I working on a service[1] for myself to track visually the first page of some news websites hourly. Hosting the agent on my RP5.

[1]https://kiosk24.site

I'm building an app that takes a screenshot every hour from some news websites. It's is a small python script running on my raspberry pi 5 and for now I'm saving the images there. I'm planning to build a front-end app to explorer how each website changes over the course of a day, focusing only on the top of the landing page.

That reminds me of this project: https://media.ccc.de/v/33c3-7912-spiegelmining_reverse_engin...

Why screenshots and not copy the source?

Thanks for sharing the link. I believe dealing with the code adds a lot more complexity to my project. On the other hand, I’d like to visually compare the websites over time, which is easier to do using screenshots.

I'm working on Austen[0] which generates story relationship graphs with Mermaid

[0] https://github.com/herol3oy/austen

I use Mermaid in my app[0] to generate books story relationships.

[0]https://github.com/herol3oy/austen

Your app looks interesting. Am curious about how is AI being used. Assuming that to discover character relationships, how is that done? Thanks.

I love the timeline! Are we limited to search for some keywords? I tried some countries but couldn’t find them in the results!

Fair point. I’ll add it.

Oh, this is a great idea. I'd love to try it and maybe I could add this feature to the current one! Thank you.

Hi, I created the app. I appreciate all of your comments. I know some may find the result of a book not concise or accurate. One issue was that I couldn't afford the price of a better LM service. At first, I tried a free tier of Gemini, but its results were disappointing. Later, I switched to DeepSeek and got better results.

It's impressive you're able to have decent results from free tiers and open source models.

This is a neat idea! My first try will be "The Sopranos".

That was exactly what I struggled with—and still do!