ContextFlow.AI

An intelligent AI research assistant powered by LLMs, RAG, and LangGraph for document understanding, reasoning, and context-aware answers.

◈ Overview

ContextFlow AI is a production-inspired AI research assistant that combines:

• Large Language Models (LLMs)
• Retrieval-Augmented Generation (RAG)
• Vector Databases
• LangGraph workflows
• Tool-assisted reasoning (AI Agent)
• Conversation/session memory

The system allows users to upload documents, retrieve relevant information, and generate grounded answers instead of relying only on an LLM's internal knowledge.

◈ Problem

Traditional LLM chatbots have limitations:

• No access to private documents
• Hallucinated information
• No semantic document search
• Limited reasoning capabilities
• No persistent conversation context

◈ Solution

ContextFlow AI addresses these challenges by integrating several modern AI techniques into a unified system.

The application:

• Understands user intent using an LLM
• Retrieves relevant information from uploaded documents using RAG
• Stores semantic document embeddings in a vector database
• Maintains conversational memory for contextual interactions
• Uses AI Agents to plan and execute multi-step tasks (calculation, web search)
• Produces grounded, explainable, and context-aware responses

◈ LangGraph Workflow

The assistant uses a conditional LangGraph workflow:

1. User sends a question.
2. Router LLM determines whether Retrieval-Augmented Generation (RAG) is required.
3. If RAG is required:
   • Retrieve relevant documents from ChromaDB.
   • Store the retrieved context in the graph state.
4. The chatbot LLM generates a response.
5. If the LLM requests a tool:
   • Execute the appropriate tool.
   • Return the tool result to the chatbot.
   • Generate the final response.
6. Return the answer to the user.

◈ ⚙️ Installation

1. Clone the repository

git clone https://github.com/<your-username>/ContextFlow-AI.git

cd ContextFlow-AI

2. Create a virtual environment

uv venv

Activate it.

Windows

.venv\Scripts\activate

Linux / macOS

source .venv/bin/activate

3. Install dependencies

uv sync

4. Configure environment variables

Create a .env file in the project root.

OPENAI_API_KEY=your_api_key_here
TAVILY_API_KEY=your_api_key_here

5. Run the backend

uv run uvicorn app.main:app --reload

6. Run the frontend

streamlit run .\frontend\app.py

Open:

http://localhost:8501/

◈ License

This project is licensed under the GNU General Public License v3.0 (GPL-3.0).

See the LICENSE file for more information.