A lab space for generating tons of content with AI

Will be a mess, and it will be fun.

Setup

Make sure you have python3 and pip installed. Make sure you have venv installed. Make sure you have uv installed.

Activate

python3 -m venv
source .venv/bin/activate
pipx install uv
pip install -r python-requirements.txt

Default to Local AI

Download:

1. MSTY
2. Ollama

YOU WILL NEED YOUR OWN API KEYS for remote LLM services, put in the .env file

1. Recraft for image generation.
2. OpenAI for text generation.
3. Jina for AI search.
4. Anthropic for AI search.
5. Groq for AI search.
6. Imagekit for image storage.

Optional Tooling

Packages Include

Data Analysis Notebooks

1. Marimo for Data Analysis Notebooks.

On Data Privacy and NDA/FrienDA.

If you're doing data analysis with Marimo or any other tool, make sure you create a private-data directory, which is already listed in the .gitignore file. This is to ensure that you don't commit any private data to the repository, which we want to default to build-in-public philsophy and practice.

AI Search with Local LLMs

1. Perplexica for AI Search.

Diagram AI Agent

2. Mermaid Diagram AI Agent

On Open Graph Images

I use OpenGraph.io to generate Open Graph images for the tooling dir. I have been in the process of moving it from script based to Observer/Watcher based, which is in the tidyverse submdoule.

To Run Python Scripts

source .venv/bin/activate python3 ai-labs/apis/recraft/generate-banner-images-recraft.py

To Run Node Scripts

node ai-labs/apis/imagekit/convertImageToImageKitUrl.js