I gave Claude Blog one prompt and it produced a complete post: cover image, audio summary, charts, real sources, internal links. Then I did the human pass on top. The post happens to be about Claude's new invisible watermark, so you get both the workflow and the news in one go. Watch: https://www.youtube.com/watch?v=-JMWPF4t82M Read the finished post: https://claude-blog.md/blog/claude-watermark-synthid-guide

@Daniel Agrici I just skimmed your new video "Claude is getting a watermark - here's how I still publish" in 2x speed and whilst it was great I did notice possibly an improvement, regarding your humanising. Have you checked out "https://github.com/blader/humanizer" and would you find it useful to incorporate into the blog itself? For instance, claude writes the blog, codex uses "humanizer" to humanise the language. Vice versa for codex>claude etc.

Found this and had to share. It's called G2. Free, open source, and it makes every chart type you can think of. Bar charts, flow diagrams, heat maps, maps, the weird circular ones you see in good reports. Look at the screenshot, all of that comes from one tool. You don't need to be a developer to use it. Drop the link into Claude, ChatGPT or Cursor, tell it what data you have and what you want it to look like, and it builds it for you. Where I'd use it: client reports, carousel graphics, blog post visuals, anything where a screenshot of a spreadsheet would kill the post. Link: github.com/antvis/G2

A lot of AI content sounds polished but never gives a clear answer. Here’s a simple way to improve it: 1. Start each section with a direct answer to one specific question. 2. Add a real example, source, or personal insight that supports the answer. 3. Link to another relevant page that helps the reader take the next step. The goal is not to publish more words. It is to make every page easier to understand and trust. What are you currently doing to make your content stand out from generic AI articles?