Properties3
| Type | Practice |
| Note created | Jun 23, 2026 |
| Last updated | Aug 4, 2026 |
The use of AI in my Digital Garden is very constrained and under control. The AI generated content is clearly marked using tags, and isolated.
tl;dr
In my Digital Garden, there is no AI-generated content that has not been, at the very least, previously read by myself.
I use extensively agentic programming at work, more specifically Claude Code, and I also use it for initial research on the web with a very strict framework for knowledge discovery. I also use it extensively for hobby projects where the technology is not that familiar to me.
For that reason, it is very natural to capture the knowledge acquired during a session into one or several notes automatically. All of that is tagged as AI generated, and actually this Garden has two different ways to tag this kind of content:
#ai-generated, which is content that has been written by some agent. The contents are AI generated and I would understand not being interested in reading it, but it serves a purpose for me: it helped me clean up some unorganized notes, create supporting ones. For example, I do that extensively with the artwork notes, which are derived from my notes when I visit a museum or exhibition. These notes are mostly verified by me but can contain some inconsistencies in the text (they are also usually extremely short).#ai-assisted, which is content derived from the use of an agent. Maybe it stems from the work I did with an agent, it helped me research a topic, or was previously AI generated and I re-wrote it to be adhere more to what I would write (and like to read). This content is not AI generated, and I wrote it, but I like to mark it because it can contain blind spots or not be the best way. For instance, I usually do this to get an idea on how a recipe is cooked in general, without having to read 12 different ones.
Below you can find an up-to-date list of all the AI notes, sorted by recency.