Weave Code
Ai Meilisearch Message Store Laravel Package
symfony/ai-meilisearch-message-store
Meilisearch Message Store integrates Meilisearch as a persistent message store for Symfony AI Chat. Index and query chat messages using Meilisearch APIs, with support for async task handling and configurable index settings via Meilisearch.
Meilisearch message store bridge for Symfony AI Chat
Frequently asked questions about Ai Meilisearch Message Store
- How do I integrate this Meilisearch message store into a Laravel app using Symfony AI Chat?
- Install via Composer (`composer require symfony/ai-meilisearch-message-store`), then register the `MeilisearchMessageStore` as a service in your `AppServiceProvider`. Bind it to Symfony AI’s `MessageStoreInterface` using Laravel’s container. Replace Symfony’s `HttpClient` with Laravel’s `Http` facade or Guzzle via a custom adapter if needed.
- What Laravel versions does this package support, and are there Symfony dependency conflicts?
- The package targets Laravel 10+ and 11, but Symfony dependencies (e.g., `symfony/ai`, `symfony/http-client`) may conflict. Use a Laravel-compatible service provider or the `spatie/laravel-symfony` bridge to resolve DI mismatches. Test with `composer check-platform-reqs` to catch version issues early.
- Can I use this for semantic search (e.g., embedding similarity) in Laravel AI?
- Meilisearch lacks native Approximate Nearest Neighbors (ANN), but you can simulate semantic search using `rankingRules` or `filter` queries with custom scoring. For production-grade vector search, pair it with a dedicated vector DB (e.g., Weaviate) via Laravel’s queue system for hybrid workflows.
- How do I handle async Meilisearch operations (e.g., index updates) in Laravel?
- Offload async tasks to Laravel queues by wrapping Meilisearch’s async operations in a `ShouldQueue` job. Use `task:wait` to poll for completion or leverage Symfony’s async task API. Example: Dispatch a `MeilisearchIndexUpdateJob` with `dispatchSync()` for critical paths.
- What’s the best way to manage multi-tenancy with separate Meilisearch indexes?
- Prefix indexes by tenant ID (e.g., `tenant_123_messages`) and validate access via Laravel middleware. Use Meilisearch’s [security rules](https://www.meilisearch.com/docs/reference/api/security) to restrict document access. For dynamic tenant routing, bind the index name to the authenticated user in your `AppServiceProvider`.
- How does this compare to Pinecone or Weaviate for Laravel AI chat history?
- Meilisearch is cost-effective for <10M monthly operations (open-core pricing) and integrates natively with Symfony AI. Pinecone/Weaviate offer better ANN support but require additional Laravel adapters. Benchmark latency (Meilisearch: ~50–150ms) vs. your RAG pipeline needs before choosing.
- Do I need to write Laravel migrations for Meilisearch schemas?
- No, Meilisearch uses dynamic schemas, but enforce consistency by syncing local DB and Meilisearch schemas via a custom `MeilisearchSchemaSync` Artisan command. Add Laravel migrations for critical fields (e.g., `updated_at`) and use Meilisearch’s [index versioning](https://www.meilisearch.com/docs/learn/core_concepts/indexes#versioning) for backward compatibility.
- How can I cache frequently accessed chat messages to reduce Meilisearch latency?
- Wrap the `MeilisearchMessageStore` in a `MessageCacheDecorator` using Laravel’s cache system (e.g., Redis). Cache responses for `find()` queries with a short TTL (e.g., 5 minutes) and invalidate on updates. Example: `cache()->remember('chat_'.$id, now()->addMinutes(5), fn() => $store->find($id));`.
- What’s the performance impact of using Meilisearch in production vs. a local DB?
- Meilisearch achieves sub-100ms retrieval for chat messages with in-memory indexes, but network hops (e.g., cloud instances) add ~50–100ms. For high throughput, batch async operations (e.g., bulk inserts) and use Laravel queues to parallelize writes. Test with your expected load (e.g., 10K+ messages/sec).
- How do I back up and restore Meilisearch indexes in Laravel?
- Use Meilisearch’s [backup API](https://www.meilisearch.com/docs/reference/api/backup) to export indexes, then restore via the [import API](https://www.meilisearch.com/docs/reference/api/backup#import). Schedule backups with Laravel’s Artisan scheduler (e.g., `schedule->command(MeilisearchBackupCommand::class)->daily()`). For disaster recovery, pair with a local DB backup of metadata.
Weaver
How can I help you explore Laravel packages today?