# Novodraft

**Version:** 0.3.0 (end - January, 2023)  
**Original Development:** 0.0.1 (January, 2022)

Novodraft: a web application that uses Artificial Intelligence (AI) to automate the drafting of legal discovery documents in
civil litigation. Written in [JavaScript](https://262.ecma-international.org/6.0/) and [Python](https://www.python.org/) using
[React](https://react.dev/) for the UI, with
[Node.js](https://nodejs.org/) and [Express.js](https://expressjs.com/) on the backend.

License: [GPL Version 3](https://opensource.org/license/gpl-3-0).

Novodraft was intended for use by attorneys and legal professionals handling
civil litigation, particularly those who draft discovery documents such as
interrogatories, requests for production, and requests for admissions. The
application automates discovery drafting, allowing attorneys to focus on tailoring substantive legal arguments as they see fit.

## Features

Novodraft generates both outgoing discovery requests and responses to incoming requests. The application:

- Accepts [PDF](https://en.wikipedia.org/wiki/PDF) uploads of incoming discovery requests
- Parses and classifies document types (interrogatories, requests for production, requests for admissions, or combined formats)
- Uses the [OpenAI](https://openai.com/)
[Application Programming Interface (API)](https://en.wikipedia.org/wiki/API) to generate legally-sound objections and responses
- Applies [Conclusion, Rule, Application, Conclusion (CRAC)](https://en.wikipedia.org/wiki/CRAC) structure where appropriate
- Formats output documents according to state-specific court styles (New York,
New Jersey, Florida, Michigan)
- Exports completed documents as [.docx](https://en.wikipedia.org/wiki/Office_Open_XML) files ready for service on opposing counsel



## Architecture

This is a client-server monorepo:

```
Novodraft/
├── ax3Client/       # React frontend application
└── ax3Services/     # Node.js/Python backend services (incl. LLM API)
```



### Frontend (ax3Client)

- A [React](https://react.dev/) 18 single-page application using:
- [React Router](https://reactrouter.com/) for navigation
- [React Bootstrap](https://react-bootstrap.github.io/) for
[User Interface (UI)](https://en.wikipedia.org/wiki/User_interface) components
- [Firebase](https://firebase.google.com/) Authentication for user management
- [Stripe](https://stripe.com/) for subscription payments
- [SCSS](https://sass-lang.com/) for styling



### Backend (ax3Services)

A collection of [Node.js](https://nodejs.org/) and
[Python](https://www.python.org/) services:

- **docGenService**: Generates formatted
[.docx](https://en.wikipedia.org/wiki/Office_Open_XML) documents using the
[python-docx](https://python-docx.readthedocs.io/) library
- **docParserService**: Parses and classifies uploaded discovery documents
- **docConvertService**: Converts PDFs to processable formats
- **tesseReaderService**: Performs
[Optical Character Recognition (OCR)](https://en.wikipedia.org/wiki/Optical_character_recognition) using [Tesseract.js](https://tesseract.projectnaptha.com/)
- **agentService**: Manages [OpenAI](https://openai.com/) API interactions and prompt templates
- **paymentService**: Handles [Stripe](https://stripe.com/) payment processing
- **storageService**: Manages document storage operations



## Getting Started



### Prerequisites

Install the following on your system:

**For Debian/Ubuntu:**

```bash
sudo apt-get update
sudo apt-get install -y nodejs npm python3 python3-pip
```

**For macOS (using [Homebrew](https://brew.sh/)):**

```bash
brew install node python3
```

Verify installations:

```bash
node --version    # Should be v18.x or higher
npm --version     # Should be v9.x or higher
python3 --version # Should be v3.8 or higher
```



### Installation

1. Clone the repository:

```bash
git clone <repository-url>
cd Novodraft
```

1. Install frontend dependencies:

```bash
cd ax3Client
npm install
```

1. Install backend dependencies:

```bash
cd ../ax3Services
npm install
pip3 install python-docx
```

1. Configure environment variables. Create `.env` files in both `ax3Client/` and
  `ax3Services/` directories with the following variables:

**ax3Client/.env:**

```
REACT_APP_API_DEV=http://localhost:3001
REACT_APP_API_PROD=<your-production-api-url>
REACT_APP_FIREBASE_API_KEY=<your-firebase-api-key>
REACT_APP_FIREBASE_AUTH_DOMAIN=<your-firebase-auth-domain>
REACT_APP_FIREBASE_PROJECT_ID=<your-firebase-project-id>
```

**ax3Services/.env:**

```
OPENAI_API_KEY=<your-openai-api-key>
STRIPE_SECRET_KEY=<your-stripe-secret-key>
```

1. Configure [Firebase](https://firebase.google.com/). Create a Firebase project
  and enable:
  - Authentication (Email/Password)
  - Cloud Firestore



### Running the Application

Start the frontend development server:

```bash
cd ax3Client
npm start
```

The frontend will be available at `http://localhost:3000`.

Start the backend services:

```bash
cd ax3Services
npm start
```



### Building for Production

```bash
cd ax3Client
npm run build
```

The production build will be output to `ax3Client/build/`.

## Supported Document Types

Novodraft handles the following discovery document types:


| Type                    | Description                                 |
| ----------------------- | ------------------------------------------- |
| Interrogatories         | Written questions requiring sworn answers   |
| Requests for Production | Demands for documents and tangible evidence |
| Requests for Admissions | Requests to admit or deny specific facts    |
| Combined/Numbered       | Documents containing multiple request types |




## State-Specific Formatting

The document generation service applies jurisdiction-specific caption headers
and formatting for:

- New York
- New Jersey
- Florida
- Michigan



## Versioning

This project uses [Semantic Versioning (SemVer)](https://semver.org/). Version
numbers follow the format MAJOR.MINOR.PATCH:

- MAJOR: Incompatible API changes
- MINOR: Backwards-compatible functionality additions
- PATCH: Backwards-compatible bug fixes



## Background

This project began in 2021 as an early experiment with
[Large Language Model (LLM)](https://en.wikipedia.org/wiki/Large_language_model)
technology to automate the creation of discovery documents in civil litigation.
The version in this repository represents work completed through January 2023.
The codebase was migrated from [GitHub](https://github.com/) to a self-hosted
[Gitea](https://about.gitea.com/) instance.

## Participation

### Contributing

Contributions are welcome, but be aware this project has largely been abandoned since 2023. The project maintainers make no guarantee that submitted pull requests will be merged. To contribute:

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Submit a pull request

Bug reports should be submitted via Git issues.

No Contributor License Agreement (CLA) is required.

### Code of Conduct

Contributors are expected to:

- Contribute in a spirit that seeks to improve society and human well-being
- Avoid harm
- Be honest and trustworthy
- Respect the work required to produce new ideas, inventions, creative works,
and computing artifacts
- Respect privacy
- Be fair and do not discriminate
- Create code that promotes these principles



### Privacy

Novodraft does not transmit user activity data off of the device on which it is
run beyond what is necessary for the core functionality (authentication,
payment processing, and AI document generation).

## Authors

- sj — [sj@sjdev.co](mailto:sj@sjdev.co)

## License

This project is licensed under the
[GNU General Public License Version 3](https://opensource.org/license/gpl-3-0).
See the LICENSE file for details.