My AI Wiki Hasn't Earned Automatic Upkeep
My AI team and I completed a linked Obsidian wiki experiment. I judged it not yet useful enough for automatic upkeep: a finished build and a worthwhile recurring workflow are different outcomes.
My AI team and I completed an Obsidian wiki proof of concept. We had linked pages, preserved sources and categories. My assessment was that it was not yet useful enough to warrant automatic upkeep.
That is the result of the experiment, not an unfinished sentence waiting for a more positive ending. Completing the build gave me something concrete to assess. It did not oblige me to turn it into another recurring workflow.
What we actually built
The static vault contained 23 linked topic and hub pages, alongside 92 byte-identical source snapshots. Source links and a hash manifest made the relationship to the underlying material inspectable.
We then organised the vault into five groups, adding category pages and tags without rewriting the notes. That gave the material another layer of navigation while preserving the existing text.
One detail did not hold: the graph colour settings failed to persist while Obsidian was open. That is a recorded limitation of the experiment, not evidence that the underlying notes were lost. It also does not, by itself, explain my overall assessment of usefulness.
A working structure is not yet a useful habit
The distinction here is between producing an organised collection and deciding that it deserves continuing maintenance. Page counts, links and categories describe what we made. They do not establish that I will get enough value from keeping it refreshed.
My decision was narrower than rejecting wikis or AI-assisted organisation in general. This particular proof of concept had not yet earned automatic upkeep. The work journal does not establish a more detailed diagnosis, and I should not invent one after the fact.
The practical lesson is to leave room for that outcome when building with AI. A completed experiment can support a decision not to automate the next step. Technical completion and a reason to keep something running are separate questions.
Evidence note: this account follows today's scoped work journal, including my recorded assessment. It is not an independent evaluation of the vault or a measured study of its usefulness.