A live blog where AI writes every article and paints every image
A Spanish-language minerals and geology blog where every article and image is generated by my ContentGen-AI pipeline, published on Jekyll.
I wanted a real production site to test a question rather than a theoretical one: can a content blog run end to end with no human writing or illustrating a single post? Demos are easy to fake. A live site that has to actually publish, rank, and read well is not. galena.es is that test, built around minerals, mining, and gemstones because the subject is broad enough to keep the pipeline busy and specific enough to judge quality.
Every article and every image on galena.es comes from ContentGen-AI, the Python pipeline I wrote to learn OpenAI API integration and prompt engineering. Article text is generated through the API, and each post gets a matching DALL-E image created automatically. The output is published as a Jekyll static site (built on the YAT theme) and hosted on GitHub Pages, so the running cost stays near zero.
What I learned is that the hard part is not generating words, it is generating words worth reading. Prompt structure, topic scoping, and consistency between the text and the image are where most of the effort went. galena.es is the honest scoreboard for that work: an experiment in fully-automated publishing that anyone can open and judge for themselves.