GitHub - precious112/prism-ai-deep-research: Open Source Deep Research and Learning Agent (Perplexity Pro Alternative). Orchestrates autonomous researchers using LangGraph, Python, and Next.js for students, developers, and lifelong learners.

An open-source AI research agent that thinks like a human analyst.

Alternative to: Perplexity Pro, OpenAI Deep Research.

🚀 Why Prism AI?

LLMs are great at summaries but bad at deep research. They hallucinate, miss key details, and struggle with long-context tasks.

Prism AI solves this by orchestrating a team of autonomous agents. Whether you're a student, a developer, or a lifelong learner, Prism helps you dive deep into any topic. Instead of a single inference pass, it uses a Plan-and-Execute architecture to:

Plan: Break down a complex query into a structured Table of Contents.
Research: Spawn multiple "Researcher Agents" to search, crawl, and read the web in parallel.
Synthesize: Aggregate findings into a cohesive, cited report.
Visualize: Generate custom charts and diagrams to explain complex data.

🧠 Beyond Research: A Powerful Learning Tool

Prism AI isn't just for professionals. It's a versatile tool for anyone curious to learn. Use it to:

Master new topics: Generate comprehensive guides on anything from quantum computing to ancient history.
Accelerate academic work: Synthesize literature reviews, find citations, and explore new perspectives.
Onboard onto codebases: Point it at a GitHub repo to understand its architecture and key components.

🎥 Demo

✨ Key Features

🧠 Plan-and-Execute Pattern: Uses a PlanningAgent to generate a research roadmap before executing.
⚡ Parallel Execution: Utilizes Python asyncio to run 5+ research agents simultaneously, reducing latency by 80%.
🔄 LangGraph State Machine: Agents aren't just chains; they are state machines that can self-correct, loop back, and retry searches if information is missing.
📊 Dynamic Visualization: The agent can decide to generate custom React components (charts, diagrams) to better explain its findings.
🔍 Transparent Sources: Every claim is cited with a direct link to the source.

🏗️ Architecture

Prism AI is built on a microservices architecture designed for scalability.

graph TD
    Input[User Query] --> Planner[Planner Agent]
    Planner --> Plan[Research Plan / ToC]
    
    Plan -->|Parallel Split| Sections{Sections}
    
    subgraph Parallel Execution
        Sections -->|Sec 1| R1[Researcher Agent 1]
        Sections -->|Sec 2| R2[Researcher Agent 2]
        Sections -->|Sec N| RN[Researcher Agent N]
    end
    
    R1 --> Agg[Conclusion Agent]
    R2 --> Agg
    RN --> Agg
    
    Agg --> Report[Final Report Stream]

Core: Python, LangGraph, LangChain.
API: Node.js, Express.
Frontend: Next.js, React, Tailwind.
Real-time: Go WebSocket Server, Redis.

⚡ Quick Start

Get up and running in minutes using Docker.

Prerequisites

Docker & Docker Compose
OpenAI API Key
Serper API Key (for Google Search)

Installation

Clone the repository

git clone https://github.com/precious112/prism-ai-deep-research.git
cd prism-ai-deep-research

Set up environment variables

cp .env.example .env
# Edit .env with your API keys

Run with Docker
```
docker-compose up --build
```

Visit http://localhost:3000 to start researching.

📚 Documentation

For detailed guides on development, deployment, and architecture, visit the docs directory.

Installation Guide - Full setup instructions.
Architecture Deep Dive - How the agents work internally.
Development Workflow - How to contribute.

🤝 Contributing

We welcome contributions! Please see our Development Workflow to get started.

📄 License

Distributed under the MIT License. See LICENSE for more information.