Prolly

This project is an AI-powered application that generates outcomes based on user input. It uses the Groq API for natural language processing and is built with FastAPI and Modal for deployment.

Project Structure

├── README.md
├── src/
│ └── app/
│   └── api/
│      └── prolly.py

File Descriptions

1. src/app/api/prolly.py: The main application script containing:

   • Configuration for the Groq and Exa API client
   • The outcomes function that processes user input and generates outcomes
   • FastAPI app setup with CORS middleware
   • Endpoint definition for the /outcomes route

Frontend

1. Clone the repository:

git clone https://github.com/willdphan/future.git
cd <repository-name>

2. Install dependencies

For Frontend:

npm install

Backend

If you want to run this locally, we're using modal labs for deploying and running the FastAPI application. Create an account and create your env variables.

Here's how you can create your secret keys in terminal:

modal secret create my-api-keys GROQ_API_KEY={key} EXA_API_KEY={key}

Modal will install these packages in the script of prolly.py

# install necessary packages
image = (
    Image.debian_slim()
    .pip_install("fastapi", "uvicorn", "groq", "pydantic", "requests")
)

To deploy this application locally, start the FastAPI server using Modal, making it accessible at http://localhost:8000 by default.

modal deploy src/app/api/prolly.py

Otherwise, it should already be running on a Modal container.

API Usage

The application exposes a single endpoint:

• POST /outcomes
  • Accepts JSON payload with a prompt field
  • Returns generated outcomes based on the input prompt

TODO

• More outcomes
• Longer, detailed node descriptions
• Better probability weighting
• Dragable nodes
• More details about entire generated flowchart
• Loading page fix
• More instructions to guide user

License

This project is licensed under the MIT License. See the LICENSE file for details.