Installation

Requirements

• PHP 8.3+
• Laravel 12 or 13
• MySQL 8+ or PostgreSQL 14+
• An AI provider API key (OpenAI, Anthropic, or Gemini)
• Pinecone account (free tier works) if using Pinecone vector store

Step 1: Install the package

composer require laraigent/larai-kit

Step 2: Run the install command

php artisan larai:install

This does three things:

1. Publishes config/larai-kit.php
2. Runs database migrations (creates ai_documents, ai_assets, ai_chunks, ai_ingestions tables)
3. Creates the storage symlink for local file uploads

Step 3: Configure .env

Add your API keys:

# Required: AI provider
LARAI_AI_PROVIDER=openai
OPENAI_API_KEY=sk-your-key-here

# Recommended: Vector store for RAG
LARAI_VECTOR_STORE=pinecone
PINECONE_API_KEY=pcsk_your-key
PINECONE_INDEX_HOST=https://your-index-abc123.svc.pinecone.io

Step 4: Verify

php artisan larai:doctor

You should see green [OK] for Database, AI Provider, and Vector Store.

Optional: PDF and DOCX support

composer require smalot/pdfparser phpoffice/phpword

Pinecone Setup

1. Sign up at pinecone.io (free tier: 100K vectors)
2. Create an index:
   • Dimensions: 1536 (matches OpenAI text-embedding-3-small)
   • Metric: cosine
3. Copy the API key from your project settings
4. Copy the index host URL (looks like https://your-index-abc123.svc.aped-1234.pinecone.io)
5. Add both to .env

pgvector Setup (alternative)

If you prefer self-hosted vector search:

DB_CONNECTION=pgsql
LARAI_VECTOR_STORE=pgvector

Requires PostgreSQL with the pgvector extension installed. On Linux:

sudo apt install postgresql-17-pgvector

Next Steps

• Configuration - All available settings
• Agents & Tools - Build your first agent
• Ingestion Pipeline - Upload documents