Introduction

AI Agentic Framework with built-in observability for PHP

Last updated 12 days ago

Introduction

AI Agentic Framework with built-in observability for PHP

Neuron AI is a framework for integrating AI Agents into your existing PHP applications.

It simplify every stage of the AI application development life cycle:

Development: You can build AI driven features into your application thanks to the Neuron core components or third party extensions.
Integration: Neuron is framework agnostic so you are free to create LLM interactions as stand alone applications, or integrate them as part of an existing architecture.
Observability: Turn your Neuron agents into production-ready applications, monitoring their performance and accuracy with built-in observability and AI Bug Fix provided by

Neuron uses a standard interface for large language models (AIProviderInterface) that can be easily implemented to support more LLMs, as well as for the other components, such as embedding, vector stores, and data loaders.

use NeuronAI\Agent;
use NeuronAI\Chat\Messages\UserMessage;
use NeuronAI\Providers\AIProviderInterface;
use NeuronAI\Providers\Anthropic\Anthropic;

class MyAgent extends Agent
{
    protected function provider(): AIProviderInterface
    {
        return new Anthropic(
            key: 'ANTHROPIC_API_KEY',
            model: 'ANTHROPIC_MODEL',
        );
    }
}

echo MyAgent::make()->chat(new UserMessage("Hi!"));
// Hi, how can I help you today?

use NeuronAI\Agent;
use NeuronAI\Chat\Messages\UserMessage;
use NeuronAI\Providers\AIProviderInterface;
use NeuronAI\Providers\Ollama\Ollama;

class MyAgent extends Agent
{
    protected function provider(): AIProviderInterface
    {
        return new Ollama(
            url: 'OLLAMA_URL',
            model: 'OLLAMA_MODEL',
        );
    }
}

echo MyAgent::make()->chat(new UserMessage("Hi!"));
// Hi, how can I help you today?

use NeuronAI\Agent;
use NeuronAI\Chat\Messages\UserMessage;
use NeuronAI\Providers\AIProviderInterface;
use NeuronAI\Providers\OpenAI\OpenAI;

class MyAgent extends Agent
{
    protected function provider(): AIProviderInterface
    {
        return new OpenAI(
            key: 'OPENAI_API_KEY',
            model: 'OPENAI_MODEL',
        );
    }
}

echo MyAgent::make()->chat(new UserMessage("Hi!"));
// Hi, how can I help you today?

use NeuronAI\Agent;
use NeuronAI\Chat\Messages\UserMessage;
use NeuronAI\Providers\AIProviderInterface;
use NeuronAI\Providers\Mistral;

class MyAgent extends Agent
{
    protected function provider(): AIProviderInterface
    {
        return new Mistral(
            key: 'MISTRAL_API_KEY',
            model: 'MISTRAL_MODEL',
        );
    }
}

echo MyAgent::make()->chat(new UserMessage("Hi!"));
// Hi, how can I help you today?

use NeuronAI\Agent;
use NeuronAI\Chat\Messages\UserMessage;
use NeuronAI\Providers\AIProviderInterface;
use NeuronAI\Providers\Deepseek;

class MyAgent extends Agent
{
    protected function provider(): AIProviderInterface
    {
        return new Deepseek(
            key: 'MISTRAL_API_KEY',
            model: 'MISTRAL_MODEL',
        );
    }
}

echo MyAgent::make()->chat(new UserMessage("Hi!"));
// Hi, how can I help you today?

With Neuron you will have a standard toolkit to implement AI driven applications drastically reducing vendor lock-in. You can switch between LLMs, vector stores, embedding providers, etc. with just a few lines of code without the need to refactor big portions of your application.

What is an AI Agent

AI agents are essentially programs that can perform tasks in response to a prompt.

Compared to a raw LLM (that primarily provide information and respond to questions within a conversation), AI agents can take independent actions to complete tasks.

While a simple LLM can answer your questions directly during a conversation, an AI agent might be able to:

Research information across multiple websites and compile it for you
Manage your email by responding to simple messages
Monitor data and alert you via email when something important happens

Why create your AI Agent

Agents can be connected to your application context and services to handle repetitive or complex processes that would otherwise require significant user effort or technical knowledge.

At Inspector we have created the "AI Bug Fix". It's an agent that automatically provides you with a bug fix proposal in real-time as an error occurs, and it's obviously built with Neuron.

It incorporates several tools and prompts, it analyzes your application data and errors, connects with your repository, etc. to achieve a very accurate result.

This is just an example, and it depends on your application domain.

Ecosystem

Register to the Neuron internal newsletter to get informative papers, articles, and best practices on how to start with AI development in PHP.

You will learn how to approach AI systems in the right way, understand the most important technical concepts behind LLMs, and how to start implementing your AI solutions into your PHP application with the Neuron AI framework.

We’re using Discussions as a place to connect with PHP developers working on Neuron to create their Agentic applications. We hope that you:

Ask questions you’re wondering about.
Share ideas.
Engage with other community members.
Welcome others and are open-minded.

Core components

Additional Resources

NextKey Concepts

Last updated 12 days ago