Team 4A1G — Tech Challenge WS23/24 — Technical University of Munich

ChatJustus

ChatJustus is an AI Chatbot to improve the Lawyer-Client UX by supporting the entire lawyer-client interaction journey. To learn more, check out our website. There, you can also try the demo.

Project Architecture

Tech Stack

• OpenAI API: gpt-3.5-turbo-1106, gpt-4-1106-preview and text-embedding-ada-002 for chat and document embeddings
• Our own gpt-wrapper for function tools, structured data generation, and more
• Qdrant for vector search, with our own VectorDB wrapper
• Assistant backend
• FastAPI WebSocket server
• Our own state synchronisation protocol
• React + Next.js frontend
• tailwindcss and NextUI interface

Documentation

Here's a comprehensive documentation of each and every important file/folder in the project. Click on the folder names to open them, and click on each summary box to switch between a Short summary and a Long description.
We have created it using our self-developed AI Recursive Summarizer tool, using the same underlying technology as ChatJustus itself! Using the tool, the entire documentation was generated in 10 minutes, what would have taken us 3-5 hours to write manually :)

Installation

For more detailed instructions, check the backend and frontend READMEs. For a quick start, follow the instructions below.

Quick Start

To install and run the ChatJustus backend server, you will need at least git and python3.10 or higher installed. To ease the installation process, we highly recommend using conda and poetry.

1. Clone the repository and navigate to the backend folder

   git clone https://github.com/4A1G/ChatJustus.git
   cd ChatJustus/backend

2. Create the tc conda environment and activate it

   conda env create -f environment.yaml
   conda activate tc

3. Install the python dependencies

   poetry install

4. Copy the .env.example file and rename it to .env. Then, fill in the OpenAI API key and the gmail account details for the email functionality.

5. Create the Legal DB and populate it with data, by running the Jupyter Notebooks create_legal_db.ipynb and create_lawyer_db.ipynb. Simply press "Run All" in the Jupyter Notebook UI. Make sure to shutdown/restart the kernel after running each notebook, because the local database storage does not allow for concurrent access.

6. Start the backend server

   poetry run expose

   This will start the server, expose it to the local network, and automatically open a browser window with a QR code that you can scan to access the server from the same network. You could also just open the localhost URL in your browser.

Other

Landing Page

The landing page is also a react + next.js app, and can be found under the landing branch. It is deployed using GitHub Pages.

Demo Server

We have a Google Cloud server running the demo, which is deployed by essentially running the same steps as in the Quick Start section.