✨ AI Local Chat Bot

• Founder: Abdullah Khawer - LinkedIn

Introduction

AI Local Chat Bot is an AI-powered chat assistant like ChatGPT running locally, safely and privately using n8n for workflow orchestration, Qdrant for vector storage, and Ollama for running local LLMs. It also includes helper scripts for downloading and processing Confluence pages as PDFs.

Purpose: To ingest it with your private data in .PDF format (for now) and then use it to ask questions and get answered like any AI chat bot.

Features

• n8n: Orchestrates chat and data ingestion workflows (http://localhost:5678)
• Qdrant: Vector database for storing document embeddings (http://localhost:6333)
• Ollama: Local LLM inference (http://localhost:11434)
• Confluence PDF Downloader: Download all pages from a Confluence space as PDFs for ingestion

Demo

System Requirements

Resource	Minimum	Recommended	Best/Production
CPU	2 cores	4+ cores	8+ cores (modern x86_64)
Memory	4 GB	8 GB	16+ GB
Disk Space	10 GB (free)	30 GB (free)	100+ GB SSD/NVMe

Notes:

• Running large LLMs with Ollama may require more memory and CPU for optimal performance.
• SSD/NVMe storage is highly recommended for best performance, especially for vector DB and model files.
• For production, consider dedicated resources and regular backups for Qdrant data.

Official Documentation

n8n

• n8n Documentation
• n8n Community
• n8n Foundational Template

Qdrant

• Qdrant Documentation

Ollama

• Ollama Documentation

Confluence PDF Downloader

• See helper-scripts/confluence_pdf_downloader.py for script usage and configuration details (also documented below in this README).

Quick Start

1. Start the stack:

   ./start.sh

2. Access n8n UI:

   • Open http://localhost:5678 in your browser.
   • Sign up for the first time.

3. n8n Workflows:

   • The workflow in n8n-workflows/ai-chat-assistant.json handles PDF ingestion and chat bot. Import it via UI to make use of it.

4. Download Confluence PDFs (optional):

   cd helper-scripts
   pip install -r requirements.txt
   python confluence_pdf_downloader.py

   Follow the prompts or set these environment variables:

   • CONFLUENCE_URL
   • CONFLUENCE_USERNAME
   • CONFLUENCE_API_TOKEN
   • CONFLUENCE_SPACE_KEY

   PDFs will be saved in helper-scripts/data/confluence_<SPACE_KEY>/.

5. Data Ingestion:

   • Upload PDFs via the n8n form trigger from the n8n UI appeared after workflow import.

6. Ask:

   • Ask your question to the chat bot and wait for the answer.

Requirements

• Docker & Docker Compose
• Python 3.8+
• For helper scripts: pip install -r helper-scripts/requirements.txt

Helper Scripts

Confluence PDF Downloader

Script: helper-scripts/confluence_pdf_downloader.py

Downloads all pages from a Confluence space as PDFs. Uses the Confluence REST API and can convert HTML to PDF if direct export is unavailable.

Usage

cd helper-scripts
python3 confluence_pdf_downloader.py

You will be prompted for credentials or can set them as environment variables.

Project Structure

• docker-compose.yml — Service definitions for n8n, Qdrant, Ollama
• n8n-workflows/ — n8n workflow JSON files
• helper-scripts/ — Python scripts and requirements for data ingestion
• start.sh — Launches the stack and sets up helper scripts

AI Models

The following models are required and will be pulled automatically by the start script:

• llama3.2 (for chat)
• mxbai-embed-large (for embeddings)

Stopping and Logs

To stop the application:

docker-compose down

To view logs for all services:

docker-compose logs -f

🤝 Contributing

Contributions are welcome!

To contribute:

• Fork this repository and create a new branch for your feature or fix.
• Make your changes and ensure code quality and documentation are updated.
• Submit a pull request with a clear explanation of your changes.

For major changes, please open an issue first to discuss what you would like to change.

📝 License

This project is licensed under the Apache License - see the LICENSE file for details.

---

😊 Any contributions, improvements and suggestions will be highly appreciated.

🌟 Star this repository to know about the latest updates.