Development

Laravel Boost: Our First Impressions of Laravel's New AI Coding Assistant.

8 minutes

Introduction

I've been using AI coding assistants for a few months now, and like many developers, I've experienced that familiar mix of excitement and frustration. One minute you're amazed at what Junie suggests, the next you're pulling your hair out because it's confidently using a deprecated Laravel method from version 5.8 in your Laravel 11 project.

Well, the Laravel team has been paying attention to this exact problem in particular Ashley Hindle, recently they released Laravel Boost - their answer to making AI assistants actually understand Laravel instead of just hallucinating their way through your codebase.

After spending some time with Boost since its public beta release, I wanted to share our first impressions and show you exactly what this "AI coding starter kit" can do.

What Exactly Is Laravel Boost?

Laravel Boost is essentially a bridge between your AI coding assistant and your Laravel application. Think of it as giving your AI assistant x-ray vision into your project - suddenly it knows your exact Laravel version, your database schema, which packages you're using, and can even run Tinker commands.

The package was announced at Laracon US 2025 and is built around three core pillars:

1. Laravel-Specific MCP Server (15+ Tools)

The Model Context Protocol (MCP) server gives AI agents direct access to your application through specialized tools. Instead of guessing, your AI can actually inspect your database schema, read your routes, check your logs, and even run Tinker commands.

2. Version-Specific Documentation API

This is where things get interesting. Laravel Boost includes access to over 17,000 pieces of vectorized Laravel ecosystem documentation, specific to the exact versions of packages you have installed. No more getting Inertia v1 examples when you're running v2.

3. Laravel-Maintained AI Guidelines

The Laravel team has curated specific guidelines for popular IDEs and AI tools (Cursor rules, Junie guidelines, GitHub Co-Pilot instructions, Claude.md files) that teach your AI to follow Laravel conventions and best practices.

Getting Started - Installation Walkthrough

Installing Boost couldn't be simpler - it took me about 2 minutes from start to finish:

shell 
1composer require laravel/boost --dev
2php artisan boost:install

The installer is surprisingly intelligent. It automatically detected that I was using Junie and asked which features I wanted to enable:

  • Boost MCP server ✓

  • Package AI guidelines ✓

  • Laravel Style AI guidelines ✓

  • Herd MCP server ✓

What I appreciated most was that it didn't force any opinionated style rules on my existing project - it respected my current setup and only added what I opted into.

First Impressions - The Good

It Actually Knows Your Project

The first thing I tested was asking my AI assistant: "What routes does this application have?" Instead of giving me generic Laravel routing documentation, it used the list-routes tool to show me the actual routes in my current project. That moment when you realize your AI assistant can see inside your app? Pretty magical.

Database Schema Awareness

This one solved a real pain point for me. Before Boost, when I asked for help with an Eloquent query, the AI would have to guess at my table structure. Now it can inspect my actual schema:

shell 
1# The AI can now run commands like this behind the scenes:
2php artisan tinker --execute="Schema::connection()->getColumnListing('users')"

Version-Specific Accuracy

I tested this by asking for help implementing deferred props in Inertia. Without Boost, my AI assistant gave me a generic solution that was "close enough." With Boost's documentation search, it found the exact Inertia v2 implementation that matched my installed version.

Automatic Test Generation

Here's where the Laravel-maintained guidelines really shine. When I asked the AI to implement rate limiting, not only did it provide the correct implementation for my Laravel version, but it also automatically generated feature tests. This is the kind of best practice nudging that makes Boost valuable.

The Reality Check - Areas for Improvement

Beta Rough Edges

Boost is currently in public beta, and it shows. I encountered a few hiccups where tools didn't respond as expected, and the documentation mentions that you should treat all generated code as draft (which you should anyway, but still).

Major Version Limitations

The documentation ingestion currently works at the major version level only. If your project depends on minor or patch version differences, you'll need to double-check the generated code more carefully.

Learning Curve for AI Prompting

I had to adjust how I interact with my AI assistant. Instead of asking "How do I create a factory in Laravel?", I now ask "Create a factory for my User model" - talking to my application through the AI rather than asking for generic help.

Real-World Example - Debugging a White Screen

The most impressive demonstration came when I intentionally introduced a bug and asked the AI to help fix it. Here's what happened:

  1. The Problem: White screen of death after a controller change

  2. My Prompt: "Help me debug this white screen issue"

  3. Boost's Response: The AI used the last-errors tool to fetch the actual Laravel error, then read the browser logs using the browser-logs tool, identified the typo, and proposed a fix.

This kind of contextual debugging is exactly what makes Boost valuable - it's not just generating code, it's actively helping you solve real problems in your actual application.

Supported IDEs and AI Tools

Boost works with all the major players:

  • Cursor

  • Claude Code

  • GitHub Copilot

  • PHPStorm (Junie)

  • VS Code

The setup process automatically detects which tools you're already using and configures them appropriately.

Tips for Getting the Most Out of Boost

1. Adjust Your Prompting Style

Instead of asking for generic Laravel help, ask about your specific application:

  • ❌ "How do I create a Laravel route?"

  • ✅ "Add a route to handle user profile updates in this application"

2. Use Application-Specific Queries

Take advantage of the tools that inspect your actual project:

  • "What's my current database schema for the users table?"

  • "Show me all the routes that handle user authentication"

  • "What Eloquent models do I have in this project?"

3. Let the AI Use Tinker

Some of the most impressive results came when I let the AI use Tinker to explore my application or create test data. Don't be afraid to ask it to "use Tinker to create some sample data."

The Bigger Picture - Where This Is Heading

Laravel Boost feels like the first step in a much larger shift. The Laravel team is clearly betting that the future of development involves AI assistance, but done thoughtfully with proper context and guardrails.

What excites me most is that this isn't trying to replace developers - it's making us more effective by reducing the busywork and providing intelligent suggestions based on our actual codebases.

Should You Install It?

If you're already using AI coding assistants with Laravel, I'd say absolutely. Boost is free, open source, and even in beta, it's providing real value. The installation is non-invasive, and you can opt into just the features you want.

The key thing to remember is that Boost isn't magic - it's a tool that makes your existing AI assistant smarter about Laravel. You still need to review the generated code, run your tests, and use your engineering judgment.

Our Experience So Far

After using Boost for a few weeks, I've found it saves me time on routine tasks and provides much more accurate suggestions. The ability to ask contextual questions about my actual application rather than getting generic Stack Overflow-style answers is genuinely useful.

The automated test generation alone has been worth the installation time - having the AI suggest appropriate tests for the code it generates feels like a significant step forward in code quality.

What's Next?

The Laravel team has made it clear that Boost is just the beginning. They're actively collecting feedback (there's even a built-in feedback tool), and version 1.0 is coming soon.

I'm particularly excited to see how this evolves as the Laravel ecosystem continues to embrace AI tooling. If this is the foundation, the future looks pretty promising.

Final Thoughts

Laravel Boost represents exactly the kind of thoughtful AI integration I want to see in development tools. Instead of trying to be everything to everyone, it solves a specific problem really well - giving AI assistants the context they need to be genuinely helpful with Laravel projects.

Is it perfect? No. Is it useful? Absolutely. And most importantly, it's free and gets better with community feedback.

Have you tried Laravel Boost yet? What's been your experience with AI coding assistants in Laravel projects? I'd love to hear about your successes (and frustrations) - drop us a line and let us know how you're finding it.

Want to get started with Laravel Boost? Check out the official repository on GitHub or read the announcement blog post for all the technical details.

Key Resources