Building AI-Driven Features in Symfony

Why AI in Symfony?

AI is no longer a futuristic concept — it's part of today's tech stack. From chatbots to content enrichment and semantic search, AI-driven features are everywhere. Thanks to the php-llm/llm-chain library, integrating these capabilities into your Symfony project has never been easier.

Introduction

php-llm/llm-chain is a PHP-native library that allows you to interact with Large Language Models (LLMs) like:

• OpenAI's GPT

• Claude by Anthropic

• LLaMA by Meta

• Gemini by Google

• DeepSeek R1

It supports a wide variety of platforms (OpenAI, Azure, Replicate, etc.) and allows you to:

• Generate content.

• Call external tools (functions) with LLMs.

• Embed and semantically search documents.

• Chain multiple AI calls together with logic.

Symfony integration is provided via php-llm/llm-chain-bundle, which gives you automatic service registration, DI support, and config-driven setup.

Installing the Symfony Bundle

Install the package via Composer:

composer require php-llm/llm-chain-bundle

Configure your .env:

OPENAI_API_KEY=your-api-key-here

Configure the service in config/packages/llm_chain.yaml:

llm_chain:
  platform:
    openai:
      api_key: '%env(OPENAI_API_KEY)%'

  chain:
    default:
      model:
        name: 'gpt4o-mini'

Using AI in Your Symfony Service

Here’s a simple example of how to create a Symfony service that sends a message to an LLM and gets a response.

use PhpLlm\LlmChain\ChainInterface;
use PhpLlm\LlmChain\Model\Message\Message;
use PhpLlm\LlmChain\Model\Message\MessageBag;
use PhpLlm\LlmChain\Model\Response\ResponseInterface;

final class SmartAssistant
{
    public function __construct(
        private ChainInterface $chain
    ) {
    }

    public function ask(string $question): ResponseInterface
    {
        $messages = new MessageBag(
            Message::forSystem('You are a helpful assistant.'),
            Message::ofUser($question),
        );

        return $this->chain->call($messages);
    }
}

You can now use this service in any controller, console command, or background worker.

Tool Calling: Make the AI Interactive

Want your LLM to call real PHP functions? Annotate them by using #[AsTool] attribute:

use PhpLlm\LlmChain\Toolbox\Attribute\AsTool;

#[AsTool('current_time', 'Returns the current server time')]
final class ClockTool
{
    public function __invoke(): string
    {
        return (new \DateTimeImmutable())->format('Y-m-d H:i:s');
    }
}

The LLM can now decide on its own when to use this function during a conversation. Think of it like ChatGPT Plugins… but in PHP.

Embeddings & Search

llm-chain also supports embeddings for semantic search. You can store vectors in providers like:

• ChromaDB

• Pinecone

• MongoDB Atlas Search

• Azure AI Search

This is great for implementing Retrieval-Augmented Generation (RAG) — a technique where you fetch contextually relevant documents before asking the LLM a question.

Try It: Symfony Demo Project

Want to test this out? The php-llm/llm-chain team provides a demo Symfony application showing chatbot interaction and vector search:

➡ php-llm/llm-chain-symfony-demo

Controlling Costs and Tokens

LLMs aren’t free, so stay efficient with:

• Cache repeating responses

• Using short prompts

• Monitoring token usage via logs

• Limiting your system prompts

Conclusion

With just a few lines of configuration and code, you can integrate powerful AI features into your Symfony app. Whether you want to automate tasks, answer questions, or enrich content — llm-chain is a solid tool to get started.

Symfony is ready for the AI age. Are you?