Weave Code
Ai Manticore Search Store Laravel Package
symfony/ai-manticore-search-store
ManticoreSearch Store integrates ManticoreSearch as a vector store for Symfony AI Store, enabling KNN/vector similarity search backed by Manticore’s engine. Includes links to Manticore KNN docs plus Symfony AI contribution and issue resources.
ManticoreSearch vector store bridge for Symfony AI
Frequently asked questions about Ai Manticore Search Store
- Can I use symfony/ai-manticore-search-store directly in Laravel without Symfony AI?
- No, this package requires Symfony AI’s StoreInterface. For Laravel, you’d need to wrap it in a facade or service provider to expose the vector store methods (e.g., `findNearest()`). Alternatively, use it as a reference to build a Laravel-native ManticoreSearch client.
- What Laravel versions support this package?
- The package targets Symfony 6.4+, which aligns with PHP 8.1+. Laravel 10+ (PHP 8.1+) should work, but test thoroughly. No Laravel-specific dependencies exist—compatibility hinges on Symfony AI’s Laravel interop.
- How do I set up ManticoreSearch for Laravel?
- First, install the package: `composer require symfony/ai manticoresearch/manticoresearch symfony/ai-manticore-search-store`. Then deploy ManticoreSearch (Docker recommended for testing) and configure the Symfony AI Store with your connection details (host, port, index). Bind the store to Laravel’s container via `AppServiceProvider`.
- Is ManticoreSearch faster than Pinecone or Weaviate for Laravel?
- Yes, ManticoreSearch offers sub-100ms latency for 1M vectors out-of-the-box, often outperforming managed services at scale. However, it lacks auto-scaling—ideal for self-hosted setups where you control infrastructure. Benchmark with your dataset before committing.
- Can I mix keyword and vector search (hybrid search) in Laravel?
- Yes, if your ManticoreSearch schema includes both text and vector fields. The package leverages Manticore’s native hybrid search capabilities. Configure your index to support `MATCH()` (keyword) + `KNN()` (vector) queries in a single request.
- What’s the migration path from Elasticsearch or pgvector to ManticoreSearch?
- You’ll need custom ETL scripts to transform embeddings into ManticoreSearch’s format (e.g., `ADD INDEX knn(index_name, 'vector_field', 'L2')`). Use Manticore’s `INSERT INTO ... VALUES` syntax. Test with a subset of data first to validate schema compatibility.
- Are there Laravel-specific alternatives to this package?
- For Laravel-native solutions, consider `laravel-ai/vector` (experimental) or build a custom client using `manticoresearch/manticoresearch`. This package’s advantage is its integration with Symfony AI’s abstractions, which may simplify RAG pipelines if you’re already using Symfony components.
- How do I handle ManticoreSearch updates in production?
- Monitor the [ManticoreSearch changelog](https://github.com/manticoresoftware/manticore/releases) and Symfony AI’s roadmap. Test updates in staging first—schema changes (e.g., index alterations) may require downtime. Use Docker for consistent local testing.
- What’s the maximum scale this setup supports in Laravel?
- ManticoreSearch handles ~10M vectors per node efficiently. For larger datasets, shard your index across multiple Manticore instances (manual configuration). Latency remains low (<50ms) if hardware is adequate (SSD-backed storage, sufficient RAM).
- How do I test vector search in Laravel before production?
- Use Docker to spin up ManticoreSearch locally (`docker-compose.yml` as in the README). Mock the Symfony AI Store in Laravel tests with a lightweight in-memory store, then swap to ManticoreSearch in production. Validate queries with known vectors (e.g., `findNearest()`) and measure response times.
Weaver
How can I help you explore Laravel packages today?