//How to install AiDox System 

(Full Functionality online & offline)

• 1. FOR ONLINE FUNCTION: go to Ollama.com and create a free account. (easy).

• this is required to get your API key for online models to work. (save the API key ollama gives you into a file on your desktop or somewhere)

• AiDox will work offline without the need for an Ollama API key.

• 2. Download AiDox using the above button. Aidox itself does not require an install, but you will need to install other stuff for it to work fully. 

3. Included with aidox is a self running script file (install_aidox) that will download everything needed including the offline models, so it will be a large download, several gigabytes. It will install Python, Ollama, Ai Models, and everything needed. You should only need to run this once, and most updates can just download and run the new AiDox version. Might have to add new installs occasionally as the software grows and gets more advanced. 

• Right Click install_aidox and select Run as Administrator

• be sure to watch for any windows you need to click "ok" on or to accept (run files as administrator, ignore windows warnings)

• once ollama opens after it installs, close ollama for the process to continue if it seems like it has stopped at that point. (exit/close app via Taskbar icon or end task via task manager) Don't just close the window.

• if windows tries to block it downloading, installing, opening, select "Learn More" if it is on the pop up warning, and there should be an option to continue, download anyway, or "Run Anyway". 

• After the script downloads and installs everything AiDox will be ready to use. 

  • Open AiDox and click Settings at the top. Select the ollama cloud option and enter you API key in the field, click SAVE at the bottom of the settings window. 

• If you don't want to do that method, then do the following:

•  download Python 3.11 here: https://www.python.org/ftp/python/3.11.0/python-3.11.0-amd64.exe

•  download Ollama here: Ollama

  • • also where you obtain your API key (enter into AiDox Settings) 

• download MoonDream: Open Terminal or PowerShell and paste this - pip install moondream 

  • alternatively feel free to do it yourself following their instructions: Getting Started | Moondream Docs 

•   download AiDox and open, should be ready to use. You will need to still install the Ai models you wish to use for offline use. I have not tested all models.. only what I have in the original settings. 

I am working on AiDox Cloud, which will be a simple to use and setup version that works the same, just with cloud models. So no performance requirements basically...

I will also look at making an all in one official installation package for the main version. It would be large. 

REad below for more info

AiDox 1.5 

The Nexus of Document Intelligence & Forensic Analysis

The world's First Fully OFfline & online capable Ai-ntelligent document suite & knowledge archive

In an era of information overload, AiDox transforms chaos into clarity. It is a powerful, AI-driven document intelligence platform built for deep analysis, structured extraction, forensic insight, and strategic research. AiDox has the ability to easily evolve with current Ai model advancements. 

Whether you're analyzing case files, processing academic materials, extracting structured data, or investigating complex datasets, AiDox provides precision, flexibility, and control.

for windows pc only currently

🚀 What Is AiDox?

AiDox is an advanced document analysis and AI research platform designed to:

• Extract structured data from documents

• Analyze and summarize complex content

• Detect patterns, inconsistencies, and anomalies

• Organize and categorize large document collections

• Operate offline using local AI models

• Integrate with multiple AI backends
  
  

AiDox is built for professionals who demand more than surface-level summaries.

🔍 Core Capabilities

📄 Intelligent Document Processing

• PDF, text, and structured document ingestion

• OCR-based extraction (for scanned documents)

• Metadata parsing and indexing

• Multi-document cross-reference analysis
  
  

🧠 AI-Powered Analysis

• Deep contextual summarization

• Entity recognition

• Pattern and anomaly detection

• Cross-document correlation

• Timeline construction

• Relationship mapping
  
  

🗂 Structured Data Extraction

• Extract names, dates, financial data, case numbers, and key entities

• Convert unstructured text into structured formats

• Export usable datasets

• Prepare import files for database systems
  
  

🔐 Offline Capability

• Run fully offline using local AI models

• Maintain complete control over sensitive documents

• Ideal for legal, investigative, or confidential environments
  
  

☁️ Multi-Model Compatibility

• Use local models

• Connect to cloud AI providers

• Switch models depending on task complexity

• Maintain flexibility in performance vs. privacy
  
  

🏛 Use Cases

AiDox is built to serve a wide range of industries and disciplines.

⚖️ Legal & Case Analysis

• Review and summarize case files

• Identify contradictions across documents

• Extract evidence references

• Build structured case chronologies

• Detect inconsistencies in statements
  
  

🕵️ Investigative & Forensic Research

• Analyze investigative reports

• Cross-reference witness statements

• Extract timelines

• Identify suspicious financial patterns

• Organize large document evidence sets
  
  

🎓 Academic & Research Work

• Summarize research papers

• Extract citations and references

• Compare multiple academic sources

• Identify themes across large literature collections

• Structure research notes automatically
  
  

🏢 Business & Intelligence

• Contract review

• Risk analysis

• Financial data extraction

• Compliance auditing

• Competitive intelligence research
  
  

🗄 Archival & Library Systems

• Digitize and analyze archives

• Index historical collections

• Extract metadata from scanned records

• Organize large institutional databases
  
  

📱 Mobile Companion Workflow (Future / Hybrid Use)

• Capture documents via mobile camera

• Cloud-based extraction and analysis

• Generate structured files

• Import processed data into desktop AiDox

• Field-to-office workflow integration
  
  

🎯 Why AiDox Is Different

Unlike generic AI chat tools, AiDox is:

• Document-centric

• Structured-data focused

• Designed for investigative depth

• Capable of running fully offline

• Built for serious research environments
  
  

It is not just an AI assistant.
It is a document intelligence engine.

🛠 Technical Highlights

• Python-based architecture

• Modular AI backend support

• Scalable document indexing

• Export-ready structured output

• Designed for desktop expansion and mobile integration
  
  

💡 Who Is AiDox For?

• Attorneys

• Investigators

• Researchers

• Librarians

• Analysts

• Archivists

• Journalists

• Educators

• Intelligence professionals

• Institutions handling large document ecosystems
  
  

🔓 Privacy & Control

With offline AI capability, AiDox ensures:

• No forced cloud dependency

• Full local control of sensitive data

• Secure document handling

• Flexible deployment environments
  
  

🌐 The Vision

AiDox bridges forensic precision and AI intelligence.
It transforms static documents into dynamic knowledge networks.

Where others summarize, AiDox dissects.
Where others respond, AiDox investigates.

included with the .exe is a readme and a setup script file for the ai models you will need for offline mode

configuration info also in the 1.4 section below

AiDox v1.5 | User Manual & Documentation

"The Ultimate Triple-Tier Document Intelligence Suite"

AiDox is a high-performance document analysis tool designed to bridge the gap between massive local datasets and cutting-edge AI. It features a triple-tier model architecture to balance speed, cost, and deep reasoning.

🛠 Features & Functions

1. Document Management

Trilateral Ai Architecture: Supports PDF, TXT, and various Image formats (PNG, JPG, BMP, TIFF).

Smart OCR: Automatically detects if a PDF is an image-scan and uses an embedded Tesseract engine to extract text.

Navigation: Browse images extracted from documents with built-in "Next/Back" navigation.

2. AI Intelligence Suite

Ainalyze Selection (Ctrl+E): Highlight any text in the content view to get an instant AI deep-dive of that specific snippet within its context.

Ainalyze Image: Uses the vision model to "see" and describe the currently displayed image.

Analyze Connections: Select multiple documents to find shared entities, timeline links, and contradictions across your entire library.

Persistent Chat: Every document has its own dedicated chat history saved to a local database (documents.db).

3. Navigation & Utilities

Find Feature (Ctrl+F): Real-time text search within the document content with Up/Down arrow navigation.

Export Master Report: Generate a professional PDF report containing AI summaries, full text, and connection analysis.

JSON Manifest: Every processed file generates a .json export in the documents_export folder for external use.

⚙️ Settings & Model Configuration

AiDox uses different models for different tasks to optimize performance. Access these via the Settings button.

Model Tiers

Tier | Recommended Model (Local) | Recommended Model (Cloud) 

Primary | llama3:latest | gpt-4o | Deep reasoning and long-form document chat.

Flash | phi3:latest | gemini-1.5-flash | Quick summarization and "Ainalyze Selection".

Vision | moondream (Required) | qwen-vl-max | Image analysis and visual data extraction.

Configuration Modes

Offline (Local): Requires [Ollama](https://ollama.com/) to be running on localhost:11434.

Ollama Cloud: Connects to a remote Ollama server via a dedicated IP address.

ChatGPT: Uses OpenAI's API (Requires an API Key).

🚀 Pro-Tips for v1.5

Background Tasks: All AI calls run in the background. You can continue navigating the UI while the "AI Assistant" is thinking.

Strict Mode: When enabled in the chat, the AI is forbidden from making things up; it can only answer using the text found in your document.

*** Moondream Setup: For the "Ainalyze Image" feature to work, you must run " ollama pull moondream " in your terminal. This model was chosen for v1.4 because it is specifically optimized for fast, accurate vision tasks on consumer hardware.

*** Ainalyze Image won't work without Moondream installed on your pc. 

* To get AiDox v1.5 fully operational with local AI, you need Ollama. This is the engine that runs the models on your own hardware without needing an internet connection.

1. Install the Ollama Engine

Go to [Ollama.com](https://ollama.com/) and download the installer for your OS (Windows, Mac, or Linux).

Run the installer. Once finished, you will see a small llama icon in your system tray (near the clock).

2. Download the Models (Terminal)

Open your terminal (Command Prompt or PowerShell on Windows) and run these three commands one by one. Note: These models are large, so ensure you have a stable connection for the initial download.

For the Primary Brain (General Chat):

ollama pull llama3

For the Flash Brain (Fast Summaries & Selection):

ollama pull phi3

For the Vision Brain (Required for Ainalyze Image):

ollama pull moondream

Also:

ollama pull llava

3. Verify Your Installation

Once the downloads are finished, type this command to make sure everything is registered:

ollama list

You should see this output:

NAME               ID            SIZE      MODIFIED

llama3:latest      365c0bd3c000  4.7 GB    1 minute ago

moondream:latest   5164293f06a4  829 MB    1 minute ago

phi3:latest        64c1188f2424  2.3 GB    1 minute ago

4. Link to AiDox

Launch AiDox v1.5.

Click the Settings button.

Set the Provider to Offline (Local).

Ensure the Local Models fields match exactly:

Primary: llama3:latest

Vision: moondream:latest

Flash: phi3:latest

Click Save.

💡 Quick Troubleshooting

"Connection Refused": Make sure the Ollama app is actually running in your system tray.

Sluggish Performance: If your computer is older, try ollama pull tinyllama for the Flash model instead—it uses significantly less RAM. ( You can change the models in AiDox

GPU Support: Ollama automatically detects NVIDIA and AMD GPUs. If you have one, these models will run 10x faster.

If the Ai returns a bad or invalid category, summary, or answer, just try it again. Remove and Reload the document, Re-submit a question or keyword. 

• Different Ai models will have different performance and capabilities. 

© 2026 AiDox. Made in USA. //superFukt 

How to Install AiDox

Because AiDox is a specialized, independently developed tool, Windows and your web browser may flag it during the first installation. Follow these steps to complete the setup:

1. Bypassing Browser Warnings

• Chrome/Edge: If the browser says the file "is not commonly downloaded," click the three dots (...) or the arrow next to the download and select Keep.

• Note: It is highly recommended to download the .zip version for the smoothest experience.

2. Running for the First Time (Windows SmartScreen)

Windows may show a blue window stating "Windows protected your PC."

• Click More Info.

• Click Run Anyway.

3. System Requirements

• OS: Windows 10 or 11 (64-bit).

• Dependencies: To use the OCR features, ensure Tesseract-OCR is installed on your system.

• AI Engine: For local privacy, ensure Ollama is running in the background before launching AiDox.

4. Database & Files

Upon first launch, AiDox will create a folder structure in the directory where the .exe is located:

• /images/: Stores your extracted forensic visuals.

• /documents_export/: Stores your JSON manifests and PDF reports.

• documents.db: Your encrypted local database.

//TROUBLESHOOTING

• The OFFLINE models are slower than the cloud based models. Be patient. 

• Each model will behave differently or offer different results. 

• If a Summary seems off or did not generate, try removing the document, and reloading it. 

• If the chat assistant does not give a response, try the question again in a different format, quote any keywords you are asking about. Also ensure you have your ai models set up correctly.