cognee

AI Memory - Cognitive Architecture, Testability, Production-Ready Apps

Open-source framework for building AI Memory, extending the limits of cognitive architecture, designed for accuracy, transparency, and control.

---

Star us on Github!

Jump into AI memory architecture, inspired by human cognitive processes, using Python. Cognee runs in iterations, from POC towards production-ready code.

To read more about the approach and details on cognitive architecture, see the blog post: AI Applications and RAGs - Cognitive Architecture, Testability, Production Ready Apps

Try it yourself on Whatsapp with one of our partners by typing /save {content you want to save} followed by /query {knowledge you saved previously}

Getting started

Run with Docker

To run cognee you need to have Docker installed on your machine.

Run Cognee in a couple of steps:

• Run cp .env.template .env in your terminal and set all the environment variables
• Run docker compose up in order to start graph and relational databases
• Run docker compose up cognee in order start Cognee

Debugging

To run Cognee with debugger attached you need to build the Cognee image with the DEBUG flag set to true.

• docker compose build cognee --no-cache --build-arg DEBUG=true
• docker compose up cognee

Run without Docker

• Run PYTHONPATH=. python cognitive_architecture/setup_database.py to setup database
• Run python -m gunicorn -w 1 -k uvicorn.workers.UvicornWorker -t 30000 --bind=127.0.0.1:8000 --log-level debug api:app

Debugging

• Run python -m debugpy --wait-for-client --listen localhost:5678 -m gunicorn -w 1 -k uvicorn.workers.UvicornWorker -t 30000 --bind=127.0.0.1:8000 --log-level debug api:app
• Attach debugger

Demo

Architecture

How Cognee Enhances Your Contextual Memory

Our framework for the OpenAI, Graph (Neo4j) and Vector (Weaviate) databases introduces three key enhancements:

• Query Classifiers: Navigate information graph using Pydantic OpenAI classifiers.
• Document Topology: Structure and store documents in public and private domains.
• Personalized Context: Provide a context object to the LLM for a better response.

Current Focus

Integration with keepi.ai and other apps

• Cognee uses the Neo4j graph database to map user data into a graph structure consisting of semantic, episodic, and procedural memory.
• Stores data and files through the WhatsApp chatbot keepi.ai
• Uses the graph to answer user queries and store new information in the graph.