Meilisearch Hybrid Search with OpenAI Embeddings — Semantic AI Search for Electronic First

Description

Implementing AI-Powered Hybrid Search on Electronic First (Digital Gaming Marketplace) — Meilisearch v1.35 + OpenAI `text-embedding-3-small` Hybrid search blends traditional keyword matching with semantic understanding, so users can search naturally — like “shopping voucher UK” or “gaming subscription” — and still get highly relevant results even when product titles don’t contain the exact same words. --- What’s covered in this video: - Setting up Meilisearch hybrid search with OpenAI embeddings - Artisan setup command (php artisan meilisearch:hybrid-setup) - Admin panel toggle to enable/disable AI search instantly - Side-by-side comparison: keyword-only vs hybrid search results - Frontend integration with InstantSearch.js - Laravel Scout backend integration Tech stack: - Laravel 9 + Laravel Scout - Meilisearch v1.35 (self-hosted) - OpenAI text-embedding-3-small (1536 dimensions) - InstantSearch.js + @meilisearch/instant-meilisearch v0.30 - Admin toggle via database settings (no redeployment needed) How it works: 1. Each product is converted into a 1536-number vector via OpenAI 2. User search queries are also converted into vectors 3. Meilisearch combines keyword matching (exact words) + semantic matching (meaning) using a configurable ratio 4. semanticRatio: 0.5 = 50% keyword + 50% AI meaning Links: - Meilisearch: https://www.meilisearch.com - Meilisearch Hybrid Search Docs: https://www.meilisearch.com/docs/guides/ai/hybrid_search - OpenAI Embeddings: https://platform.openai.com/docs/guides/embeddings - Laravel Scout: https://laravel.com/docs/scout - Electronic First: https://www.electronicfirst.com #Meilisearch #HybridSearch #AI #OpenAI #Embeddings #Laravel #SemanticSearch #Ecommerce #SearchEngine #LaravelScout