Prototype of using GPT index to answer questions in technical documentation.

Create .env file and add OpenAI API key to it:

# .env
OPENAI_API_KEY=<sk-...>

Install dependencies:

pip install -r requirements.txt

Run:

python main.py

Right now the index just contains a couple of random documentation pages about Atlas Functions.

You can find these pages in gen_index.py in the urls list. If you want to change the pages indexed, you can just edit this list of links.

To regenerate the index, run:

python gen_index.py

• basic web interface
  • consider using https://playground.arwes.dev/core/
• UI toggle for human languages (Spanish, Portuguese, Japanese, Pirate)
• type of content (code block, how to, conceptual overview)
• create simple server to handle requests. flask?
• demo
• presentation

• multi-tier index (global search, and search by site)
• continuous conversation
• and more?

• deploy python backend to Heroku or some other infra PaaS
• deploy frontend to Netlify
• set up git lfs to track those big index files.