Local-first AI-powered document intelligence platform for investigative journalism. ArkhamMirror provides journalists and researchers with advanced NLP and Vision AI capabilities for analyzing documents entirely offline, ensuring absolute privacy and transforming unstructured data into actionable insights through semantic search, knowledge graphs, and contradiction detection.

How It Works

The platform leverages offline Retrieval-Augmented Generation (RAG) for conversational data analysis, enabling semantic search that goes beyond keywords to find concepts. It constructs knowledge graphs to visualize connections between entities (People, Orgs, Places) and uses Vision AI models for extracting complex tables from PDFs and images. Additionally, it features auto-timelines for event reconstruction and contradiction detection to flag conflicting information across documents, all while maintaining zero cloud dependencies.

Quick Start & Requirements

ArkhamMirror offers a Smart Installer that automates the setup of Python, Docker, and database dependencies. Windows users can run setup.bat, while Mac/Linux users should execute bash chmod +x setup.sh && ./setup.sh. A standalone, browser-based Analysis of Competing Hypotheses (ACH) tool is also available for quick analyses without full installation, storing data locally in browser localStorage. Optional AI assistance requires users to provide their own API keys (OpenAI, Groq, Anthropic, or local LLMs).

• User Guide: [Link to User Guide]
• Network & Privacy: [Link to Network & Privacy Guide]
• ACH Analysis Guide: [Link to ACH Analysis Guide]

Highlighted Details

• Local AI & Offline RAG: Chat with data using advanced NLP models that run entirely on the user's machine.
• Vision AI for Tables: Visual table extraction from PDFs and images using sophisticated vision models.
• Knowledge Graph: Automatically generates and visualizes connections between entities like people, organizations, and locations.
• Contradiction Detection: Automatically identifies and flags conflicting statements within and across documents.
• Absolute Privacy: Designed for air-gapped operation with zero cloud dependencies, ensuring data never leaves the local "Data Silo".

Maintenance & Community

The project is presented as a personal endeavor aimed at providing powerful forensics tools without cloud costs or privacy risks. No specific community links (Discord, Slack) or details on contributors/sponsorships are provided in the README.

Licensing & Compatibility

The README does not explicitly state the software license. Compatibility for commercial use or closed-source linking is not detailed.

Limitations & Caveats

The project appears to be actively developed, with a focus on core features. Specific limitations, known bugs, or alpha status are not detailed in the provided text. The reliance on user-provided API keys for advanced AI assistance means external costs or setup are required for those features.