Weave Code
Ai Azure Search Store Laravel Package
symfony/ai-azure-search-store
Azure AI Search vector store integration for Symfony AI Store. Index and query embeddings using Azure’s vector search capabilities, enabling semantic retrieval for RAG and AI apps. Links to official docs plus Symfony AI contribution and issue resources.
Azure AI Search vector store bridge for Symfony AI
Frequently asked questions about Ai Azure Search Store
- Can I use this package directly in Laravel without Symfony, or do I need to install symfony/ai?
- You’ll need to install `symfony/ai` (≥v0.8.0) as a dependency since this package implements Symfony AI’s `StoreInterface`. For Laravel, use the `symfony/http-client` bridge and bind the store in your service container via a custom provider. Laravel’s facades or service container will then manage the AzureSearchStore instance.
- What Laravel versions are officially supported, and are there any PHP version requirements?
- This package targets PHP 8.2+ and aligns with Laravel 10+. While not Laravel-specific, its Symfony AI dependency requires Laravel 10+ due to PHP 8.2+ constraints. Test thoroughly in your Laravel version, as Symfony AI’s evolving API may introduce compatibility gaps.
- How do I configure Azure AI Search credentials and endpoint in Laravel’s config?
- Add an `azure-search` array to your `config/services.php` with keys like `endpoint` (e.g., `https://your-service.search.windows.net`) and `key` (your Azure Search admin key). The store constructor expects these values, so inject them via Laravel’s service container binding or environment variables.
- Does this support hybrid search (vector + metadata filters) like Azure AI Search’s $filter?
- Yes. The package abstracts Azure Search’s hybrid capabilities, letting you combine vector similarity with metadata filters (e.g., `WHERE category = 'tech'`). This is exposed via Symfony AI’s `StoreInterface` methods like `findNearest()`, where you pass a `Filter` object for metadata constraints.
- What’s the performance impact of using Azure AI Search vs. a local vector store like FAISS?
- Azure AI Search typically adds 100–300ms latency per query due to network round trips, compared to FAISS’s sub-50ms local execution. Mitigate this with client-side caching (e.g., Redis) or batch queries. Benchmark with your workload—high-frequency searches may require reserved capacity or regional optimization.
- How do I handle Azure Search’s eventual consistency in Laravel’s synchronous workflows?
- Azure Search’s asynchronous writes may conflict with Laravel’s transactions. Use explicit retries for critical operations or implement a compensating transaction pattern. For example, wrap store operations in a `try-catch` and log failures to a queue for reprocessing.
- Are there Laravel-specific features like Eloquent model integration or query builder support?
- No. This package is Symfony-centric and lacks Eloquent integration or Laravel’s query builder. You’ll need to manually map Laravel models to embeddings or use a facade to abstract the store. For Eloquent-like patterns, consider a custom repository layer or a package like `spatie/laravel-activitylog` for audit trails.
- What are the cost implications of using Azure AI Search for high-volume queries?
- Azure AI Search charges per operation (e.g., ~$0.0004 per query in the US East region). For 1M queries, expect ~$400/month, compared to Pinecone’s $0.60/1M. Optimize costs with reserved capacity, batch operations, or client-side caching. Always validate pricing with Azure’s [pricing calculator](https://azure.microsoft.com/en-us/pricing/calculator/).
- How do I migrate from this package to another vector store (e.g., Pinecone or Weaviate) later?
- Design a facade or abstract class wrapping the `StoreInterface` to isolate Azure-specific logic. For example, inject `AzureSearchStore` via dependency injection and mock it during testing. Schema migrations may be needed if you rely on Azure-specific features like hybrid filters.
- Is there official Laravel documentation or a tutorial for integrating this with Azure OpenAI?
- No Laravel-specific docs exist, but you can combine this package with Azure OpenAI via Symfony AI’s `ClientInterface`. Example: Use `AzureSearchStore` to retrieve embeddings, then pass them to `AzureOpenAIClient` for completions. Follow Symfony AI’s [RAG example](https://symfony.com/doc/current/ai/rag.html) and adapt for Laravel’s service container.
Weaver
How can I help you explore Laravel packages today?