IntraMind Documentation

Complete guide to deploying, configuring, and using IntraMind for your organization

What is IntraMind?

IntraMind is a local intelligence system that transforms your organization's internal documents into a smart, searchable knowledge network. It runs entirely on your infrastructure, ensuring complete data privacy while delivering AI-powered insights.

Privacy-First

Offline-Capable

AI-Powered

Multi-Format

System Architecture

Documents

Ingestion

Embeddings

Vector DB

Query

LLM

Response

Installation

Prerequisites

Python 3.9 or higher
8GB RAM minimum (16GB recommended)
10GB free disk space
Ollama installed (for local LLM)

Quick Start

# Clone the repository
git clone https://github.com/crux-ecosystem/IntraMind-Showcase.git
cd IntraMind-Showcase
# Install dependencies
pip install -r requirements.txt
# Configure your settings
cp config.example.yaml config.yaml
# Start IntraMind
python main.py

Configuration

config.yaml

1# IntraMind Configuration
2model:
3  embedding: "all-MiniLM-L6-v2"
4  llm: "llama3"
5  
6storage:
7  vector_db: "chromadb"
8  documents_path: "./documents"
9  
10server:
11  host: "0.0.0.0"
12  port: 8080
13  lan_only: true

Model Options

Embedding Models:

• all-MiniLM-L6-v2 (Fast, 80MB)
• all-mpnet-base-v2 (Higher quality, 420MB)
• multi-qa-MiniLM-L6-cos-v1 (Q&A optimized)

LLM Models:

• llama3 (Balanced, 8B)
• mistral (Faster, 7B)
• llama3:70b (Highest quality, 70B)
• codellama (Technical docs)

Performance Metrics

Query Performance

45ms

Embedding

120ms

180ms

LLM

345ms

Total

Average response time: <350ms (local deployment)

Usage

Adding Documents

Terminal

# Add documents to the folder
cp /path/to/documents/* ./documents/
# Trigger re-indexing
python main.py --reindex

API Example

Terminal

import requests
response = requests.post(
    'http://localhost:8080/api/query',
    json={'query': 'What is our policy?'}
)
print(response.json()['answer'])

System Components

Document Ingestion

Processes PDFs, DOCX, images with OCR

Text Extraction

Normalizes and chunks documents

Embedding Generation

Creates vector representations

Vector Storage

Stores in ChromaDB

Query Engine

Retrieves relevant context

LLM Inference

Generates answers using local models

Response Formatting

Presents results with citations

Deployment

LAN Deployment

Set lan_only: true in config
Configure firewall rules
Set up SSL certificates (optional)
Deploy on internal server

Docker Deployment

Terminal

# Build the image
docker build -t intramind .
# Run the container
docker run -d -p 8080:8080 \
  -v ./documents:/app/documents \
  intramind

API Reference

POST /api/query

Request:

Terminal

{
  "query": "Your question here",
  "max_results": 5
}

Response:

Terminal

{
  "answer": "Generated answer",
  "sources": [
    {
      "document": "filename.pdf",
      "page": 3,
      "relevance": 0.95
    }
  ]
}

Best Practices & Troubleshooting

Best Practices

Structure documents in logical folders
Re-index when adding new documents
Choose models based on your hardware
Regularly backup vector database

Troubleshooting

Slow query responses

Reduce chunk size or use smaller embedding model

Out of memory

Increase RAM or use smaller LLM model

Documents not indexing

Check file permissions and supported formats

Need Help?

Visit our GitHub repository for issues, documentation, and updates

View GitHub Repository →

# Clone the repository

git clone https://github.com/crux-ecosystem/IntraMind-Showcase.git

cd IntraMind-Showcase

# Install dependencies

pip install -r requirements.txt

# Configure your settings

cp config.example.yaml config.yaml

# Start IntraMind

python main.py

1# IntraMind Configuration

2model:

3 embedding: "all-MiniLM-L6-v2"

4 llm: "llama3"

6storage:

7 vector_db: "chromadb"

8 documents_path: "./documents"

10server:

11 host: "0.0.0.0"

12 port: 8080

13 lan_only: true