I really like how he automates his newsletter off of his blog, notes, TIL, etc., but it's also amazing how complicated it is. Imagine the blog would be on Markdown, nothing of this is needed as the "content" would just be locally for a Claude to read, it could do the SQL-query with a cmd-line instead, or you could build a small wrapper around if you want. Plus if the Newsletter supports Markdown, too (I use Listmonk), then the conversion would be straight forward too, also with no bugs.
This inspires me to automate one day more of my newsletter too, maybe as I have everything in place for it. And I automate blog post to send out automatically, but not a collection of notes and blogs.
I found Presenterm [1] to be optimal for me. Simple and works in the terminal, yet powerful to export to PDF and HTML. It supports Mermaid and images. I'm also collecting a list [2] with other Markdown-first presentation tools, and according to the git stars, reveal.js seems to be the most popular. Tough for me, it was too heavy.
i made a toolset i call "mdslides" for making pure HTML+CSS (no JavaScript) presentations in Markdown. it's just a CSS stylesheet and an 8 line Awk preprocessor for a slide delimiter, adding just enough HTML wrapping to work with the stylesheet. the stylesheet adds page breaks at each slide so you can get a PDF by asking your browser to print/save-as-PDF. it should work with any CommonMark Markdown formatter (i use "md2html" from the MD4C project).
I tried to use pandoc+revealJS, then tried presenterm (which was really nice but didn't give me enough control over font sizes), and then settled on Marp, which worked great.
yeah, avoiding all the serialization and deserialization, as I'm already working in Markdown and open text for almost all my stuff. The Claude Skill only seems to make sense for people who don't have their data in multiple different proprietary formats, then it might sense to packaging them into another one. But this can get messy pretty quick!
You can also read here, in case that works better: https://www.ssp.sh/blog/why-pivot-tables-never-die/. And it's definitely not LLM written. And I tried to answer the why in the chapter "Why Pivot Tables Endure":
> The enduring power of pivot tables is their robustness, simple usage, and fast, interactive response. It's the Lingua Franca of data if you are not fluent in the language of SQL or Python. A common language everyone understands: the top management, domain experts, and developers. It's an interface to data; it's the first no-code interface. Instead of the multidimensional query language MDX or the newer DAX, people can use a simple drag-and-drop interface. It democratized data analysis.
I'm a BI engineer, and I always convince people to use something else. But when I need to calculate my finances, and does not fit into Obsidian, I find myself using Excel again too :D Great product. Maybe not so much if you need to align on financial numbers - as everyone has their own truth :)
