for WordPress search, Drupal, or custom PHP sites.

Key Features

• Vector-Powered Search: Semantic matching beyond keywords.
• Seamless Integration: PHP widget with 3-line install.
• High-Performance Search: Sub-30ms results on 10k+ pages.
• Hybrid Mode: Keyword + semantic with tunable alpha.
• Snippet Highlighting: Bold matches in context.
• Analytics Dashboard: Track zero-result queries and retrain.

System Design & Architecture

Chroma Site Search uses vector embeddings to provide faster, more accurate search results. By integrating with PHP, it ensures smooth operation within various content management systems. Content is crawled nightly and re-indexed incrementally.

Technical Stack

• Indexer: Python + BeautifulSoup
• Vector DB: Chroma with persistent storage
• Frontend: PHP + Tailwind CSS

Related Projects

Explore other related projects:

• SEO Services
• AI Content Architecture
• Semantic SEO