This project is inspired by Open Deep Research, which uses LangGraph for implementation. Other implementations exist also for llamaindex, and others. Our version leverages Neuron to create a powerful, modular workflow for research and analysis.
Neuron Open Deep Research provides a structured approach to generating comprehensive research reports on any topic using large language models, with a focus on modularity, extensibility, and real-time results.
Neuron is an agentic framework that allows you to create full-featured AI Agents in PHP. It definitively fills the gap for AI Agents development between PHP and other ecosystems like Python or Javascript.
It provides you with 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.
If you are new to AI Agents development, or you already have experience, Neuron can be the perfect playground to move your idea from experiments to reliable production implementations.
Check out the documentation: https://docs.neuron-ai.dev
- Modular Workflow Architecture: Easily extensible with nested workflows
- Automated Research: Generate queries and perform web searches
- Structured Reports: Create well-organized reports with customizable sections
- Performance Monitoring: Track execution time of workflow steps
- Streaming Results: Get real-time updates as the report is generated
Download the project on your machine and open your terminal in the project directory. First, install the composer dependencies:
composer install
Create a .env file in your project root (see .env.example for a template), and provides the API keys based on
the service you want to connect with.
# At least one required
ANTHROPIC_API_KEY=
GEMINI_API_KEY=
OPENAI_API_KEY=
# Required
TAVILY_API_KEY=
# Optional
INSPECTOR_INGESTION_KEY=
INSPECTOR_TRANSPORT=syncExecute the agent with the command below:
php research.php
- Planning: Creates the structure of the report
- GenerateSectionContent: Generates content for each section using search results
- Format: Compiles the final report
- GenerateQueries: Creates search queries based on section topics
- SearchTheWeb: Executes parallel searches and processes results
Integrating AI Agents into your application, you're not working only with functions and deterministic code, you program your agent also influencing probability distributions. Same input ≠ output. That means reproducibility, versioning, and debugging become real problems.
Many of the Agents you build with Neuron will contain multiple steps with multiple invocations of LLM calls, tool usage, access to external memories, etc. As these applications get more and more complex, it becomes crucial to be able to inspect what exactly your agent is doing and why.
Why is the model taking certain decisions? What data is the model reacting to? Prompting is not programming in the common sense. No static types, small changes break output, long prompts cost latency, and no two models behave exactly the same with the same prompt.
The best way to do this is with Inspector. After you sign up,
make sure to set the INSPECTOR_INGESTION_KEY variable in the application environment file to start monitoring:
INSPECTOR_INGESTION_KEY=fwe45gtxxxxxxxxxxxxxxxxxxxxxxxxxxxxAfter configuring the environment variable, you will see the agent execution timeline in your Inspector dashboard.
