Artificial Intelligence applications are evolving rapidly, but traditional chatbots still struggle with one major problem:

They cannot access your private or domain-specific data.

This is where Retrieval-Augmented Generation (RAG) becomes extremely powerful.

A RAG system allows an AI model to search your own knowledge base before generating a response. Instead of relying only on pretrained knowledge, the AI can answer questions using your documents, PDFs, articles, database records, or company knowledge.

In this tutorial series, you will learn how to build a complete RAG pipeline in Laravel using:

Laravel 12
PostgreSQL + pgvector
Open source embedding models
Ollama
Llama 3
Vector similarity search
AI-powered chat API

By the end of this series, you will have a production-ready AI assistant capable of answering questions using your own data.

What is RAG?

RAG stands for Retrieval-Augmented Generation.

The workflow looks like this:

User sends a question
System converts the question into embeddings
Vector database searches for similar documents
Relevant context is injected into the prompt
AI model generates the final answer

Without RAG:

text
Question → AI Model → Answer

With RAG:

text
Question → Embedding → Vector Search → Context → AI Model → Answer

This dramatically improves:

Accuracy
Freshness of information
Domain-specific responses
Hallucination reduction

RAG Architecture Overview

Our Laravel RAG system architecture:

text
Frontend Chat UI
       ↓
Laravel API
       ↓
Embedding Service
       ↓
PostgreSQL + pgvector
       ↓
Context Retrieval
       ↓
Ollama + Llama 3
       ↓
AI Response

Why Use Laravel for AI Applications?

Laravel is surprisingly excellent for AI applications because it already provides:

Queue system
API development
Database ORM
Authentication
Event system
Broadcasting
Background jobs
File storage
Scheduling

This makes Laravel ideal for:

AI chat applications
Knowledge base systems
AI agents
Document indexing
Semantic search
AI SaaS products

Choosing the Vector Database

There are several vector database options:

Database	Pros	Cons
Pinecone	Managed	Paid
Weaviate	Powerful	More complex
Qdrant	Fast	Separate infrastructure
Elasticsearch	Mature	Heavy
PostgreSQL pgvector	Simple + powerful	Less specialized

For Laravel developers, PostgreSQL + pgvector is one of the best choices because:

Easy integration
SQL support
Open source
Cheap infrastructure
Great performance
Familiar ecosystem

Installing PostgreSQL with pgvector

Install PostgreSQL first.

Ubuntu:

bash
sudo apt update
sudo apt install postgresql postgresql-contrib

Install pgvector:

bash
sudo apt install postgresql-16-pgvector

Enable extension:

sql
CREATE EXTENSION vector;

Verify installation:

sql
SELECT * FROM pg_extension;

Creating Laravel Project

Create new Laravel project:

bash
composer create-project laravel/laravel laravel-rag

Enter project:

bash
cd laravel-rag

Configure Database

Update .env:

env
DB_CONNECTION=pgsql
DB_HOST=127.0.0.1
DB_PORT=5432
DB_DATABASE=laravel_rag
DB_USERNAME=postgres
DB_PASSWORD=secret

Creating the Documents Table

Create migration:

bash
php artisan make:migration create_documents_table

Migration:

php
<?php
use Illuminate\Database\Migrations\Migration;
use Illuminate\Database\Schema\Blueprint;
use Illuminate\Support\Facades\DB;
use Illuminate\Support\Facades\Schema;

return new class extends Migration
{
    public function up(): void
    {
        Schema::create('documents', function (Blueprint $table) {
            $table->id();
            $table->string('title');
            $table->text('content');
            $table->timestamps();
        });

        DB::statement('ALTER TABLE documents ADD COLUMN embedding vector(768)');
    }

    public function down(): void
    {
        Schema::dropIfExists('documents');
    }
};

Why Embeddings Matter

Embeddings are numerical vector representations of text.

Example:

text
"Laravel is a PHP framework"
↓
[0.213, -0.881, 0.115, ...]

Similar texts produce similar vectors.

This allows semantic search instead of keyword search.

Example:

Query: "How to authenticate users?", "AI chatbot"
Semantic Match: "Laravel authentication tutorial", "Conversational assistant"

Installing Ollama

Ollama allows running open source AI models locally.

Install Ollama:

bash
curl -fsSL https://ollama.com/install.sh | sh

Run model:

bash
ollama run llama3

Pull embedding model:

bash
ollama pull nomic-embed-text

Creating Embedding Service in Laravel

Create service:

bash
php artisan make:service EmbeddingService

Create file manually if needed.

php
<?php
namespace App\Services;

use Illuminate\Support\Facades\Http;

class EmbeddingService
{
    public function embed(string $text): array
    {
        $response = Http::post('http://localhost:11434/api/embeddings', [
            'model' => 'nomic-embed-text',
            'prompt' => $text,
        ]);

        return $response->json()['embedding'];
    }
}

Storing Embeddings

Create model:

bash
php artisan make:model Document

Insert document with embedding:

php
<?php
use App\Models\Document;
use App\Services\EmbeddingService;

$embedding = app(EmbeddingService::class)
    ->embed($content);

Document::create([
    'title' => $title,
    'content' => $content,
    'embedding' => '[' . implode(',', $embedding) . ']',
]);

Next Part

In the next article, we will build:

Vector similarity search
Semantic retrieval
AI chat endpoint
Prompt engineering
Context injection
Streaming responses

Building a RAG System in Laravel with Vector Search and Open Source AI Models