Referencing the Danish tabloid BT, I created "BOT-T" — a tabloid-style website that automatically generates news using input from web scrapers. It’s a playful experiment in AI-driven media, where headlines and stories emerge from real-world data but are twisted into absurd, overly tabloid-esque headlines, exaggerated ad absurdum. The goal wasn’t accuracy, but to explore what happens when automation meets sensationalism — turning the internet’s noise into something both ridiculous and strangely familiar.
Behind the scenes, BOT-T runs on a simple Python backend. A custom script scrapes real websites for fresh data, then feeds that input to OpenAI’s language model to generate a new story — not a factual summary, but a wild, tabloid-style spin on the original. The content is managed through a headless CMS, which the script updates automatically. As soon as a new story is generated, it gets published to the site without any manual steps. It’s a fully automated pipeline: from raw web data to live tabloid headline, in minutes.
While the site ran automatically, I also built a lightweight GUI to manually add stories and control the pipeline. It lets me input custom prompts, trigger the generation process, and review or tweak content before it goes live. This gives me the option to intervene — whether to steer the tone, experiment with different prompts, or just have a bit more fun with the headlines. It’s a way to stay in the loop without disrupting the automation.
