CortexDoc is an advanced AI-powered document intelligence framework designed for deep semantic understanding, reasoning, and analysis of complex documents.
Unlike traditional OCR pipelines that merely extract text, CortexDoc transforms raw documents into structured cognitive memory that can be queried, reasoned upon, debated, and analyzed through autonomous AI agents.
The system combines:
- OCR intelligence
- semantic retrieval
- vector memory
- multi-agent cognition
- local LLM reasoning
- strategic document analysis
All running locally.
Uses PaddleOCR for high-accuracy text extraction from:
- PDFs
- scanned documents
- images
- low-quality scans
Documents are converted into vectorized semantic memory using ChromaDB for:
- contextual retrieval
- long-term memory
- intelligent querying
- similarity search
CortexDoc introduces a society of reasoning agents capable of:
- analysis
- critique
- debate
- evaluation
- contradiction detection
- reflective reasoning
Agents simulate distributed cognition over document knowledge.
Integrated with Ollama-based local language models such as:
- Phi-3
- Mistral
- Llama3
No cloud dependency required.
Designed especially for:
- defence studies
- geopolitical analysis
- maritime strategy
- doctrine comparison
- intelligence extraction
- policy analysis
- PaddleOCR
- Ollama
- ChromaDB
- Sentence Transformers
- Python
- Gradio
git clone https://github.com/YOUR_USERNAME/CortexDoc.git
cd CortexDocpython -m venv venv
source venv/bin/activatepip install -r requirements.txtDownload Ollama:
Pull a model:
ollama pull phi3python app.pyOpen browser:
http://127.0.0.1:7860
What is this document about?
Summarize the strategic concepts discussed.
Identify contradictions in the doctrine.
Compare strategic priorities across documents.
Extract military or geopolitical themes.
- OCR extraction
- semantic retrieval
- local reasoning
- vector memory
- contextual QA
- multi-agent architecture
- strategic analysis pipeline
- Multi-document cognition
- autonomous research agents
- knowledge graph visualization
- contradiction heatmaps
- agent memory evolution
- real-time collaborative reasoning
- doctrine intelligence engine
- strategic forecasting models
Most document AI systems stop at extraction.
CortexDoc moves toward:
- cognition
- reasoning
- memory
- strategic interpretation
- autonomous analytical systems
The project explores how AI systems can evolve from passive text processors into active cognitive architectures.
MIT License
ROHIT PATIL